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

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1305 - (show annotations)
Mon Apr 1 14:50:45 2013 UTC (6 years, 8 months ago) by ph10
File size: 245767 byte(s)
Fix open parens in MAKE/SKIP/PRUNE/THEN name bug.
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
134 Version 8.32 30-November-2012
135 -----------------------------
136
137 1. Improved JIT compiler optimizations for first character search and single
138 character iterators.
139
140 2. Supporting IBM XL C compilers for PPC architectures in the JIT compiler.
141 Patch by Daniel Richard G.
142
143 3. Single character iterator optimizations in the JIT compiler.
144
145 4. Improved JIT compiler optimizations for character ranges.
146
147 5. Rename the "leave" variable names to "quit" to improve WinCE compatibility.
148 Reported by Giuseppe D'Angelo.
149
150 6. The PCRE_STARTLINE bit, indicating that a match can occur only at the start
151 of a line, was being set incorrectly in cases where .* appeared inside
152 atomic brackets at the start of a pattern, or where there was a subsequent
153 *PRUNE or *SKIP.
154
155 7. Improved instruction cache flush for POWER/PowerPC.
156 Patch by Daniel Richard G.
157
158 8. Fixed a number of issues in pcregrep, making it more compatible with GNU
159 grep:
160
161 (a) There is now no limit to the number of patterns to be matched.
162
163 (b) An error is given if a pattern is too long.
164
165 (c) Multiple uses of --exclude, --exclude-dir, --include, and --include-dir
166 are now supported.
167
168 (d) --exclude-from and --include-from (multiple use) have been added.
169
170 (e) Exclusions and inclusions now apply to all files and directories, not
171 just to those obtained from scanning a directory recursively.
172
173 (f) Multiple uses of -f and --file-list are now supported.
174
175 (g) In a Windows environment, the default for -d has been changed from
176 "read" (the GNU grep default) to "skip", because otherwise the presence
177 of a directory in the file list provokes an error.
178
179 (h) The documentation has been revised and clarified in places.
180
181 9. Improve the matching speed of capturing brackets.
182
183 10. Changed the meaning of \X so that it now matches a Unicode extended
184 grapheme cluster.
185
186 11. Patch by Daniel Richard G to the autoconf files to add a macro for sorting
187 out POSIX threads when JIT support is configured.
188
189 12. Added support for PCRE_STUDY_EXTRA_NEEDED.
190
191 13. In the POSIX wrapper regcomp() function, setting re_nsub field in the preg
192 structure could go wrong in environments where size_t is not the same size
193 as int.
194
195 14. Applied user-supplied patch to pcrecpp.cc to allow PCRE_NO_UTF8_CHECK to be
196 set.
197
198 15. The EBCDIC support had decayed; later updates to the code had included
199 explicit references to (e.g.) \x0a instead of CHAR_LF. There has been a
200 general tidy up of EBCDIC-related issues, and the documentation was also
201 not quite right. There is now a test that can be run on ASCII systems to
202 check some of the EBCDIC-related things (but is it not a full test).
203
204 16. The new PCRE_STUDY_EXTRA_NEEDED option is now used by pcregrep, resulting
205 in a small tidy to the code.
206
207 17. Fix JIT tests when UTF is disabled and both 8 and 16 bit mode are enabled.
208
209 18. If the --only-matching (-o) option in pcregrep is specified multiple
210 times, each one causes appropriate output. For example, -o1 -o2 outputs the
211 substrings matched by the 1st and 2nd capturing parentheses. A separating
212 string can be specified by --om-separator (default empty).
213
214 19. Improving the first n character searches.
215
216 20. Turn case lists for horizontal and vertical white space into macros so that
217 they are defined only once.
218
219 21. This set of changes together give more compatible Unicode case-folding
220 behaviour for characters that have more than one other case when UCP
221 support is available.
222
223 (a) The Unicode property table now has offsets into a new table of sets of
224 three or more characters that are case-equivalent. The MultiStage2.py
225 script that generates these tables (the pcre_ucd.c file) now scans
226 CaseFolding.txt instead of UnicodeData.txt for character case
227 information.
228
229 (b) The code for adding characters or ranges of characters to a character
230 class has been abstracted into a generalized function that also handles
231 case-independence. In UTF-mode with UCP support, this uses the new data
232 to handle characters with more than one other case.
233
234 (c) A bug that is fixed as a result of (b) is that codepoints less than 256
235 whose other case is greater than 256 are now correctly matched
236 caselessly. Previously, the high codepoint matched the low one, but not
237 vice versa.
238
239 (d) The processing of \h, \H, \v, and \ in character classes now makes use
240 of the new class addition function, using character lists defined as
241 macros alongside the case definitions of 20 above.
242
243 (e) Caseless back references now work with characters that have more than
244 one other case.
245
246 (f) General caseless matching of characters with more than one other case
247 is supported.
248
249 22. Unicode character properties were updated from Unicode 6.2.0
250
251 23. Improved CMake support under Windows. Patch by Daniel Richard G.
252
253 24. Add support for 32-bit character strings, and UTF-32
254
255 25. Major JIT compiler update (code refactoring and bugfixing).
256 Experimental Sparc 32 support is added.
257
258 26. Applied a modified version of Daniel Richard G's patch to create
259 pcre.h.generic and config.h.generic by "make" instead of in the
260 PrepareRelease script.
261
262 27. Added a definition for CHAR_NULL (helpful for the z/OS port), and use it in
263 pcre_compile.c when checking for a zero character.
264
265 28. Introducing a native interface for JIT. Through this interface, the compiled
266 machine code can be directly executed. The purpose of this interface is to
267 provide fast pattern matching, so several sanity checks are not performed.
268 However, feature tests are still performed. The new interface provides
269 1.4x speedup compared to the old one.
270
271 29. If pcre_exec() or pcre_dfa_exec() was called with a negative value for
272 the subject string length, the error given was PCRE_ERROR_BADOFFSET, which
273 was confusing. There is now a new error PCRE_ERROR_BADLENGTH for this case.
274
275 30. In 8-bit UTF-8 mode, pcretest failed to give an error for data codepoints
276 greater than 0x7fffffff (which cannot be represented in UTF-8, even under
277 the "old" RFC 2279). Instead, it ended up passing a negative length to
278 pcre_exec().
279
280 31. Add support for GCC's visibility feature to hide internal functions.
281
282 32. Running "pcretest -C pcre8" or "pcretest -C pcre16" gave a spurious error
283 "unknown -C option" after outputting 0 or 1.
284
285 33. There is now support for generating a code coverage report for the test
286 suite in environments where gcc is the compiler and lcov is installed. This
287 is mainly for the benefit of the developers.
288
289 34. If PCRE is built with --enable-valgrind, certain memory regions are marked
290 unaddressable using valgrind annotations, allowing valgrind to detect
291 invalid memory accesses. This is mainly for the benefit of the developers.
292
293 25. (*UTF) can now be used to start a pattern in any of the three libraries.
294
295 26. Give configure error if --enable-cpp but no C++ compiler found.
296
297
298 Version 8.31 06-July-2012
299 -------------------------
300
301 1. Fixing a wrong JIT test case and some compiler warnings.
302
303 2. Removed a bashism from the RunTest script.
304
305 3. Add a cast to pcre_exec.c to fix the warning "unary minus operator applied
306 to unsigned type, result still unsigned" that was given by an MS compiler
307 on encountering the code "-sizeof(xxx)".
308
309 4. Partial matching support is added to the JIT compiler.
310
311 5. Fixed several bugs concerned with partial matching of items that consist
312 of more than one character:
313
314 (a) /^(..)\1/ did not partially match "aba" because checking references was
315 done on an "all or nothing" basis. This also applied to repeated
316 references.
317
318 (b) \R did not give a hard partial match if \r was found at the end of the
319 subject.
320
321 (c) \X did not give a hard partial match after matching one or more
322 characters at the end of the subject.
323
324 (d) When newline was set to CRLF, a pattern such as /a$/ did not recognize
325 a partial match for the string "\r".
326
327 (e) When newline was set to CRLF, the metacharacter "." did not recognize
328 a partial match for a CR character at the end of the subject string.
329
330 6. If JIT is requested using /S++ or -s++ (instead of just /S+ or -s+) when
331 running pcretest, the text "(JIT)" added to the output whenever JIT is
332 actually used to run the match.
333
334 7. Individual JIT compile options can be set in pcretest by following -s+[+]
335 or /S+[+] with a digit between 1 and 7.
336
337 8. OP_NOT now supports any UTF character not just single-byte ones.
338
339 9. (*MARK) control verb is now supported by the JIT compiler.
340
341 10. The command "./RunTest list" lists the available tests without actually
342 running any of them. (Because I keep forgetting what they all are.)
343
344 11. Add PCRE_INFO_MAXLOOKBEHIND.
345
346 12. Applied a (slightly modified) user-supplied patch that improves performance
347 when the heap is used for recursion (compiled with --disable-stack-for-
348 recursion). Instead of malloc and free for each heap frame each time a
349 logical recursion happens, frames are retained on a chain and re-used where
350 possible. This sometimes gives as much as 30% improvement.
351
352 13. As documented, (*COMMIT) is now confined to within a recursive subpattern
353 call.
354
355 14. As documented, (*COMMIT) is now confined to within a positive assertion.
356
357 15. It is now possible to link pcretest with libedit as an alternative to
358 libreadline.
359
360 16. (*COMMIT) control verb is now supported by the JIT compiler.
361
362 17. The Unicode data tables have been updated to Unicode 6.1.0.
363
364 18. Added --file-list option to pcregrep.
365
366 19. Added binary file support to pcregrep, including the -a, --binary-files,
367 -I, and --text options.
368
369 20. The madvise function is renamed for posix_madvise for QNX compatibility
370 reasons. Fixed by Giuseppe D'Angelo.
371
372 21. Fixed a bug for backward assertions with REVERSE 0 in the JIT compiler.
373
374 22. Changed the option for creating symbolic links for 16-bit man pages from
375 -s to -sf so that re-installing does not cause issues.
376
377 23. Support PCRE_NO_START_OPTIMIZE in JIT as (*MARK) support requires it.
378
379 24. Fixed a very old bug in pcretest that caused errors with restarted DFA
380 matches in certain environments (the workspace was not being correctly
381 retained). Also added to pcre_dfa_exec() a simple plausibility check on
382 some of the workspace data at the beginning of a restart.
383
384 25. \s*\R was auto-possessifying the \s* when it should not, whereas \S*\R
385 was not doing so when it should - probably a typo introduced by SVN 528
386 (change 8.10/14).
387
388 26. When PCRE_UCP was not set, \w+\x{c4} was incorrectly auto-possessifying the
389 \w+ when the character tables indicated that \x{c4} was a word character.
390 There were several related cases, all because the tests for doing a table
391 lookup were testing for characters less than 127 instead of 255.
392
393 27. If a pattern contains capturing parentheses that are not used in a match,
394 their slots in the ovector are set to -1. For those that are higher than
395 any matched groups, this happens at the end of processing. In the case when
396 there were back references that the ovector was too small to contain
397 (causing temporary malloc'd memory to be used during matching), and the
398 highest capturing number was not used, memory off the end of the ovector
399 was incorrectly being set to -1. (It was using the size of the temporary
400 memory instead of the true size.)
401
402 28. To catch bugs like 27 using valgrind, when pcretest is asked to specify an
403 ovector size, it uses memory at the end of the block that it has got.
404
405 29. Check for an overlong MARK name and give an error at compile time. The
406 limit is 255 for the 8-bit library and 65535 for the 16-bit library.
407
408 30. JIT compiler update.
409
410 31. JIT is now supported on jailbroken iOS devices. Thanks for Ruiger
411 Rill for the patch.
412
413 32. Put spaces around SLJIT_PRINT_D in the JIT compiler. Required by CXX11.
414
415 33. Variable renamings in the PCRE-JIT compiler. No functionality change.
416
417 34. Fixed typos in pcregrep: in two places there was SUPPORT_LIBZ2 instead of
418 SUPPORT_LIBBZ2. This caused a build problem when bzip2 but not gzip (zlib)
419 was enabled.
420
421 35. Improve JIT code generation for greedy plus quantifier.
422
423 36. When /((?:a?)*)*c/ or /((?>a?)*)*c/ was matched against "aac", it set group
424 1 to "aa" instead of to an empty string. The bug affected repeated groups
425 that could potentially match an empty string.
426
427 37. Optimizing single character iterators in JIT.
428
429 38. Wide characters specified with \uxxxx in JavaScript mode are now subject to
430 the same checks as \x{...} characters in non-JavaScript mode. Specifically,
431 codepoints that are too big for the mode are faulted, and in a UTF mode,
432 disallowed codepoints are also faulted.
433
434 39. If PCRE was compiled with UTF support, in three places in the DFA
435 matcher there was code that should only have been obeyed in UTF mode, but
436 was being obeyed unconditionally. In 8-bit mode this could cause incorrect
437 processing when bytes with values greater than 127 were present. In 16-bit
438 mode the bug would be provoked by values in the range 0xfc00 to 0xdc00. In
439 both cases the values are those that cannot be the first data item in a UTF
440 character. The three items that might have provoked this were recursions,
441 possessively repeated groups, and atomic groups.
442
443 40. Ensure that libpcre is explicitly listed in the link commands for pcretest
444 and pcregrep, because some OS require shared objects to be explicitly
445 passed to ld, causing the link step to fail if they are not.
446
447 41. There were two incorrect #ifdefs in pcre_study.c, meaning that, in 16-bit
448 mode, patterns that started with \h* or \R* might be incorrectly matched.
449
450
451 Version 8.30 04-February-2012
452 -----------------------------
453
454 1. Renamed "isnumber" as "is_a_number" because in some Mac environments this
455 name is defined in ctype.h.
456
457 2. Fixed a bug in fixed-length calculation for lookbehinds that would show up
458 only in quite long subpatterns.
459
460 3. Removed the function pcre_info(), which has been obsolete and deprecated
461 since it was replaced by pcre_fullinfo() in February 2000.
462
463 4. For a non-anchored pattern, if (*SKIP) was given with a name that did not
464 match a (*MARK), and the match failed at the start of the subject, a
465 reference to memory before the start of the subject could occur. This bug
466 was introduced by fix 17 of release 8.21.
467
468 5. A reference to an unset group with zero minimum repetition was giving
469 totally wrong answers (in non-JavaScript-compatibility mode). For example,
470 /(another)?(\1?)test/ matched against "hello world test". This bug was
471 introduced in release 8.13.
472
473 6. Add support for 16-bit character strings (a large amount of work involving
474 many changes and refactorings).
475
476 7. RunGrepTest failed on msys because \r\n was replaced by whitespace when the
477 command "pattern=`printf 'xxx\r\njkl'`" was run. The pattern is now taken
478 from a file.
479
480 8. Ovector size of 2 is also supported by JIT based pcre_exec (the ovector size
481 rounding is not applied in this particular case).
482
483 9. The invalid Unicode surrogate codepoints U+D800 to U+DFFF are now rejected
484 if they appear, or are escaped, in patterns.
485
486 10. Get rid of a number of -Wunused-but-set-variable warnings.
487
488 11. The pattern /(?=(*:x))(q|)/ matches an empty string, and returns the mark
489 "x". The similar pattern /(?=(*:x))((*:y)q|)/ did not return a mark at all.
490 Oddly, Perl behaves the same way. PCRE has been fixed so that this pattern
491 also returns the mark "x". This bug applied to capturing parentheses,
492 non-capturing parentheses, and atomic parentheses. It also applied to some
493 assertions.
494
495 12. Stephen Kelly's patch to CMakeLists.txt allows it to parse the version
496 information out of configure.ac instead of relying on pcre.h.generic, which
497 is not stored in the repository.
498
499 13. Applied Dmitry V. Levin's patch for a more portable method for linking with
500 -lreadline.
501
502 14. ZH added PCRE_CONFIG_JITTARGET; added its output to pcretest -C.
503
504 15. Applied Graycode's patch to put the top-level frame on the stack rather
505 than the heap when not using the stack for recursion. This gives a
506 performance improvement in many cases when recursion is not deep.
507
508 16. Experimental code added to "pcretest -C" to output the stack frame size.
509
510
511 Version 8.21 12-Dec-2011
512 ------------------------
513
514 1. Updating the JIT compiler.
515
516 2. JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
517 are added as well.
518
519 3. Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
520 PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
521 calling _pcre_jit_exec. Some extra comments are added.
522
523 4. (*MARK) settings inside atomic groups that do not contain any capturing
524 parentheses, for example, (?>a(*:m)), were not being passed out. This bug
525 was introduced by change 18 for 8.20.
526
527 5. Supporting of \x, \U and \u in JavaScript compatibility mode based on the
528 ECMA-262 standard.
529
530 6. Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
531 erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
532 This bug was probably introduced by change 9 of 8.13.
533
534 7. While fixing 6 above, I noticed that a number of other items were being
535 incorrectly rejected as "not fixed length". This arose partly because newer
536 opcodes had not been added to the fixed-length checking code. I have (a)
537 corrected the bug and added tests for these items, and (b) arranged for an
538 error to occur if an unknown opcode is encountered while checking for fixed
539 length instead of just assuming "not fixed length". The items that were
540 rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
541 (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
542 repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
543
544 8. A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
545 being incorrectly compiled and would have given unpredicatble results.
546
547 9. A possessively repeated subpattern with minimum repeat count greater than
548 one behaved incorrectly. For example, (A){2,}+ behaved as if it was
549 (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
550 the first (A) could occur when it should not.
551
552 10. Add a cast and remove a redundant test from the code.
553
554 11. JIT should use pcre_malloc/pcre_free for allocation.
555
556 12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
557 best practice nowadays, and helps with cross-compiling. (If the exec_prefix
558 is anything other than /usr, -L is still shown).
559
560 13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
561
562 14. Perl does not support \N without a following name in a [] class; PCRE now
563 also gives an error.
564
565 15. If a forward reference was repeated with an upper limit of around 2000,
566 it caused the error "internal error: overran compiling workspace". The
567 maximum number of forward references (including repeats) was limited by the
568 internal workspace, and dependent on the LINK_SIZE. The code has been
569 rewritten so that the workspace expands (via pcre_malloc) if necessary, and
570 the default depends on LINK_SIZE. There is a new upper limit (for safety)
571 of around 200,000 forward references. While doing this, I also speeded up
572 the filling in of repeated forward references.
573
574 16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
575 incorrectly expecting the subject to contain another "a" after the start.
576
577 17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
578 in the match, the SKIP should be ignored. This was not happening; instead
579 the SKIP was being treated as NOMATCH. For patterns such as
580 /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
581 tested.
582
583 18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
584 now much more compatible with Perl, in particular in cases where the result
585 is a non-match for a non-anchored pattern. For example, if
586 /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
587 "m", where previously it did not return a name. A side effect of this
588 change is that for partial matches, the last encountered mark name is
589 returned, as for non matches. A number of tests that were previously not
590 Perl-compatible have been moved into the Perl-compatible test files. The
591 refactoring has had the pleasing side effect of removing one argument from
592 the match() function, thus reducing its stack requirements.
593
594 19. If the /S+ option was used in pcretest to study a pattern using JIT,
595 subsequent uses of /S (without +) incorrectly behaved like /S+.
596
597 21. Retrieve executable code size support for the JIT compiler and fixing
598 some warnings.
599
600 22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
601 not work when the shorter character appeared right at the end of the
602 subject string.
603
604 23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
605 systems.
606
607 24. Added PCRE_INFO_JITSIZE to pass on the value from (21) above, and also
608 output it when the /M option is used in pcretest.
609
610 25. The CheckMan script was not being included in the distribution. Also, added
611 an explicit "perl" to run Perl scripts from the PrepareRelease script
612 because this is reportedly needed in Windows.
613
614 26. If study data was being save in a file and studying had not found a set of
615 "starts with" bytes for the pattern, the data written to the file (though
616 never used) was taken from uninitialized memory and so caused valgrind to
617 complain.
618
619 27. Updated RunTest.bat as provided by Sheri Pierce.
620
621 28. Fixed a possible uninitialized memory bug in pcre_jit_compile.c.
622
623 29. Computation of memory usage for the table of capturing group names was
624 giving an unnecessarily large value.
625
626
627 Version 8.20 21-Oct-2011
628 ------------------------
629
630 1. Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
631 a POSIX class. After further experiments with Perl, which convinced me that
632 Perl has bugs and confusions, a closing square bracket is no longer allowed
633 in a POSIX name. This bug also affected patterns with classes that started
634 with full stops.
635
636 2. If a pattern such as /(a)b|ac/ is matched against "ac", there is no
637 captured substring, but while checking the failing first alternative,
638 substring 1 is temporarily captured. If the output vector supplied to
639 pcre_exec() was not big enough for this capture, the yield of the function
640 was still zero ("insufficient space for captured substrings"). This cannot
641 be totally fixed without adding another stack variable, which seems a lot
642 of expense for a edge case. However, I have improved the situation in cases
643 such as /(a)(b)x|abc/ matched against "abc", where the return code
644 indicates that fewer than the maximum number of slots in the ovector have
645 been set.
646
647 3. Related to (2) above: when there are more back references in a pattern than
648 slots in the output vector, pcre_exec() uses temporary memory during
649 matching, and copies in the captures as far as possible afterwards. It was
650 using the entire output vector, but this conflicts with the specification
651 that only 2/3 is used for passing back captured substrings. Now it uses
652 only the first 2/3, for compatibility. This is, of course, another edge
653 case.
654
655 4. Zoltan Herczeg's just-in-time compiler support has been integrated into the
656 main code base, and can be used by building with --enable-jit. When this is
657 done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
658 runtime --no-jit option is given.
659
660 5. When the number of matches in a pcre_dfa_exec() run exactly filled the
661 ovector, the return from the function was zero, implying that there were
662 other matches that did not fit. The correct "exactly full" value is now
663 returned.
664
665 6. If a subpattern that was called recursively or as a subroutine contained
666 (*PRUNE) or any other control that caused it to give a non-standard return,
667 invalid errors such as "Error -26 (nested recursion at the same subject
668 position)" or even infinite loops could occur.
669
670 7. If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
671 computing the minimum length on reaching *ACCEPT, and so ended up with the
672 wrong value of 1 rather than 0. Further investigation indicates that
673 computing a minimum subject length in the presence of *ACCEPT is difficult
674 (think back references, subroutine calls), and so I have changed the code
675 so that no minimum is registered for a pattern that contains *ACCEPT.
676
677 8. If (*THEN) was present in the first (true) branch of a conditional group,
678 it was not handled as intended. [But see 16 below.]
679
680 9. Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
681 Sheri Pierce.
682
683 10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
684 the first byte in a match must be "a".
685
686 11. Change 17 for 8.13 increased the recursion depth for patterns like
687 /a(?:.)*?a/ drastically. I've improved things by remembering whether a
688 pattern contains any instances of (*THEN). If it does not, the old
689 optimizations are restored. It would be nice to do this on a per-group
690 basis, but at the moment that is not feasible.
691
692 12. In some environments, the output of pcretest -C is CRLF terminated. This
693 broke RunTest's code that checks for the link size. A single white space
694 character after the value is now allowed for.
695
696 13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
697 For "fr", it uses the Windows-specific input and output files.
698
699 14. If (*THEN) appeared in a group that was called recursively or as a
700 subroutine, it did not work as intended. [But see next item.]
701
702 15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
703 pattern fragments (but not containing any | characters). If A and B are
704 matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
705 was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
706 D. In other words, Perl considers parentheses that do not contain any |
707 characters to be part of a surrounding alternative, whereas PCRE was
708 treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
709 differently. PCRE now behaves in the same way as Perl, except in the case
710 of subroutine/recursion calls such as (?1) which have in any case always
711 been different (but PCRE had them first :-).
712
713 16. Related to 15 above: Perl does not treat the | in a conditional group as
714 creating alternatives. Such a group is treated in the same way as an
715 ordinary group without any | characters when processing (*THEN). PCRE has
716 been changed to match Perl's behaviour.
717
718 17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
719 RunGrepTest script failed.
720
721 18. Change 22 for version 13 caused atomic groups to use more stack. This is
722 inevitable for groups that contain captures, but it can lead to a lot of
723 stack use in large patterns. The old behaviour has been restored for atomic
724 groups that do not contain any capturing parentheses.
725
726 19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
727 suppress the check for a minimum subject length at run time. (If it was
728 given to pcre_exec() or pcre_dfa_exec() it did work.)
729
730 20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
731 fail to work when decoding hex characters in data strings in EBCDIC
732 environments.
733
734 21. It appears that in at least one Mac OS environment, the isxdigit() function
735 is implemented as a macro that evaluates to its argument more than once,
736 contravening the C 90 Standard (I haven't checked a later standard). There
737 was an instance in pcretest which caused it to go wrong when processing
738 \x{...} escapes in subject strings. The has been rewritten to avoid using
739 things like p++ in the argument of isxdigit().
740
741
742 Version 8.13 16-Aug-2011
743 ------------------------
744
745 1. The Unicode data tables have been updated to Unicode 6.0.0.
746
747 2. Two minor typos in pcre_internal.h have been fixed.
748
749 3. Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
750 pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
751 in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
752
753 4. There were a number of related bugs in the code for matching backrefences
754 caselessly in UTF-8 mode when codes for the characters concerned were
755 different numbers of bytes. For example, U+023A and U+2C65 are an upper
756 and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
757 (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
758 code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
759 2-byte code at the end of the subject (it thought there wasn't enough data
760 left).
761
762 5. Comprehensive information about what went wrong is now returned by
763 pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
764 as the output vector has at least 2 elements. The offset of the start of
765 the failing character and a reason code are placed in the vector.
766
767 6. When the UTF-8 string check fails for pcre_compile(), the offset that is
768 now returned is for the first byte of the failing character, instead of the
769 last byte inspected. This is an incompatible change, but I hope it is small
770 enough not to be a problem. It makes the returned offset consistent with
771 pcre_exec() and pcre_dfa_exec().
772
773 7. pcretest now gives a text phrase as well as the error number when
774 pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
775 failure, the offset and reason code are output.
776
777 8. When \R was used with a maximizing quantifier it failed to skip backwards
778 over a \r\n pair if the subsequent match failed. Instead, it just skipped
779 back over a single character (\n). This seems wrong (because it treated the
780 two characters as a single entity when going forwards), conflicts with the
781 documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
782 behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
783 has been changed.
784
785 9. Some internal refactoring has changed the processing so that the handling
786 of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
787 time (the PCRE_DOTALL option was changed this way some time ago: version
788 7.7 change 16). This has made it possible to abolish the OP_OPT op code,
789 which was always a bit of a fudge. It also means that there is one less
790 argument for the match() function, which reduces its stack requirements
791 slightly. This change also fixes an incompatibility with Perl: the pattern
792 (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
793
794 10. More internal refactoring has drastically reduced the number of recursive
795 calls to match() for possessively repeated groups such as (abc)++ when
796 using pcre_exec().
797
798 11. While implementing 10, a number of bugs in the handling of groups were
799 discovered and fixed:
800
801 (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
802 (a|)*(?1) gave a compile-time internal error.
803 ((a|)+)+ did not notice that the outer group could match an empty string.
804 (^a|^)+ was not marked as anchored.
805 (.*a|.*)+ was not marked as matching at start or after a newline.
806
807 12. Yet more internal refactoring has removed another argument from the match()
808 function. Special calls to this function are now indicated by setting a
809 value in a variable in the "match data" data block.
810
811 13. Be more explicit in pcre_study() instead of relying on "default" for
812 opcodes that mean there is no starting character; this means that when new
813 ones are added and accidentally left out of pcre_study(), testing should
814 pick them up.
815
816 14. The -s option of pcretest has been documented for ages as being an old
817 synonym of -m (show memory usage). I have changed it to mean "force study
818 for every regex", that is, assume /S for every regex. This is similar to -i
819 and -d etc. It's slightly incompatible, but I'm hoping nobody is still
820 using it. It makes it easier to run collections of tests with and without
821 study enabled, and thereby test pcre_study() more easily. All the standard
822 tests are now run with and without -s (but some patterns can be marked as
823 "never study" - see 20 below).
824
825 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
826 restoration of the capturing data to the outer values was not happening
827 correctly.
828
829 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
830 empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
831 pattern had matched an empty string, and so incorrectly returned a no
832 match.
833
834 17. There was optimizing code for the last branch of non-capturing parentheses,
835 and also for the obeyed branch of a conditional subexpression, which used
836 tail recursion to cut down on stack usage. Unfortunately, now that there is
837 the possibility of (*THEN) occurring in these branches, tail recursion is
838 no longer possible because the return has to be checked for (*THEN). These
839 two optimizations have therefore been removed. [But see 8.20/11 above.]
840
841 18. If a pattern containing \R was studied, it was assumed that \R always
842 matched two bytes, thus causing the minimum subject length to be
843 incorrectly computed because \R can also match just one byte.
844
845 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
846 was incorrectly computed.
847
848 20. If /S is present twice on a test pattern in pcretest input, it now
849 *disables* studying, thereby overriding the use of -s on the command line
850 (see 14 above). This is necessary for one or two tests to keep the output
851 identical in both cases.
852
853 21. When (*ACCEPT) was used in an assertion that matched an empty string and
854 PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
855
856 22. When an atomic group that contained a capturing parenthesis was
857 successfully matched, but the branch in which it appeared failed, the
858 capturing was not being forgotten if a higher numbered group was later
859 captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
860 group 1 to "a", when in fact it should be unset. This applied to multi-
861 branched capturing and non-capturing groups, repeated or not, and also to
862 positive assertions (capturing in negative assertions does not happen
863 in PCRE) and also to nested atomic groups.
864
865 23. Add the ++ qualifier feature to pcretest, to show the remainder of the
866 subject after a captured substring, to make it easier to tell which of a
867 number of identical substrings has been captured.
868
869 24. The way atomic groups are processed by pcre_exec() has been changed so that
870 if they are repeated, backtracking one repetition now resets captured
871 values correctly. For example, if ((?>(a+)b)+aabab) is matched against
872 "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
873 "aaa". Previously, it would have been "a". As part of this code
874 refactoring, the way recursive calls are handled has also been changed.
875
876 25. If an assertion condition captured any substrings, they were not passed
877 back unless some other capturing happened later. For example, if
878 (?(?=(a))a) was matched against "a", no capturing was returned.
879
880 26. When studying a pattern that contained subroutine calls or assertions,
881 the code for finding the minimum length of a possible match was handling
882 direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
883 group 1 called group 2 while simultaneously a separate group 2 called group
884 1). A stack overflow occurred in this case. I have fixed this by limiting
885 the recursion depth to 10.
886
887 27. Updated RunTest.bat in the distribution to the version supplied by Tom
888 Fortmann. This supports explicit test numbers on the command line, and has
889 argument validation and error reporting.
890
891 28. An instance of \X with an unlimited repeat could fail if at any point the
892 first character it looked at was a mark character.
893
894 29. Some minor code refactoring concerning Unicode properties and scripts
895 should reduce the stack requirement of match() slightly.
896
897 30. Added the '=' option to pcretest to check the setting of unused capturing
898 slots at the end of the pattern, which are documented as being -1, but are
899 not included in the return count.
900
901 31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
902 compiled something random. Now it gives a compile-time error (as does
903 Perl).
904
905 32. A *MARK encountered during the processing of a positive assertion is now
906 recorded and passed back (compatible with Perl).
907
908 33. If --only-matching or --colour was set on a pcregrep call whose pattern
909 had alternative anchored branches, the search for a second match in a line
910 was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
911 matched the line "0102" twice. The same bug affected patterns that started
912 with a backwards assertion. For example /\b01|\b02/ also matched "0102"
913 twice.
914
915 34. Previously, PCRE did not allow quantification of assertions. However, Perl
916 does, and because of capturing effects, quantifying parenthesized
917 assertions may at times be useful. Quantifiers are now allowed for
918 parenthesized assertions.
919
920 35. A minor code tidy in pcre_compile() when checking options for \R usage.
921
922 36. \g was being checked for fancy things in a character class, when it should
923 just be a literal "g".
924
925 37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
926 appearance of a nested POSIX class supersedes an apparent external class.
927 For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
928 unescaped square brackets may also appear as part of class names. For
929 example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
930 more like Perl. (But see 8.20/1 above.)
931
932 38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
933 was because it thought it was \N{name}, which is not supported).
934
935 39. Add minix to OS list not supporting the -S option in pcretest.
936
937 40. PCRE tries to detect cases of infinite recursion at compile time, but it
938 cannot analyze patterns in sufficient detail to catch mutual recursions
939 such as ((?1))((?2)). There is now a runtime test that gives an error if a
940 subgroup is called recursively as a subpattern for a second time at the
941 same position in the subject string. In previous releases this might have
942 been caught by the recursion limit, or it might have run out of stack.
943
944 41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
945 happen only once. PCRE was, however incorrectly giving a compile time error
946 "recursive call could loop indefinitely" because it cannot analyze the
947 pattern in sufficient detail. The compile time test no longer happens when
948 PCRE is compiling a conditional subpattern, but actual runaway loops are
949 now caught at runtime (see 40 above).
950
951 42. It seems that Perl allows any characters other than a closing parenthesis
952 to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
953 has been changed to be the same.
954
955 43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
956 as not to get warnings when autogen.sh is called. Also changed
957 AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
958
959 44. To help people who use pcregrep to scan files containing exceedingly long
960 lines, the following changes have been made:
961
962 (a) The default value of the buffer size parameter has been increased from
963 8K to 20K. (The actual buffer used is three times this size.)
964
965 (b) The default can be changed by ./configure --with-pcregrep-bufsize when
966 PCRE is built.
967
968 (c) A --buffer-size=n option has been added to pcregrep, to allow the size
969 to be set at run time.
970
971 (d) Numerical values in pcregrep options can be followed by K or M, for
972 example --buffer-size=50K.
973
974 (e) If a line being scanned overflows pcregrep's buffer, an error is now
975 given and the return code is set to 2.
976
977 45. Add a pointer to the latest mark to the callout data block.
978
979 46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
980 partial match of an empty string instead of no match. This was specific to
981 the use of ".".
982
983 47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
984 complete match instead of a partial match. This bug was dependent on both
985 the PCRE_UTF8 and PCRE_DOTALL options being set.
986
987 48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
988 starting byte set, because \b was not being ignored.
989
990
991 Version 8.12 15-Jan-2011
992 ------------------------
993
994 1. Fixed some typos in the markup of the man pages, and wrote a script that
995 checks for such things as part of the documentation building process.
996
997 2. On a big-endian 64-bit system, pcregrep did not correctly process the
998 --match-limit and --recursion-limit options (added for 8.11). In
999 particular, this made one of the standard tests fail. (The integer value
1000 went into the wrong half of a long int.)
1001
1002 3. If the --colour option was given to pcregrep with -v (invert match), it
1003 did strange things, either producing crazy output, or crashing. It should,
1004 of course, ignore a request for colour when reporting lines that do not
1005 match.
1006
1007 4. Another pcregrep bug caused similar problems if --colour was specified with
1008 -M (multiline) and the pattern match finished with a line ending.
1009
1010 5. In pcregrep, when a pattern that ended with a literal newline sequence was
1011 matched in multiline mode, the following line was shown as part of the
1012 match. This seems wrong, so I have changed it.
1013
1014 6. Another pcregrep bug in multiline mode, when --colour was specified, caused
1015 the check for further matches in the same line (so they could be coloured)
1016 to overrun the end of the current line. If another match was found, it was
1017 incorrectly shown (and then shown again when found in the next line).
1018
1019 7. If pcregrep was compiled under Windows, there was a reference to the
1020 function pcregrep_exit() before it was defined. I am assuming this was
1021 the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
1022 reported by a user. I've moved the definition above the reference.
1023
1024
1025 Version 8.11 10-Dec-2010
1026 ------------------------
1027
1028 1. (*THEN) was not working properly if there were untried alternatives prior
1029 to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
1030 backtracked to try for "b" instead of moving to the next alternative branch
1031 at the same level (in this case, to look for "c"). The Perl documentation
1032 is clear that when (*THEN) is backtracked onto, it goes to the "next
1033 alternative in the innermost enclosing group".
1034
1035 2. (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
1036 such as (A(*COMMIT)B(*THEN)C|D) any failure after matching A should
1037 result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
1038 (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
1039 (*THEN).
1040
1041 3. If \s appeared in a character class, it removed the VT character from
1042 the class, even if it had been included by some previous item, for example
1043 in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
1044 of \s, but is part of the POSIX "space" class.)
1045
1046 4. A partial match never returns an empty string (because you can always
1047 match an empty string at the end of the subject); however the checking for
1048 an empty string was starting at the "start of match" point. This has been
1049 changed to the "earliest inspected character" point, because the returned
1050 data for a partial match starts at this character. This means that, for
1051 example, /(?<=abc)def/ gives a partial match for the subject "abc"
1052 (previously it gave "no match").
1053
1054 5. Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
1055 of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
1056 previously a full match would be given. However, setting PCRE_PARTIAL_HARD
1057 has an implication that the given string is incomplete (because a partial
1058 match is preferred over a full match). For this reason, these items now
1059 give a partial match in this situation. [Aside: previously, the one case
1060 /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
1061 match rather than a full match, which was wrong by the old rules, but is
1062 now correct.]
1063
1064 6. There was a bug in the handling of #-introduced comments, recognized when
1065 PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
1066 If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
1067 UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
1068 scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
1069 but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
1070 places in pcre_compile().
1071
1072 7. Related to (6) above, when pcre_compile() was skipping #-introduced
1073 comments when looking ahead for named forward references to subpatterns,
1074 the only newline sequence it recognized was NL. It now handles newlines
1075 according to the set newline convention.
1076
1077 8. SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
1078 former, but used strtoul(), whereas pcretest avoided strtoul() but did not
1079 cater for a lack of strerror(). These oversights have been fixed.
1080
1081 9. Added --match-limit and --recursion-limit to pcregrep.
1082
1083 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
1084
1085 11. When the -o option was used, pcregrep was setting a return code of 1, even
1086 when matches were found, and --line-buffered was not being honoured.
1087
1088 12. Added an optional parentheses number to the -o and --only-matching options
1089 of pcregrep.
1090
1091 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
1092 can match an empty string. The code to do it in pcretest and pcredemo
1093 needed fixing:
1094
1095 (a) When the newline convention was "crlf", pcretest got it wrong, skipping
1096 only one byte after an empty string match just before CRLF (this case
1097 just got forgotten; "any" and "anycrlf" were OK).
1098
1099 (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
1100 mode when an empty string match preceded an ASCII character followed by
1101 a non-ASCII character. (The code for advancing by one character rather
1102 than one byte was nonsense.)
1103
1104 (c) The pcredemo.c sample program did not have any code at all to handle
1105 the cases when CRLF is a valid newline sequence.
1106
1107 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
1108 as a starting offset was within the subject string. There is now a new
1109 error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
1110 negative or greater than the length of the string. In order to test this,
1111 pcretest is extended to allow the setting of negative starting offsets.
1112
1113 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
1114 starting offset points to the beginning of a UTF-8 character was
1115 unnecessarily clumsy. I tidied it up.
1116
1117 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
1118 bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
1119
1120 17. Nobody had reported that the --include_dir option, which was added in
1121 release 7.7 should have been called --include-dir (hyphen, not underscore)
1122 for compatibility with GNU grep. I have changed it to --include-dir, but
1123 left --include_dir as an undocumented synonym, and the same for
1124 --exclude-dir, though that is not available in GNU grep, at least as of
1125 release 2.5.4.
1126
1127 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
1128 characters from a string of bytes) have been redefined so as not to use
1129 loops, in order to improve performance in some environments. At the same
1130 time, I abstracted some of the common code into auxiliary macros to save
1131 repetition (this should not affect the compiled code).
1132
1133 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
1134 compile-time error is now given if \c is not followed by an ASCII
1135 character, that is, a byte less than 128. (In EBCDIC mode, the code is
1136 different, and any byte value is allowed.)
1137
1138 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
1139 START_OPTIMIZE option, which is now allowed at compile time - but just
1140 passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
1141 to pcregrep and other applications that have no direct access to PCRE
1142 options. The new /Y option in pcretest sets this option when calling
1143 pcre_compile().
1144
1145 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
1146 back references. Groups containing recursive back references were forced to
1147 be atomic by that change, but in the case of named groups, the amount of
1148 memory required was incorrectly computed, leading to "Failed: internal
1149 error: code overflow". This has been fixed.
1150
1151 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
1152 pcretest.c, to avoid build problems in some Borland environments.
1153
1154
1155 Version 8.10 25-Jun-2010
1156 ------------------------
1157
1158 1. Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
1159 THEN.
1160
1161 2. (*ACCEPT) was not working when inside an atomic group.
1162
1163 3. Inside a character class, \B is treated as a literal by default, but
1164 faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
1165 causes the error). The code is unchanged, but I tidied the documentation.
1166
1167 4. Inside a character class, PCRE always treated \R and \X as literals,
1168 whereas Perl faults them if its -w option is set. I have changed PCRE so
1169 that it faults them when PCRE_EXTRA is set.
1170
1171 5. Added support for \N, which always matches any character other than
1172 newline. (It is the same as "." when PCRE_DOTALL is not set.)
1173
1174 6. When compiling pcregrep with newer versions of gcc which may have
1175 FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
1176 declared with attribute warn_unused_result" were given. Just casting the
1177 result to (void) does not stop the warnings; a more elaborate fudge is
1178 needed. I've used a macro to implement this.
1179
1180 7. Minor change to pcretest.c to avoid a compiler warning.
1181
1182 8. Added four artifical Unicode properties to help with an option to make
1183 \s etc use properties (see next item). The new properties are: Xan
1184 (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
1185
1186 9. Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
1187 use Unicode properties. (*UCP) at the start of a pattern can be used to set
1188 this option. Modified pcretest to add /W to test this facility. Added
1189 REG_UCP to make it available via the POSIX interface.
1190
1191 10. Added --line-buffered to pcregrep.
1192
1193 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
1194 studied, and the match started with a letter with a code point greater than
1195 127 whose first byte was different to the first byte of the other case of
1196 the letter, the other case of this starting letter was not recognized
1197 (#976).
1198
1199 12. If a pattern that was studied started with a repeated Unicode property
1200 test, for example, \p{Nd}+, there was the theoretical possibility of
1201 setting up an incorrect bitmap of starting bytes, but fortunately it could
1202 not have actually happened in practice until change 8 above was made (it
1203 added property types that matched character-matching opcodes).
1204
1205 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
1206 possible starting bytes for non-anchored patterns.
1207
1208 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
1209 \R, and also a number of cases that involve Unicode properties, both
1210 explicit and implicit when PCRE_UCP is set.
1211
1212 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
1213 input, it could crash or give wrong results if characters with values
1214 greater than 0xc0 were present in the subject string. (Detail: it assumed
1215 UTF-8 input when processing these items.)
1216
1217 16. Added a lot of (int) casts to avoid compiler warnings in systems where
1218 size_t is 64-bit (#991).
1219
1220 17. Added a check for running out of memory when PCRE is compiled with
1221 --disable-stack-for-recursion (#990).
1222
1223 18. If the last data line in a file for pcretest does not have a newline on
1224 the end, a newline was missing in the output.
1225
1226 19. The default pcre_chartables.c file recognizes only ASCII characters (values
1227 less than 128) in its various bitmaps. However, there is a facility for
1228 generating tables according to the current locale when PCRE is compiled. It
1229 turns out that in some environments, 0x85 and 0xa0, which are Unicode space
1230 characters, are recognized by isspace() and therefore were getting set in
1231 these tables, and indeed these tables seem to approximate to ISO 8859. This
1232 caused a problem in UTF-8 mode when pcre_study() was used to create a list
1233 of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
1234 which of course cannot start UTF-8 characters. I have changed the code so
1235 that only real ASCII characters (less than 128) and the correct starting
1236 bytes for UTF-8 encodings are set for characters greater than 127 when in
1237 UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
1238 altogether.)
1239
1240 20. Added the /T option to pcretest so as to be able to run tests with non-
1241 standard character tables, thus making it possible to include the tests
1242 used for 19 above in the standard set of tests.
1243
1244 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
1245 reference to a subpattern the other side of a comment that contains an
1246 opening parenthesis caused either an internal compiling error, or a
1247 reference to the wrong subpattern.
1248
1249
1250 Version 8.02 19-Mar-2010
1251 ------------------------
1252
1253 1. The Unicode data tables have been updated to Unicode 5.2.0.
1254
1255 2. Added the option --libs-cpp to pcre-config, but only when C++ support is
1256 configured.
1257
1258 3. Updated the licensing terms in the pcregexp.pas file, as agreed with the
1259 original author of that file, following a query about its status.
1260
1261 4. On systems that do not have stdint.h (e.g. Solaris), check for and include
1262 inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
1263
1264 5. A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
1265 quantifier applied to a forward-referencing subroutine call, could compile
1266 incorrect code or give the error "internal error: previously-checked
1267 referenced subpattern not found".
1268
1269 6. Both MS Visual Studio and Symbian OS have problems with initializing
1270 variables to point to external functions. For these systems, therefore,
1271 pcre_malloc etc. are now initialized to local functions that call the
1272 relevant global functions.
1273
1274 7. There were two entries missing in the vectors called coptable and poptable
1275 in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
1276 I've fixed the data, and added a kludgy way of testing at compile time that
1277 the lengths are correct (equal to the number of opcodes).
1278
1279 8. Following on from 7, I added a similar kludge to check the length of the
1280 eint vector in pcreposix.c.
1281
1282 9. Error texts for pcre_compile() are held as one long string to avoid too
1283 much relocation at load time. To find a text, the string is searched,
1284 counting zeros. There was no check for running off the end of the string,
1285 which could happen if a new error number was added without updating the
1286 string.
1287
1288 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
1289
1290 11. \K was not working if it appeared in an atomic group or in a group that
1291 was called as a "subroutine", or in an assertion. Perl 5.11 documents that
1292 \K is "not well defined" if used in an assertion. PCRE now accepts it if
1293 the assertion is positive, but not if it is negative.
1294
1295 12. Change 11 fortuitously reduced the size of the stack frame used in the
1296 "match()" function of pcre_exec.c by one pointer. Forthcoming
1297 implementation of support for (*MARK) will need an extra pointer on the
1298 stack; I have reserved it now, so that the stack frame size does not
1299 decrease.
1300
1301 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
1302 item in branch that calls a recursion is a subroutine call - as in the
1303 second branch in the above example - was incorrectly given the compile-
1304 time error "recursive call could loop indefinitely" because pcre_compile()
1305 was not correctly checking the subroutine for matching a non-empty string.
1306
1307 14. The checks for overrunning compiling workspace could trigger after an
1308 overrun had occurred. This is a "should never occur" error, but it can be
1309 triggered by pathological patterns such as hundreds of nested parentheses.
1310 The checks now trigger 100 bytes before the end of the workspace.
1311
1312 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
1313
1314
1315 Version 8.01 19-Jan-2010
1316 ------------------------
1317
1318 1. If a pattern contained a conditional subpattern with only one branch (in
1319 particular, this includes all (*DEFINE) patterns), a call to pcre_study()
1320 computed the wrong minimum data length (which is of course zero for such
1321 subpatterns). This could cause incorrect "no match" results.
1322
1323 2. For patterns such as (?i)a(?-i)b|c where an option setting at the start of
1324 the pattern is reset in the first branch, pcre_compile() failed with
1325 "internal error: code overflow at offset...". This happened only when
1326 the reset was to the original external option setting. (An optimization
1327 abstracts leading options settings into an external setting, which was the
1328 cause of this.)
1329
1330 3. A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
1331 of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
1332 assertion pattern did not match (meaning that the assertion was true), it
1333 was incorrectly treated as false if the SKIP had been reached during the
1334 matching. This also applied to assertions used as conditions.
1335
1336 4. If an item that is not supported by pcre_dfa_exec() was encountered in an
1337 assertion subpattern, including such a pattern used as a condition,
1338 unpredictable results occurred, instead of the error return
1339 PCRE_ERROR_DFA_UITEM.
1340
1341 5. The C++ GlobalReplace function was not working like Perl for the special
1342 situation when an empty string is matched. It now does the fancy magic
1343 stuff that is necessary.
1344
1345 6. In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
1346 removed. (These were left over from very, very early versions of PCRE.)
1347
1348 7. Some cosmetic changes to the code to make life easier when compiling it
1349 as part of something else:
1350
1351 (a) Change DEBUG to PCRE_DEBUG.
1352
1353 (b) In pcre_compile(), rename the member of the "branch_chain" structure
1354 called "current" as "current_branch", to prevent a collision with the
1355 Linux macro when compiled as a kernel module.
1356
1357 (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
1358 prevent a collision with the Linux macro when compiled as a kernel
1359 module.
1360
1361 8. In pcre_compile() there are some checks for integer overflows that used to
1362 cast potentially large values to (double). This has been changed to that
1363 when building, a check for int64_t is made, and if it is found, it is used
1364 instead, thus avoiding the use of floating point arithmetic. (There is no
1365 other use of FP in PCRE.) If int64_t is not found, the fallback is to
1366 double.
1367
1368 9. Added two casts to avoid signed/unsigned warnings from VS Studio Express
1369 2005 (difference between two addresses compared to an unsigned value).
1370
1371 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
1372 custom one, because of the following reported problem in Windows:
1373
1374 - libbz2 uses the Pascal calling convention (WINAPI) for the functions
1375 under Win32.
1376 - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
1377 therefore missing the function definition.
1378 - The compiler thus generates a "C" signature for the test function.
1379 - The linker fails to find the "C" function.
1380 - PCRE fails to configure if asked to do so against libbz2.
1381
1382 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
1383 messages were output:
1384
1385 Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
1386 rerunning libtoolize, to keep the correct libtool macros in-tree.
1387 Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
1388
1389 I have done both of these things.
1390
1391 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
1392 most of the time, it *can* run out if it is given a pattern that contains a
1393 runaway infinite recursion. I updated the discussion in the pcrestack man
1394 page.
1395
1396 13. Now that we have gone to the x.xx style of version numbers, the minor
1397 version may start with zero. Using 08 or 09 is a bad idea because users
1398 might check the value of PCRE_MINOR in their code, and 08 or 09 may be
1399 interpreted as invalid octal numbers. I've updated the previous comment in
1400 configure.ac, and also added a check that gives an error if 08 or 09 are
1401 used.
1402
1403 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
1404 causing partial matching to fail when the end of the subject matched \W
1405 in a UTF-8 pattern where \W was quantified with a minimum of 3.
1406
1407 15. There were some discrepancies between the declarations in pcre_internal.h
1408 of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
1409 their definitions. The declarations used "const uschar *" and the
1410 definitions used USPTR. Even though USPTR is normally defined as "const
1411 unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
1412 reported that: "This difference in casting confuses some C++ compilers, for
1413 example, SunCC recognizes above declarations as different functions and
1414 generates broken code for hbpcre." I have changed the declarations to use
1415 USPTR.
1416
1417 16. GNU libtool is named differently on some systems. The autogen.sh script now
1418 tries several variants such as glibtoolize (MacOSX) and libtoolize1x
1419 (FreeBSD).
1420
1421 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
1422 (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
1423 comment: "Figure out how to create a longlong from a string: strtoll and
1424 equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
1425 instance, but it only takes 2 args instead of 3!"
1426
1427 18. A subtle bug concerned with back references has been fixed by a change of
1428 specification, with a corresponding code fix. A pattern such as
1429 ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
1430 refers, was giving matches when it shouldn't. For example, xa=xaaa would
1431 match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
1432 same bug. Such groups have to be quantified to be useful, or contained
1433 inside another quantified group. (If there's no repetition, the reference
1434 can never match.) The problem arises because, having left the group and
1435 moved on to the rest of the pattern, a later failure that backtracks into
1436 the group uses the captured value from the final iteration of the group
1437 rather than the correct earlier one. I have fixed this in PCRE by forcing
1438 any group that contains a reference to itself to be an atomic group; that
1439 is, there cannot be any backtracking into it once it has completed. This is
1440 similar to recursive and subroutine calls.
1441
1442
1443 Version 8.00 19-Oct-09
1444 ----------------------
1445
1446 1. The table for translating pcre_compile() error codes into POSIX error codes
1447 was out-of-date, and there was no check on the pcre_compile() error code
1448 being within the table. This could lead to an OK return being given in
1449 error.
1450
1451 2. Changed the call to open a subject file in pcregrep from fopen(pathname,
1452 "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
1453 in a Windows environment.
1454
1455 3. The pcregrep --count option prints the count for each file even when it is
1456 zero, as does GNU grep. However, pcregrep was also printing all files when
1457 --files-with-matches was added. Now, when both options are given, it prints
1458 counts only for those files that have at least one match. (GNU grep just
1459 prints the file name in this circumstance, but including the count seems
1460 more useful - otherwise, why use --count?) Also ensured that the
1461 combination -clh just lists non-zero counts, with no names.
1462
1463 4. The long form of the pcregrep -F option was incorrectly implemented as
1464 --fixed_strings instead of --fixed-strings. This is an incompatible change,
1465 but it seems right to fix it, and I didn't think it was worth preserving
1466 the old behaviour.
1467
1468 5. The command line items --regex=pattern and --regexp=pattern were not
1469 recognized by pcregrep, which required --regex pattern or --regexp pattern
1470 (with a space rather than an '='). The man page documented the '=' forms,
1471 which are compatible with GNU grep; these now work.
1472
1473 6. No libpcreposix.pc file was created for pkg-config; there was just
1474 libpcre.pc and libpcrecpp.pc. The omission has been rectified.
1475
1476 7. Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
1477 when UCP support is not needed, by modifying the Python script that
1478 generates it from Unicode data files. This should not matter if the module
1479 is correctly used as a library, but I received one complaint about 50K of
1480 unwanted data. My guess is that the person linked everything into his
1481 program rather than using a library. Anyway, it does no harm.
1482
1483 8. A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
1484 was a minimum greater than 1 for a wide character in a possessive
1485 repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
1486 which had an unlimited repeat of a nested, fixed maximum repeat of a wide
1487 character. Chaos in the form of incorrect output or a compiling loop could
1488 result.
1489
1490 9. The restrictions on what a pattern can contain when partial matching is
1491 requested for pcre_exec() have been removed. All patterns can now be
1492 partially matched by this function. In addition, if there are at least two
1493 slots in the offset vector, the offset of the earliest inspected character
1494 for the match and the offset of the end of the subject are set in them when
1495 PCRE_ERROR_PARTIAL is returned.
1496
1497 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
1498 synonymous with PCRE_PARTIAL, for backwards compatibility, and
1499 PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
1500 and may be more useful for multi-segment matching.
1501
1502 11. Partial matching with pcre_exec() is now more intuitive. A partial match
1503 used to be given if ever the end of the subject was reached; now it is
1504 given only if matching could not proceed because another character was
1505 needed. This makes a difference in some odd cases such as Z(*FAIL) with the
1506 string "Z", which now yields "no match" instead of "partial match". In the
1507 case of pcre_dfa_exec(), "no match" is given if every matching path for the
1508 final character ended with (*FAIL).
1509
1510 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
1511 if the pattern had a "must contain" character that was already found in the
1512 earlier partial match, unless partial matching was again requested. For
1513 example, with the pattern /dog.(body)?/, the "must contain" character is
1514 "g". If the first part-match was for the string "dog", restarting with
1515 "sbody" failed. This bug has been fixed.
1516
1517 13. The string returned by pcre_dfa_exec() after a partial match has been
1518 changed so that it starts at the first inspected character rather than the
1519 first character of the match. This makes a difference only if the pattern
1520 starts with a lookbehind assertion or \b or \B (\K is not supported by
1521 pcre_dfa_exec()). It's an incompatible change, but it makes the two
1522 matching functions compatible, and I think it's the right thing to do.
1523
1524 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
1525 so that the demonstration program is easily available in environments where
1526 PCRE has not been installed from source.
1527
1528 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
1529 libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
1530 library.
1531
1532 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
1533 It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
1534 is not the first non-POSIX option to be added. Clearly some people find
1535 these options useful.
1536
1537 17. If a caller to the POSIX matching function regexec() passes a non-zero
1538 value for nmatch with a NULL value for pmatch, the value of
1539 nmatch is forced to zero.
1540
1541 18. RunGrepTest did not have a test for the availability of the -u option of
1542 the diff command, as RunTest does. It now checks in the same way as
1543 RunTest, and also checks for the -b option.
1544
1545 19. If an odd number of negated classes containing just a single character
1546 interposed, within parentheses, between a forward reference to a named
1547 subpattern and the definition of the subpattern, compilation crashed with
1548 an internal error, complaining that it could not find the referenced
1549 subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
1550 [The bug was that it was starting one character too far in when skipping
1551 over the character class, thus treating the ] as data rather than
1552 terminating the class. This meant it could skip too much.]
1553
1554 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
1555 /g option in pcretest when the pattern contains \K, which makes it possible
1556 to have an empty string match not at the start, even when the pattern is
1557 anchored. Updated pcretest and pcredemo to use this option.
1558
1559 21. If the maximum number of capturing subpatterns in a recursion was greater
1560 than the maximum at the outer level, the higher number was returned, but
1561 with unset values at the outer level. The correct (outer level) value is
1562 now given.
1563
1564 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
1565 PCRE did not set those parentheses (unlike Perl). I have now found a way to
1566 make it do so. The string so far is captured, making this feature
1567 compatible with Perl.
1568
1569 23. The tests have been re-organized, adding tests 11 and 12, to make it
1570 possible to check the Perl 5.10 features against Perl 5.10.
1571
1572 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
1573 pattern matches a fixed length string. PCRE did not allow this; now it
1574 does. Neither allows recursion.
1575
1576 25. I finally figured out how to implement a request to provide the minimum
1577 length of subject string that was needed in order to match a given pattern.
1578 (It was back references and recursion that I had previously got hung up
1579 on.) This code has now been added to pcre_study(); it finds a lower bound
1580 to the length of subject needed. It is not necessarily the greatest lower
1581 bound, but using it to avoid searching strings that are too short does give
1582 some useful speed-ups. The value is available to calling programs via
1583 pcre_fullinfo().
1584
1585 26. While implementing 25, I discovered to my embarrassment that pcretest had
1586 not been passing the result of pcre_study() to pcre_dfa_exec(), so the
1587 study optimizations had never been tested with that matching function.
1588 Oops. What is worse, even when it was passed study data, there was a bug in
1589 pcre_dfa_exec() that meant it never actually used it. Double oops. There
1590 were also very few tests of studied patterns with pcre_dfa_exec().
1591
1592 27. If (?| is used to create subpatterns with duplicate numbers, they are now
1593 allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
1594 on the other side of the coin, they are no longer allowed to have different
1595 names, because these cannot be distinguished in PCRE, and this has caused
1596 confusion. (This is a difference from Perl.)
1597
1598 28. When duplicate subpattern names are present (necessarily with different
1599 numbers, as required by 27 above), and a test is made by name in a
1600 conditional pattern, either for a subpattern having been matched, or for
1601 recursion in such a pattern, all the associated numbered subpatterns are
1602 tested, and the overall condition is true if the condition is true for any
1603 one of them. This is the way Perl works, and is also more like the way
1604 testing by number works.
1605
1606
1607 Version 7.9 11-Apr-09
1608 ---------------------
1609
1610 1. When building with support for bzlib/zlib (pcregrep) and/or readline
1611 (pcretest), all targets were linked against these libraries. This included
1612 libpcre, libpcreposix, and libpcrecpp, even though they do not use these
1613 libraries. This caused unwanted dependencies to be created. This problem
1614 has been fixed, and now only pcregrep is linked with bzlib/zlib and only
1615 pcretest is linked with readline.
1616
1617 2. The "typedef int BOOL" in pcre_internal.h that was included inside the
1618 "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
1619 moved outside it again, because FALSE and TRUE are already defined in AIX,
1620 but BOOL is not.
1621
1622 3. The pcre_config() function was treating the PCRE_MATCH_LIMIT and
1623 PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
1624
1625 4. The pcregrep documentation said spaces were inserted as well as colons (or
1626 hyphens) following file names and line numbers when outputting matching
1627 lines. This is not true; no spaces are inserted. I have also clarified the
1628 wording for the --colour (or --color) option.
1629
1630 5. In pcregrep, when --colour was used with -o, the list of matching strings
1631 was not coloured; this is different to GNU grep, so I have changed it to be
1632 the same.
1633
1634 6. When --colo(u)r was used in pcregrep, only the first matching substring in
1635 each matching line was coloured. Now it goes on to look for further matches
1636 of any of the test patterns, which is the same behaviour as GNU grep.
1637
1638 7. A pattern that could match an empty string could cause pcregrep to loop; it
1639 doesn't make sense to accept an empty string match in pcregrep, so I have
1640 locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
1641 seems to be how GNU grep behaves.
1642
1643 8. The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
1644 start or after a newline", because the conditional assertion was not being
1645 correctly handled. The rule now is that both the assertion and what follows
1646 in the first alternative must satisfy the test.
1647
1648 9. If auto-callout was enabled in a pattern with a conditional group whose
1649 condition was an assertion, PCRE could crash during matching, both with
1650 pcre_exec() and pcre_dfa_exec().
1651
1652 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
1653 used for matching.
1654
1655 11. Unicode property support in character classes was not working for
1656 characters (bytes) greater than 127 when not in UTF-8 mode.
1657
1658 12. Added the -M command line option to pcretest.
1659
1660 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
1661
1662 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
1663
1664 16. Added comments and documentation about mis-use of no_arg in the C++
1665 wrapper.
1666
1667 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
1668 from Martin Jerabek that uses macro names for all relevant character and
1669 string constants.
1670
1671 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
1672 SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
1673 SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
1674 these, but not everybody uses configure.
1675
1676 19. A conditional group that had only one branch was not being correctly
1677 recognized as an item that could match an empty string. This meant that an
1678 enclosing group might also not be so recognized, causing infinite looping
1679 (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
1680 with the subject "ab", where knowledge that the repeated group can match
1681 nothing is needed in order to break the loop.
1682
1683 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
1684 exec(), but without supplying a callout function, matching went wrong.
1685
1686 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
1687 leak if the size of the offset vector was greater than 30. When the vector
1688 is smaller, the saved offsets during recursion go onto a local stack
1689 vector, but for larger vectors malloc() is used. It was failing to free
1690 when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
1691 error, in fact).
1692
1693 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
1694 heapframe that is used only when UTF-8 support is enabled. This caused no
1695 problem, but was untidy.
1696
1697 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
1698 CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
1699 included within another project.
1700
1701 24. Steven Van Ingelgem's patches to add more options to the CMake support,
1702 slightly modified by me:
1703
1704 (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
1705 not building pcregrep.
1706
1707 (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
1708 if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
1709
1710 25. Forward references, both numeric and by name, in patterns that made use of
1711 duplicate group numbers, could behave incorrectly or give incorrect errors,
1712 because when scanning forward to find the reference group, PCRE was not
1713 taking into account the duplicate group numbers. A pattern such as
1714 ^X(?3)(a)(?|(b)|(q))(Y) is an example.
1715
1716 26. Changed a few more instances of "const unsigned char *" to USPTR, making
1717 the feature of a custom pointer more persuasive (as requested by a user).
1718
1719 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
1720 pcretest.c, inside #ifndefs, because it seems they are sometimes already
1721 pre-defined.
1722
1723 28. Added support for (*UTF8) at the start of a pattern.
1724
1725 29. Arrange for flags added by the "release type" setting in CMake to be shown
1726 in the configuration summary.
1727
1728
1729 Version 7.8 05-Sep-08
1730 ---------------------
1731
1732 1. Replaced UCP searching code with optimized version as implemented for Ad
1733 Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
1734 stage table and inline lookup instead of a function, giving speed ups of 2
1735 to 5 times on some simple patterns that I tested. Permission was given to
1736 distribute the MultiStage2.py script that generates the tables (it's not in
1737 the tarball, but is in the Subversion repository).
1738
1739 2. Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
1740 scripts.
1741
1742 3. Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
1743 a group with a zero qualifier. The result of the study could be incorrect,
1744 or the function might crash, depending on the pattern.
1745
1746 4. Caseless matching was not working for non-ASCII characters in back
1747 references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
1748 It now works when Unicode Property Support is available.
1749
1750 5. In pcretest, an escape such as \x{de} in the data was always generating
1751 a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
1752 non-UTF-8 mode. If the value is greater than 255, it gives a warning about
1753 truncation.
1754
1755 6. Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
1756
1757 7. Added two (int) casts to pcregrep when printing the difference of two
1758 pointers, in case they are 64-bit values.
1759
1760 8. Added comments about Mac OS X stack usage to the pcrestack man page and to
1761 test 2 if it fails.
1762
1763 9. Added PCRE_CALL_CONVENTION just before the names of all exported functions,
1764 and a #define of that name to empty if it is not externally set. This is to
1765 allow users of MSVC to set it if necessary.
1766
1767 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
1768 the convenience functions in the pcre_get.c source file.
1769
1770 11. An option change at the start of a pattern that had top-level alternatives
1771 could cause overwriting and/or a crash. This command provoked a crash in
1772 some environments:
1773
1774 printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
1775
1776 This potential security problem was recorded as CVE-2008-2371.
1777
1778 12. For a pattern where the match had to start at the beginning or immediately
1779 after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
1780 pcre_dfa_exec() could read past the end of the passed subject if there was
1781 no match. To help with detecting such bugs (e.g. with valgrind), I modified
1782 pcretest so that it places the subject at the end of its malloc-ed buffer.
1783
1784 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
1785 exec() might read past the end of the data buffer in UTF-8 mode.
1786
1787 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
1788 the data contained the byte 0x85 as part of a UTF-8 character within its
1789 first line. This applied both to normal and DFA matching.
1790
1791 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
1792 /^[^d]*?$/8 failed to match "abc".
1793
1794 16. Added a missing copyright notice to pcrecpp_internal.h.
1795
1796 17. Make it more clear in the documentation that values returned from
1797 pcre_exec() in ovector are byte offsets, not character counts.
1798
1799 18. Tidied a few places to stop certain compilers from issuing warnings.
1800
1801 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
1802 supplied by Stefan Weber. I made a further small update for 7.8 because
1803 there is a change of source arrangements: the pcre_searchfuncs.c module is
1804 replaced by pcre_ucd.c.
1805
1806
1807 Version 7.7 07-May-08
1808 ---------------------
1809
1810 1. Applied Craig's patch to sort out a long long problem: "If we can't convert
1811 a string to a long long, pretend we don't even have a long long." This is
1812 done by checking for the strtoq, strtoll, and _strtoi64 functions.
1813
1814 2. Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
1815 pre-7.6 versions, which defined a global no_arg variable instead of putting
1816 it in the RE class. (See also #8 below.)
1817
1818 3. Remove a line of dead code, identified by coverity and reported by Nuno
1819 Lopes.
1820
1821 4. Fixed two related pcregrep bugs involving -r with --include or --exclude:
1822
1823 (1) The include/exclude patterns were being applied to the whole pathnames
1824 of files, instead of just to the final components.
1825
1826 (2) If there was more than one level of directory, the subdirectories were
1827 skipped unless they satisfied the include/exclude conditions. This is
1828 inconsistent with GNU grep (and could even be seen as contrary to the
1829 pcregrep specification - which I improved to make it absolutely clear).
1830 The action now is always to scan all levels of directory, and just
1831 apply the include/exclude patterns to regular files.
1832
1833 5. Added the --include_dir and --exclude_dir patterns to pcregrep, and used
1834 --exclude_dir in the tests to avoid scanning .svn directories.
1835
1836 6. Applied Craig's patch to the QuoteMeta function so that it escapes the
1837 NUL character as backslash + 0 rather than backslash + NUL, because PCRE
1838 doesn't support NULs in patterns.
1839
1840 7. Added some missing "const"s to declarations of static tables in
1841 pcre_compile.c and pcre_dfa_exec.c.
1842
1843 8. Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
1844 caused by fix #2 above. (Subsequently also a second patch to fix the
1845 first patch. And a third patch - this was a messy problem.)
1846
1847 9. Applied Craig's patch to remove the use of push_back().
1848
1849 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
1850 matching function regexec().
1851
1852 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
1853 which, however, unlike Perl's \g{...}, are subroutine calls, not back
1854 references. PCRE supports relative numbers with this syntax (I don't think
1855 Oniguruma does).
1856
1857 12. Previously, a group with a zero repeat such as (...){0} was completely
1858 omitted from the compiled regex. However, this means that if the group
1859 was called as a subroutine from elsewhere in the pattern, things went wrong
1860 (an internal error was given). Such groups are now left in the compiled
1861 pattern, with a new opcode that causes them to be skipped at execution
1862 time.
1863
1864 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
1865 to the way PCRE behaves:
1866
1867 (a) A lone ] character is dis-allowed (Perl treats it as data).
1868
1869 (b) A back reference to an unmatched subpattern matches an empty string
1870 (Perl fails the current match path).
1871
1872 (c) A data ] in a character class must be notated as \] because if the
1873 first data character in a class is ], it defines an empty class. (In
1874 Perl it is not possible to have an empty class.) The empty class []
1875 never matches; it forces failure and is equivalent to (*FAIL) or (?!).
1876 The negative empty class [^] matches any one character, independently
1877 of the DOTALL setting.
1878
1879 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
1880 non-existent subpattern following a character class starting with ']' and
1881 containing () gave an internal compiling error instead of "reference to
1882 non-existent subpattern". Fortunately, when the pattern did exist, the
1883 compiled code was correct. (When scanning forwards to check for the
1884 existencd of the subpattern, it was treating the data ']' as terminating
1885 the class, so got the count wrong. When actually compiling, the reference
1886 was subsequently set up correctly.)
1887
1888 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
1889 it was being rejected as not supported by pcre_dfa_exec(), even though
1890 other assertions are supported. I have made pcre_dfa_exec() support
1891 (*FAIL).
1892
1893 16. The implementation of 13c above involved the invention of a new opcode,
1894 OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
1895 cannot be changed at match time, I realized I could make a small
1896 improvement to matching performance by compiling OP_ALLANY instead of
1897 OP_ANY for "." when DOTALL was set, and then removing the runtime tests
1898 on the OP_ANY path.
1899
1900 17. Compiling pcretest on Windows with readline support failed without the
1901 following two fixes: (1) Make the unistd.h include conditional on
1902 HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
1903
1904 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
1905 ncurses library to be included for pcretest when ReadLine support is
1906 requested, but also to allow for it to be overridden. This patch came from
1907 Daniel Bergström.
1908
1909 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
1910 as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
1911 any errors with the current Unicode tables. Thanks to Peter Kankowski for
1912 spotting this.
1913
1914
1915 Version 7.6 28-Jan-08
1916 ---------------------
1917
1918 1. A character class containing a very large number of characters with
1919 codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
1920 overflow.
1921
1922 2. Patch to cut out the "long long" test in pcrecpp_unittest when
1923 HAVE_LONG_LONG is not defined.
1924
1925 3. Applied Christian Ehrlicher's patch to update the CMake build files to
1926 bring them up to date and include new features. This patch includes:
1927
1928 - Fixed PH's badly added libz and libbz2 support.
1929 - Fixed a problem with static linking.
1930 - Added pcredemo. [But later removed - see 7 below.]
1931 - Fixed dftables problem and added an option.
1932 - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
1933 HAVE_LONG_LONG.
1934 - Added readline support for pcretest.
1935 - Added an listing of the option settings after cmake has run.
1936
1937 4. A user submitted a patch to Makefile that makes it easy to create
1938 "pcre.dll" under mingw when using Configure/Make. I added stuff to
1939 Makefile.am that cause it to include this special target, without
1940 affecting anything else. Note that the same mingw target plus all
1941 the other distribution libraries and programs are now supported
1942 when configuring with CMake (see 6 below) instead of with
1943 Configure/Make.
1944
1945 5. Applied Craig's patch that moves no_arg into the RE class in the C++ code.
1946 This is an attempt to solve the reported problem "pcrecpp::no_arg is not
1947 exported in the Windows port". It has not yet been confirmed that the patch
1948 solves the problem, but it does no harm.
1949
1950 6. Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
1951 NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
1952 with CMake, and also correct the comment about stack recursion.
1953
1954 7. Remove the automatic building of pcredemo from the ./configure system and
1955 from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
1956 of a program that users should build themselves after PCRE is installed, so
1957 building it automatically is not really right. What is more, it gave
1958 trouble in some build environments.
1959
1960 8. Further tidies to CMakeLists.txt from Sheri and Christian.
1961
1962
1963 Version 7.5 10-Jan-08
1964 ---------------------
1965
1966 1. Applied a patch from Craig: "This patch makes it possible to 'ignore'
1967 values in parens when parsing an RE using the C++ wrapper."
1968
1969 2. Negative specials like \S did not work in character classes in UTF-8 mode.
1970 Characters greater than 255 were excluded from the class instead of being
1971 included.
1972
1973 3. The same bug as (2) above applied to negated POSIX classes such as
1974 [:^space:].
1975
1976 4. PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
1977 defined or documented. It seems to have been a typo for PCRE_STATIC, so
1978 I have changed it.
1979
1980 5. The construct (?&) was not diagnosed as a syntax error (it referenced the
1981 first named subpattern) and a construct such as (?&a) would reference the
1982 first named subpattern whose name started with "a" (in other words, the
1983 length check was missing). Both these problems are fixed. "Subpattern name
1984 expected" is now given for (?&) (a zero-length name), and this patch also
1985 makes it give the same error for \k'' (previously it complained that that
1986 was a reference to a non-existent subpattern).
1987
1988 6. The erroneous patterns (?+-a) and (?-+a) give different error messages;
1989 this is right because (?- can be followed by option settings as well as by
1990 digits. I have, however, made the messages clearer.
1991
1992 7. Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
1993 than the number used in the conditional) now cause a compile-time error.
1994 This is actually not compatible with Perl, which accepts such patterns, but
1995 treats the conditional as always being FALSE (as PCRE used to), but it
1996 seems to me that giving a diagnostic is better.
1997
1998 8. Change "alphameric" to the more common word "alphanumeric" in comments
1999 and messages.
2000
2001 9. Fix two occurrences of "backslash" in comments that should have been
2002 "backspace".
2003
2004 10. Remove two redundant lines of code that can never be obeyed (their function
2005 was moved elsewhere).
2006
2007 11. The program that makes PCRE's Unicode character property table had a bug
2008 which caused it to generate incorrect table entries for sequences of
2009 characters that have the same character type, but are in different scripts.
2010 It amalgamated them into a single range, with the script of the first of
2011 them. In other words, some characters were in the wrong script. There were
2012 thirteen such cases, affecting characters in the following ranges:
2013
2014 U+002b0 - U+002c1
2015 U+0060c - U+0060d
2016 U+0061e - U+00612
2017 U+0064b - U+0065e
2018 U+0074d - U+0076d
2019 U+01800 - U+01805
2020 U+01d00 - U+01d77
2021 U+01d9b - U+01dbf
2022 U+0200b - U+0200f
2023 U+030fc - U+030fe
2024 U+03260 - U+0327f
2025 U+0fb46 - U+0fbb1
2026 U+10450 - U+1049d
2027
2028 12. The -o option (show only the matching part of a line) for pcregrep was not
2029 compatible with GNU grep in that, if there was more than one match in a
2030 line, it showed only the first of them. It now behaves in the same way as
2031 GNU grep.
2032
2033 13. If the -o and -v options were combined for pcregrep, it printed a blank
2034 line for every non-matching line. GNU grep prints nothing, and pcregrep now
2035 does the same. The return code can be used to tell if there were any
2036 non-matching lines.
2037
2038 14. Added --file-offsets and --line-offsets to pcregrep.
2039
2040 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
2041 infinitely looping recursion. The bug was that positive lookaheads were not
2042 being skipped when checking for a possible empty match (negative lookaheads
2043 and both kinds of lookbehind were skipped).
2044
2045 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
2046 inclusion of <windows.h> to before rather than after the definition of
2047 INVALID_FILE_ATTRIBUTES (patch from David Byron).
2048
2049 17. Specifying a possessive quantifier with a specific limit for a Unicode
2050 character property caused pcre_compile() to compile bad code, which led at
2051 runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
2052 are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
2053 caused the error; without that there was no problem.
2054
2055 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
2056
2057 19. Added --enable-pcretest-libreadline.
2058
2059 20. In pcrecpp.cc, the variable 'count' was incremented twice in
2060 RE::GlobalReplace(). As a result, the number of replacements returned was
2061 double what it should be. I removed one of the increments, but Craig sent a
2062 later patch that removed the other one (the right fix) and added unit tests
2063 that check the return values (which was not done before).
2064
2065 21. Several CMake things:
2066
2067 (1) Arranged that, when cmake is used on Unix, the libraries end up with
2068 the names libpcre and libpcreposix, not just pcre and pcreposix.
2069
2070 (2) The above change means that pcretest and pcregrep are now correctly
2071 linked with the newly-built libraries, not previously installed ones.
2072
2073 (3) Added PCRE_SUPPORT_LIBREADLINE, PCRE_SUPPORT_LIBZ, PCRE_SUPPORT_LIBBZ2.
2074
2075 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
2076 crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
2077 UTF-8 newline character). The key issue is that the pattern starts .*;
2078 this means that the match must be either at the beginning, or after a
2079 newline. The bug was in the code for advancing after a failed match and
2080 checking that the new position followed a newline. It was not taking
2081 account of UTF-8 characters correctly.
2082
2083 23. PCRE was behaving differently from Perl in the way it recognized POSIX
2084 character classes. PCRE was not treating the sequence [:...:] as a
2085 character class unless the ... were all letters. Perl, however, seems to
2086 allow any characters between [: and :], though of course it rejects as
2087 unknown any "names" that contain non-letters, because all the known class
2088 names consist only of letters. Thus, Perl gives an error for [[:1234:]],
2089 for example, whereas PCRE did not - it did not recognize a POSIX character
2090 class. This seemed a bit dangerous, so the code has been changed to be
2091 closer to Perl. The behaviour is not identical to Perl, because PCRE will
2092 diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
2093 treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
2094 Perl does, and where it didn't before.
2095
2096 24. Rewrite so as to remove the single use of %n from pcregrep because in some
2097 Windows environments %n is disabled by default.
2098
2099
2100 Version 7.4 21-Sep-07
2101 ---------------------
2102
2103 1. Change 7.3/28 was implemented for classes by looking at the bitmap. This
2104 means that a class such as [\s] counted as "explicit reference to CR or
2105 LF". That isn't really right - the whole point of the change was to try to
2106 help when there was an actual mention of one of the two characters. So now
2107 the change happens only if \r or \n (or a literal CR or LF) character is
2108 encountered.
2109
2110 2. The 32-bit options word was also used for 6 internal flags, but the numbers
2111 of both had grown to the point where there were only 3 bits left.
2112 Fortunately, there was spare space in the data structure, and so I have
2113 moved the internal flags into a new 16-bit field to free up more option
2114 bits.
2115
2116 3. The appearance of (?J) at the start of a pattern set the DUPNAMES option,
2117 but did not set the internal JCHANGED flag - either of these is enough to
2118 control the way the "get" function works - but the PCRE_INFO_JCHANGED
2119 facility is supposed to tell if (?J) was ever used, so now (?J) at the
2120 start sets both bits.
2121
2122 4. Added options (at build time, compile time, exec time) to change \R from
2123 matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
2124
2125 5. doc/pcresyntax.html was missing from the distribution.
2126
2127 6. Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
2128 compatibility, even though it is no longer used.
2129
2130 7. Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
2131 strtoull to pcrecpp.cc to select the available functions in WIN32 when the
2132 windows.h file is present (where different names are used). [This was
2133 reversed later after testing - see 16 below.]
2134
2135 8. Changed all #include <config.h> to #include "config.h". There were also
2136 some further <pcre.h> cases that I changed to "pcre.h".
2137
2138 9. When pcregrep was used with the --colour option, it missed the line ending
2139 sequence off the lines that it output.
2140
2141 10. It was pointed out to me that arrays of string pointers cause lots of
2142 relocations when a shared library is dynamically loaded. A technique of
2143 using a single long string with a table of offsets can drastically reduce
2144 these. I have refactored PCRE in four places to do this. The result is
2145 dramatic:
2146
2147 Originally: 290
2148 After changing UCP table: 187
2149 After changing error message table: 43
2150 After changing table of "verbs" 36
2151 After changing table of Posix names 22
2152
2153 Thanks to the folks working on Gregex for glib for this insight.
2154
2155 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
2156 unicode-properties was also set.
2157
2158 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
2159
2160 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
2161 checked only for CRLF.
2162
2163 14. Added casts to pcretest.c to avoid compiler warnings.
2164
2165 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
2166
2167 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
2168 and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
2169 entirely. This removes changes made in 7 above.
2170
2171 17. The CMake files have been updated, and there is now more information about
2172 building with CMake in the NON-UNIX-USE document.
2173
2174
2175 Version 7.3 28-Aug-07
2176 ---------------------
2177
2178 1. In the rejigging of the build system that eventually resulted in 7.1, the
2179 line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
2180 brackets there is not right, since it causes compilers to look for an
2181 installed pcre.h, not the version that is in the source that is being
2182 compiled (which of course may be different). I have changed it back to:
2183
2184 #include "pcre.h"
2185
2186 I have a vague recollection that the change was concerned with compiling in
2187 different directories, but in the new build system, that is taken care of
2188 by the VPATH setting the Makefile.
2189
2190 2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
2191 when the subject happened to end in the byte 0x85 (e.g. if the last
2192 character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
2193 characters but of course it shouldn't be taken as a newline when it is part
2194 of another character. The bug was that, for an unlimited repeat of . in
2195 not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
2196 characters when looking for a newline.
2197
2198 3. A small performance improvement in the DOTALL UTF-8 mode .* case.
2199
2200 4. Debugging: adjusted the names of opcodes for different kinds of parentheses
2201 in debug output.
2202
2203 5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
2204 long printing in the pcrecpp unittest when running under MinGW.
2205
2206 6. ESC_K was left out of the EBCDIC table.
2207
2208 7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
2209 parentheses; I made it 1000, which seemed large enough. Unfortunately, the
2210 limit also applies to "virtual nesting" when a pattern is recursive, and in
2211 this case 1000 isn't so big. I have been able to remove this limit at the
2212 expense of backing off one optimization in certain circumstances. Normally,
2213 when pcre_exec() would call its internal match() function recursively and
2214 immediately return the result unconditionally, it uses a "tail recursion"
2215 feature to save stack. However, when a subpattern that can match an empty
2216 string has an unlimited repetition quantifier, it no longer makes this
2217 optimization. That gives it a stack frame in which to save the data for
2218 checking that an empty string has been matched. Previously this was taken
2219 from the 1000-entry workspace that had been reserved. So now there is no
2220 explicit limit, but more stack is used.
2221
2222 8. Applied Daniel's patches to solve problems with the import/export magic
2223 syntax that is required for Windows, and which was going wrong for the
2224 pcreposix and pcrecpp parts of the library. These were overlooked when this
2225 problem was solved for the main library.
2226
2227 9. There were some crude static tests to avoid integer overflow when computing
2228 the size of patterns that contain repeated groups with explicit upper
2229 limits. As the maximum quantifier is 65535, the maximum group length was
2230 set at 30,000 so that the product of these two numbers did not overflow a
2231 32-bit integer. However, it turns out that people want to use groups that
2232 are longer than 30,000 bytes (though not repeat them that many times).
2233 Change 7.0/17 (the refactoring of the way the pattern size is computed) has
2234 made it possible to implement the integer overflow checks in a much more
2235 dynamic way, which I have now done. The artificial limitation on group
2236 length has been removed - we now have only the limit on the total length of
2237 the compiled pattern, which depends on the LINK_SIZE setting.
2238
2239 10. Fixed a bug in the documentation for get/copy named substring when
2240 duplicate names are permitted. If none of the named substrings are set, the
2241 functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
2242 empty string.
2243
2244 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
2245 instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
2246 because the ] is interpreted as the first data character and the
2247 terminating ] is not found. PCRE has been made compatible with Perl in this
2248 regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
2249 cause memory overwriting.
2250
2251 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
2252 string has been matched (to stop an infinite loop). It was not recognizing
2253 a conditional subpattern that could match an empty string if that
2254 subpattern was within another subpattern. For example, it looped when
2255 trying to match (((?(1)X|))*) but it was OK with ((?(1)X|)*) where the
2256 condition was not nested. This bug has been fixed.
2257
2258 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
2259 past the start of the subject in the presence of bytes with the top bit
2260 set, for example "\x8aBCD".
2261
2262 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
2263 (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
2264
2265 14. Optimized (?!) to (*FAIL).
2266
2267 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
2268 This restricts code points to be within the range 0 to 0x10FFFF, excluding
2269 the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
2270 full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
2271 does: it's just the validity check that is more restrictive.
2272
2273 16. Inserted checks for integer overflows during escape sequence (backslash)
2274 processing, and also fixed erroneous offset values for syntax errors during
2275 backslash processing.
2276
2277 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
2278 for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
2279
2280 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
2281 caused an overrun.
2282
2283 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
2284 something other than just ASCII characters) inside a group that had an
2285 unlimited repeat caused a loop at compile time (while checking to see
2286 whether the group could match an empty string).
2287
2288 20. Debugging a pattern containing \p or \P could cause a crash. For example,
2289 [\P{Any}] did so. (Error in the code for printing property names.)
2290
2291 21. An orphan \E inside a character class could cause a crash.
2292
2293 22. A repeated capturing bracket such as (A)? could cause a wild memory
2294 reference during compilation.
2295
2296 23. There are several functions in pcre_compile() that scan along a compiled
2297 expression for various reasons (e.g. to see if it's fixed length for look
2298 behind). There were bugs in these functions when a repeated \p or \P was
2299 present in the pattern. These operators have additional parameters compared
2300 with \d, etc, and these were not being taken into account when moving along
2301 the compiled data. Specifically:
2302
2303 (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
2304 length.
2305
2306 (b) An item such as \pL+ within a repeated group could cause crashes or
2307 loops.
2308
2309 (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
2310 "reference to non-existent subpattern" error.
2311
2312 (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
2313
2314 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
2315 characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
2316
2317 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
2318
2319 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
2320 character were causing crashes (broken optimization).
2321
2322 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
2323 \p or \P) caused a compile-time loop.
2324
2325 28. More problems have arisen in unanchored patterns when CRLF is a valid line
2326 break. For example, the unstudied pattern [\r\n]A does not match the string
2327 "\r\nA" because change 7.0/46 below moves the current point on by two
2328 characters after failing to match at the start. However, the pattern \nA
2329 *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
2330 the same is true. There doesn't seem any very clean way out of this, but
2331 what I have chosen to do makes the common cases work: PCRE now takes note
2332 of whether there can be an explicit match for \r or \n anywhere in the
2333 pattern, and if so, 7.0/46 no longer applies. As part of this change,
2334 there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
2335 pattern has explicit CR or LF references.
2336
2337 29. Added (*CR) etc for changing newline setting at start of pattern.
2338
2339
2340 Version 7.2 19-Jun-07
2341 ---------------------
2342
2343 1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
2344 which is apparently normally available under Windows.
2345
2346 2. Re-jig the pcregrep tests with different newline settings in an attempt
2347 to make them independent of the local environment's newline setting.
2348
2349 3. Add code to configure.ac to remove -g from the CFLAGS default settings.
2350
2351 4. Some of the "internals" tests were previously cut out when the link size
2352 was not 2, because the output contained actual offsets. The recent new
2353 "Z" feature of pcretest means that these can be cut out, making the tests
2354 usable with all link sizes.
2355
2356 5. Implemented Stan Switzer's goto replacement for longjmp() when not using
2357 stack recursion. This gives a massive performance boost under BSD, but just
2358 a small improvement under Linux. However, it saves one field in the frame
2359 in all cases.
2360
2361 6. Added more features from the forthcoming Perl 5.10:
2362
2363 (a) (?-n) (where n is a string of digits) is a relative subroutine or
2364 recursion call. It refers to the nth most recently opened parentheses.
2365
2366 (b) (?+n) is also a relative subroutine call; it refers to the nth next
2367 to be opened parentheses.
2368
2369 (c) Conditions that refer to capturing parentheses can be specified
2370 relatively, for example, (?(-2)... or (?(+3)...
2371
2372 (d) \K resets the start of the current match so that everything before
2373 is not part of it.
2374
2375 (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
2376
2377 (f) \g{name} is another synonym - part of Perl 5.10's unification of
2378 reference syntax.
2379
2380 (g) (?| introduces a group in which the numbering of parentheses in each
2381 alternative starts with the same number.
2382
2383 (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
2384
2385 7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
2386 PCRE_INFO_JCHANGED.
2387
2388 8. A pattern such as (.*(.)?)* caused pcre_exec() to fail by either not
2389 terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
2390 for detecting groups that can match an empty string.
2391
2392 9. A pattern with a very large number of alternatives (more than several
2393 hundred) was running out of internal workspace during the pre-compile
2394 phase, where pcre_compile() figures out how much memory will be needed. A
2395 bit of new cunning has reduced the workspace needed for groups with
2396 alternatives. The 1000-alternative test pattern now uses 12 bytes of
2397 workspace instead of running out of the 4096 that are available.
2398
2399 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
2400
2401 11. Applied patch from Google to remove an optimization that didn't quite work.
2402 The report of the bug said:
2403
2404 pcrecpp::RE("a*").FullMatch("aaa") matches, while
2405 pcrecpp::RE("a*?").FullMatch("aaa") does not, and
2406 pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
2407
2408 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
2409 it matched the wrong number of bytes.
2410
2411
2412 Version 7.1 24-Apr-07
2413 ---------------------
2414
2415 1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
2416 that is more "standard", making use of automake and other Autotools. There
2417 is some re-arrangement of the files and adjustment of comments consequent
2418 on this.
2419
2420 2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
2421 for recursive directory scanning broke on some systems because the files
2422 are not scanned in any specific order and on different systems the order
2423 was different. A call to "sort" has been inserted into RunGrepTest for the
2424 approprate test as a short-term fix. In the longer term there may be an
2425 alternative.
2426
2427 3. I had an email from Eric Raymond about problems translating some of PCRE's
2428 man pages to HTML (despite the fact that I distribute HTML pages, some
2429 people do their own conversions for various reasons). The problems
2430 concerned the use of low-level troff macros .br and .in. I have therefore
2431 removed all such uses from the man pages (some were redundant, some could
2432 be replaced by .nf/.fi pairs). The 132html script that I use to generate
2433 HTML has been updated to handle .nf/.fi and to complain if it encounters
2434 .br or .in.
2435
2436 4. Updated comments in configure.ac that get placed in config.h.in and also
2437 arranged for config.h to be included in the distribution, with the name
2438 config.h.generic, for the benefit of those who have to compile without
2439 Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
2440
2441 5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
2442 Weber: (1) pcre_internal.h was missing some function renames; (2) updated
2443 makevp.bat for the current PCRE, using the additional files
2444 makevp_c.txt, makevp_l.txt, and pcregexp.pas.
2445
2446 6. A Windows user reported a minor discrepancy with test 2, which turned out
2447 to be caused by a trailing space on an input line that had got lost in his
2448 copy. The trailing space was an accident, so I've just removed it.
2449
2450 7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
2451 that is needed.
2452
2453 8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
2454 as "const" (a) because they are and (b) because it helps the PHP
2455 maintainers who have recently made a script to detect big data structures
2456 in the php code that should be moved to the .rodata section. I remembered
2457 to update Builducptable as well, so it won't revert if ucptable.h is ever
2458 re-created.
2459
2460 9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
2461 pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
2462 order to be able to cut out the UTF-8 tables in the latter when UTF-8
2463 support is not required. This saves 1.5-2K of code, which is important in
2464 some applications.
2465
2466 Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
2467 so as not to refer to the tables, even though these functions will never be
2468 called when UTF-8 support is disabled. Otherwise there are problems with a
2469 shared library.
2470
2471 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
2472
2473 (a) It was defining its arguments as char * instead of void *.
2474
2475 (b) It was assuming that all moves were upwards in memory; this was true
2476 a long time ago when I wrote it, but is no longer the case.
2477
2478 The emulated memove() is provided for those environments that have neither
2479 memmove() nor bcopy(). I didn't think anyone used it these days, but that
2480 is clearly not the case, as these two bugs were recently reported.
2481
2482 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
2483 and Detrail to create the HTML documentation, the .txt form of the man
2484 pages, and it removes trailing spaces from listed files. It also creates
2485 pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
2486 case, it wraps all the #defines with #ifndefs. This script should be run
2487 before "make dist".
2488
2489 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
2490 with Unicode property support.
2491
2492 (a) For a maximizing quantifier, if the two different cases of the
2493 character were of different lengths in their UTF-8 codings (there are
2494 some cases like this - I found 11), and the matching function had to
2495 back up over a mixture of the two cases, it incorrectly assumed they
2496 were both the same length.
2497
2498 (b) When PCRE was configured to use the heap rather than the stack for
2499 recursion during matching, it was not correctly preserving the data for
2500 the other case of a UTF-8 character when checking ahead for a match
2501 while processing a minimizing repeat. If the check also involved
2502 matching a wide character, but failed, corruption could cause an
2503 erroneous result when trying to check for a repeat of the original
2504 character.
2505
2506 13. Some tidying changes to the testing mechanism:
2507
2508 (a) The RunTest script now detects the internal link size and whether there
2509 is UTF-8 and UCP support by running ./pcretest -C instead of relying on
2510 values substituted by "configure". (The RunGrepTest script already did
2511 this for UTF-8.) The configure.ac script no longer substitutes the
2512 relevant variables.
2513
2514 (b) The debugging options /B and /D in pcretest show the compiled bytecode
2515 with length and offset values. This means that the output is different
2516 for different internal link sizes. Test 2 is skipped for link sizes
2517 other than 2 because of this, bypassing the problem. Unfortunately,
2518 there was also a test in test 3 (the locale tests) that used /B and
2519 failed for link sizes other than 2. Rather than cut the whole test out,
2520 I have added a new /Z option to pcretest that replaces the length and
2521 offset values with spaces. This is now used to make test 3 independent
2522 of link size. (Test 2 will be tidied up later.)
2523
2524 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
2525 segmentation fault instead of returning the appropriate error message.
2526
2527 15. In multiline mode when the newline sequence was set to "any", the pattern
2528 ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
2529 This doesn't seem right; it now treats the CRLF combination as the line
2530 ending, and so does not match in that case. It's only a pattern such as ^$
2531 that would hit this one: something like ^ABC$ would have failed after \r
2532 and then tried again after \r\n.
2533
2534 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
2535 in an attempt to make files that differ only in their line terminators
2536 compare equal. This works on Linux.
2537
2538 17. Under certain error circumstances pcregrep might try to free random memory
2539 as it exited. This is now fixed, thanks to valgrind.
2540
2541 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
2542 "abc\r\n\r\n", it found an unwanted second match after the second \r. This
2543 was because its rules for how to advance for /g after matching an empty
2544 string at the end of a line did not allow for this case. They now check for
2545 it specially.
2546
2547 20. pcretest is supposed to handle patterns and data of any length, by
2548 extending its buffers when necessary. It was getting this wrong when the
2549 buffer for a data line had to be extended.
2550
2551 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
2552 CRLF as a newline sequence.
2553
2554 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
2555 out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
2556 I have nevertheless tidied it up.
2557
2558 23. Added some casts to kill warnings from HP-UX ia64 compiler.
2559
2560 24. Added a man page for pcre-config.
2561
2562
2563 Version 7.0 19-Dec-06
2564 ---------------------
2565
2566 1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
2567 moving to gcc 4.1.1.
2568
2569 2. The -S option for pcretest uses setrlimit(); I had omitted to #include
2570 sys/time.h, which is documented as needed for this function. It doesn't
2571 seem to matter on Linux, but it showed up on some releases of OS X.
2572
2573 3. It seems that there are systems where bytes whose values are greater than
2574 127 match isprint() in the "C" locale. The "C" locale should be the
2575 default when a C program starts up. In most systems, only ASCII printing
2576 characters match isprint(). This difference caused the output from pcretest
2577 to vary, making some of the tests fail. I have changed pcretest so that:
2578
2579 (a) When it is outputting text in the compiled version of a pattern, bytes
2580 other than 32-126 are always shown as hex escapes.
2581
2582 (b) When it is outputting text that is a matched part of a subject string,
2583 it does the same, unless a different locale has been set for the match
2584 (using the /L modifier). In this case, it uses isprint() to decide.
2585
2586 4. Fixed a major bug that caused incorrect computation of the amount of memory
2587 required for a compiled pattern when options that changed within the
2588 pattern affected the logic of the preliminary scan that determines the
2589 length. The relevant options are -x, and -i in UTF-8 mode. The result was
2590 that the computed length was too small. The symptoms of this bug were
2591 either the PCRE error "internal error: code overflow" from pcre_compile(),
2592 or a glibc crash with a message such as "pcretest: free(): invalid next
2593 size (fast)". Examples of patterns that provoked this bug (shown in
2594 pcretest format) are:
2595
2596 /(?-x: )/x
2597 /(?x)(?-x: \s*#\s*)/
2598 /((?i)[\x{c0}])/8
2599 /(?i:[\x{c0}])/8
2600
2601 HOWEVER: Change 17 below makes this fix obsolete as the memory computation
2602 is now done differently.
2603
2604 5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
2605 wrapper classes; (b) implement a new function in the C++ scanner that is
2606 more efficient than the old way of doing things because it avoids levels of
2607 recursion in the regex matching; (c) add a paragraph to the documentation
2608 for the FullMatch() function.
2609
2610 6. The escape sequence \n was being treated as whatever was defined as
2611 "newline". Not only was this contrary to the documentation, which states
2612 that \n is character 10 (hex 0A), but it also went horribly wrong when
2613 "newline" was defined as CRLF. This has been fixed.
2614
2615 7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
2616 was being set to -1 for the "end of line" case (supposedly a value that no
2617 character can have). Though this value is never used (the check for end of
2618 line is "zero bytes in current character"), it caused compiler complaints.
2619 I've changed it to 0xffffffff.
2620
2621 8. In pcre_version.c, the version string was being built by a sequence of
2622 C macros that, in the event of PCRE_PRERELEASE being defined as an empty
2623 string (as it is for production releases) called a macro with an empty
2624 argument. The C standard says the result of this is undefined. The gcc
2625 compiler treats it as an empty string (which was what was wanted) but it is
2626 reported that Visual C gives an error. The source has been hacked around to
2627 avoid this problem.
2628
2629 9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
2630 builds of pcretest, and changed the call to _setmode() to use _O_BINARY
2631 instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
2632 of them did).
2633
2634 10. Originally, pcretest opened its input and output without "b"; then I was
2635 told that "b" was needed in some environments, so it was added for release
2636 5.0 to both the input and output. (It makes no difference on Unix-like
2637 systems.) Later I was told that it is wrong for the input on Windows. I've
2638 now abstracted the modes into two macros, to make it easier to fiddle with
2639 them, and removed "b" from the input mode under Windows.
2640
2641 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
2642
2643 12. Added -help and --help to pcretest as an official way of being reminded
2644 of the options.
2645
2646 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
2647 and pcrecpp.cc because they annoy compilers at high warning levels.
2648
2649 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
2650
2651 15. Fixed an occurrence of == in configure.ac that should have been = (shell
2652 scripts are not C programs :-) and which was not noticed because it works
2653 on Linux.
2654
2655 16. pcretest is supposed to handle any length of pattern and data line (as one
2656 line or as a continued sequence of lines) by extending its input buffer if
2657 necessary. This feature was broken for very long pattern lines, leading to
2658 a string of junk being passed to pcre_compile() if the pattern was longer
2659 than about 50K.
2660
2661 17. I have done a major re-factoring of the way pcre_compile() computes the
2662 amount of memory needed for a compiled pattern. Previously, there was code
2663 that made a preliminary scan of the pattern in order to do this. That was
2664 OK when PCRE was new, but as the facilities have expanded, it has become
2665 harder and harder to keep it in step with the real compile phase, and there
2666 have been a number of bugs (see for example, 4 above). I have now found a
2667 cunning way of running the real compile function in a "fake" mode that
2668 enables it to compute how much memory it would need, while actually only
2669 ever using a few hundred bytes of working memory and without too many
2670 tests of the mode. This should make future maintenance and development
2671 easier. A side effect of this work is that the limit of 200 on the nesting
2672 depth of parentheses has been removed (though this was never a serious
2673 limitation, I suspect). However, there is a downside: pcre_compile() now
2674 runs more slowly than before (30% or more, depending on the pattern). I
2675 hope this isn't a big issue. There is no effect on runtime performance.
2676
2677 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
2678 newline (only possible for the last line of a file) and it was a
2679 pattern that set a locale (followed by /Lsomething), pcretest crashed.
2680
2681 19. Added additional timing features to pcretest. (1) The -tm option now times
2682 matching only, not compiling. (2) Both -t and -tm can be followed, as a
2683 separate command line item, by a number that specifies the number of
2684 repeats to use when timing. The default is 50000; this gives better
2685 precision, but takes uncomfortably long for very large patterns.
2686
2687 20. Extended pcre_study() to be more clever in cases where a branch of a
2688 subpattern has no definite first character. For example, (a*|b*)[cd] would
2689 previously give no result from pcre_study(). Now it recognizes that the
2690 first character must be a, b, c, or d.
2691
2692 21. There was an incorrect error "recursive call could loop indefinitely" if
2693 a subpattern (or the entire pattern) that was being tested for matching an
2694 empty string contained only one non-empty item after a nested subpattern.
2695 For example, the pattern (?>\x{100}*)\d(?R) provoked this error
2696 incorrectly, because the \d was being skipped in the check.
2697
2698 22. The pcretest program now has a new pattern option /B and a command line
2699 option -b, which is equivalent to adding /B to every pattern. This causes
2700 it to show the compiled bytecode, without the additional information that
2701 -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
2702 is the same as /B/I).
2703
2704 23. A new optimization is now able automatically to treat some sequences such
2705 as a*b as a*+b. More specifically, if something simple (such as a character
2706 or a simple class like \d) has an unlimited quantifier, and is followed by
2707 something that cannot possibly match the quantified thing, the quantifier
2708 is automatically "possessified".
2709
2710 24. A recursive reference to a subpattern whose number was greater than 39
2711 went wrong under certain circumstances in UTF-8 mode. This bug could also
2712 have affected the operation of pcre_study().
2713
2714 25. Realized that a little bit of performance could be had by replacing
2715 (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
2716
2717 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
2718
2719 27. Possessive quantifiers such as a++ were previously implemented by turning
2720 them into atomic groups such as ($>a+). Now they have their own opcodes,
2721 which improves performance. This includes the automatically created ones
2722 from 23 above.
2723
2724 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
2725 lookahead was broken if it was not anchored. PCRE was mistakenly expecting
2726 the first matched character to be a colon. This applied both to named and
2727 numbered groups.
2728
2729 29. The ucpinternal.h header file was missing its idempotency #ifdef.
2730
2731 30. I was sent a "project" file called libpcre.a.dev which I understand makes
2732 building PCRE on Windows easier, so I have included it in the distribution.
2733
2734 31. There is now a check in pcretest against a ridiculously large number being
2735 returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
2736 loop, the loop is abandoned.
2737
2738 32. Forward references to subpatterns in conditions such as (?(2)...) where
2739 subpattern 2 is defined later cause pcre_compile() to search forwards in
2740 the pattern for the relevant set of parentheses. This search went wrong
2741 when there were unescaped parentheses in a character class, parentheses
2742 escaped with \Q...\E, or parentheses in a #-comment in /x mode.
2743
2744 33. "Subroutine" calls and backreferences were previously restricted to
2745 referencing subpatterns earlier in the regex. This restriction has now
2746 been removed.
2747
2748 34. Added a number of extra features that are going to be in Perl 5.10. On the
2749 whole, these are just syntactic alternatives for features that PCRE had
2750 previously implemented using the Python syntax or my own invention. The
2751 other formats are all retained for compatibility.
2752
2753 (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
2754 as (?P<name>...). The new forms, as well as being in Perl 5.10, are
2755 also .NET compatible.
2756
2757 (b) A recursion or subroutine call to a named group can now be defined as
2758 (?&name) as well as (?P>name).
2759
2760 (c) A backreference to a named group can now be defined as \k<name> or
2761 \k'name' as well as (?P=name). The new forms, as well as being in Perl
2762 5.10, are also .NET compatible.
2763
2764 (d) A conditional reference to a named group can now use the syntax
2765 (?(<name>) or (?('name') as well as (?(name).
2766
2767 (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
2768 groups (named and numbered) that are never evaluated inline, but can be
2769 called as "subroutines" from elsewhere. In effect, the DEFINE condition
2770 is always false. There may be only one alternative in such a group.
2771
2772 (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
2773 as the simple (?(R). The condition is true only if the most recent
2774 recursion is that of the given number or name. It does not search out
2775 through the entire recursion stack.
2776
2777 (g) The escape \gN or \g{N} has been added, where N is a positive or
2778 negative number, specifying an absolute or relative reference.
2779
2780 35. Tidied to get rid of some further signed/unsigned compiler warnings and
2781 some "unreachable code" warnings.
2782
2783 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
2784 things, this adds five new scripts.
2785
2786 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
2787 There were also incompatibilities regarding the handling of \Q..\E inside
2788 character classes, for example with patterns like [\Qa\E-\Qz\E] where the
2789 hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
2790
2791 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
2792 matches an empty string, and forcibly breaks the loop. There were bugs in
2793 this code in non-simple cases. For a pattern such as ^(a()*)* matched
2794 against aaaa the result was just "a" rather than "aaaa", for example. Two
2795 separate and independent bugs (that affected different cases) have been
2796 fixed.
2797
2798 39. Refactored the code to abolish the use of different opcodes for small
2799 capturing bracket numbers. This is a tidy that I avoided doing when I
2800 removed the limit on the number of capturing brackets for 3.5 back in 2001.
2801 The new approach is not only tidier, it makes it possible to reduce the
2802 memory needed to fix the previous bug (38).
2803
2804 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
2805 sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
2806 processing dot, circumflex, or dollar metacharacters, or #-comments in /x
2807 mode.
2808
2809 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
2810 report.
2811
2812 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
2813 copy construction and assignment in the C++ wrapper.
2814
2815 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
2816 couple of bugs that could have given wrong results in the "--newline=crlf"
2817 case.
2818
2819 44. Added a number of casts and did some reorganization of signed/unsigned int
2820 variables following suggestions from Dair Grant. Also renamed the variable
2821 "this" as "item" because it is a C++ keyword.
2822
2823 45. Arranged for dftables to add
2824
2825 #include "pcre_internal.h"
2826
2827 to pcre_chartables.c because without it, gcc 4.x may remove the array
2828 definition from the final binary if PCRE is built into a static library and
2829 dead code stripping is activated.
2830
2831 46. For an unanchored pattern, if a match attempt fails at the start of a
2832 newline sequence, and the newline setting is CRLF or ANY, and the next two
2833 characters are CRLF, advance by two characters instead of one.
2834
2835
2836 Version 6.7 04-Jul-06
2837 ---------------------
2838
2839 1. In order to handle tests when input lines are enormously long, pcretest has
2840 been re-factored so that it automatically extends its buffers when
2841 necessary. The code is crude, but this _is_ just a test program. The
2842 default size has been increased from 32K to 50K.
2843
2844 2. The code in pcre_study() was using the value of the re argument before
2845 testing it for NULL. (Of course, in any sensible call of the function, it
2846 won't be NULL.)
2847
2848 3. The memmove() emulation function in pcre_internal.h, which is used on
2849 systems that lack both memmove() and bcopy() - that is, hardly ever -
2850 was missing a "static" storage class specifier.
2851
2852 4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
2853 containing an extended class (one that cannot be represented by a bitmap
2854 because it contains high-valued characters or Unicode property items, e.g.
2855 [\pZ]). Almost always one would set UTF-8 mode when processing such a
2856 pattern, but PCRE should not loop if you do not (it no longer does).
2857 [Detail: two cases were found: (a) a repeated subpattern containing an
2858 extended class; (b) a recursive reference to a subpattern that followed a
2859 previous extended class. It wasn't skipping over the extended class
2860 correctly when UTF-8 mode was not set.]
2861
2862 5. A negated single-character class was not being recognized as fixed-length
2863 in lookbehind assertions such as (?<=[^f]), leading to an incorrect
2864 compile error "lookbehind assertion is not fixed length".
2865
2866 6. The RunPerlTest auxiliary script was showing an unexpected difference
2867 between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
2868 write a Perl script that can interpret lines of an input file either as
2869 byte characters or as UTF-8, which is what "perltest" was being required to
2870 do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
2871 can't do is switch easily at run time between having the "use utf8;" pragma
2872 or not. In the end, I fudged it by using the RunPerlTest script to insert
2873 "use utf8;" explicitly for the UTF-8 tests.
2874
2875 7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
2876 the end of the subject string, contrary to the documentation and to what
2877 Perl does. This was true of both matching functions. Now it matches only at
2878 the start of the subject and immediately after *internal* newlines.
2879
2880 8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
2881 a pointer to an int instead of a pointer to an unsigned long int. This
2882 caused problems on 64-bit systems.
2883
2884 9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
2885 instance of the 'standard' template library not being so standard".
2886
2887 10. There was no check on the number of named subpatterns nor the maximum
2888 length of a subpattern name. The product of these values is used to compute
2889 the size of the memory block for a compiled pattern. By supplying a very
2890 long subpattern name and a large number of named subpatterns, the size
2891 computation could be caused to overflow. This is now prevented by limiting
2892 the length of names to 32 characters, and the number of named subpatterns
2893 to 10,000.
2894
2895 11. Subpatterns that are repeated with specific counts have to be replicated in
2896 the compiled pattern. The size of memory for this was computed from the
2897 length of the subpattern and the repeat count. The latter is limited to
2898 65535, but there was no limit on the former, meaning that integer overflow
2899 could in principle occur. The compiled length of a repeated subpattern is
2900 now limited to 30,000 bytes in order to prevent this.
2901
2902 12. Added the optional facility to have named substrings with the same name.
2903
2904 13. Added the ability to use a named substring as a condition, using the
2905 Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
2906 are numbers (not recommended). Forward references are permitted.
2907
2908 14. Added forward references in named backreferences (if you see what I mean).
2909
2910 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
2911 pattern could run off the end of the subject. For example, the pattern
2912 "(?s)(.{1,5})"8 did this with the subject "ab".
2913
2914 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
2915 PCRE_CASELESS was set when matching characters that were quantified with ?
2916 or *.
2917
2918 17. A character class other than a single negated character that had a minimum
2919 but no maximum quantifier - for example [ab]{6,} - was not handled
2920 correctly by pce_dfa_exec(). It would match only one character.
2921
2922 18. A valid (though odd) pattern that looked like a POSIX character
2923 class but used an invalid character after [ (for example [[,abc,]]) caused
2924 pcre_compile() to give the error "Failed: internal error: code overflow" or
2925 in some cases to crash with a glibc free() error. This could even happen if
2926 the pattern terminated after [[ but there just happened to be a sequence of
2927 letters, a binary zero, and a closing ] in the memory that followed.
2928
2929 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
2930 over the years. Originally (before any Unicode support), just the bottom 8
2931 bits were taken. Thus, for example, \500 really meant \100. Nowadays the
2932 output from "man perlunicode" includes this:
2933
2934 The regular expression compiler produces polymorphic opcodes. That
2935 is, the pattern adapts to the data and automatically switches to
2936 the Unicode character scheme when presented with Unicode data--or
2937 instead uses a traditional byte scheme when presented with byte
2938 data.
2939
2940 Sadly, a wide octal escape does not cause a switch, and in a string with
2941 no other multibyte characters, these octal escapes are treated as before.
2942 Thus, in Perl, the pattern /\500/ actually matches \100 but the pattern
2943 /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
2944 Unicode string.
2945
2946 I have not perpetrated such confusion in PCRE. Up till now, it took just
2947 the bottom 8 bits, as in old Perl. I have now made octal escapes with
2948 values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
2949 translate to the appropriate multibyte character.
2950
2951 29. Applied some refactoring to reduce the number of warnings from Microsoft
2952 and Borland compilers. This has included removing the fudge introduced
2953 seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
2954 a warning about an unused variable.
2955
2956 21. PCRE has not included VT (character 0x0b) in the set of whitespace
2957 characters since release 4.0, because Perl (from release 5.004) does not.
2958 [Or at least, is documented not to: some releases seem to be in conflict
2959 with the documentation.] However, when a pattern was studied with
2960 pcre_study() and all its branches started with \s, PCRE still included VT
2961 as a possible starting character. Of course, this did no harm; it just
2962 caused an unnecessary match attempt.
2963
2964 22. Removed a now-redundant internal flag bit that recorded the fact that case
2965 dependency changed within the pattern. This was once needed for "required
2966 byte" processing, but is no longer used. This recovers a now-scarce options
2967 bit. Also moved the least significant internal flag bit to the most-
2968 significant bit of the word, which was not previously used (hangover from
2969 the days when it was an int rather than a uint) to free up another bit for
2970 the future.
2971
2972 23. Added support for CRLF line endings as well as CR and LF. As well as the
2973 default being selectable at build time, it can now be changed at runtime
2974 via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
2975 specify that it is scanning data with non-default line endings.
2976
2977 24. Changed the definition of CXXLINK to make it agree with the definition of
2978 LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
2979
2980 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
2981 recursions. This makes a big different to stack usage for some patterns.
2982
2983 26. If a subpattern containing a named recursion or subroutine reference such
2984 as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
2985 the space required for the compiled pattern went wrong and gave too small a
2986 value. Depending on the environment, this could lead to "Failed: internal
2987 error: code overflow at offset 49" or "glibc detected double free or
2988 corruption" errors.
2989
2990 27. Applied patches from Google (a) to support the new newline modes and (b) to
2991 advance over multibyte UTF-8 characters in GlobalReplace.
2992
2993 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
2994 difference for some implementation of PCRE in some Windows version.
2995
2996 29. Added some extra testing facilities to pcretest:
2997
2998 \q<number> in a data line sets the "match limit" value
2999 \Q<number> in a data line sets the "match recursion limt" value
3000 -S <number> sets the stack size, where <number> is in megabytes
3001
3002 The -S option isn't available for Windows.
3003
3004
3005 Version 6.6 06-Feb-06
3006 ---------------------
3007
3008 1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
3009 in pcreposix.h. I have copied the definition from pcre.h.
3010
3011 2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
3012 because pcre.h is no longer a built file.
3013
3014 3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
3015 not normally included in the compiled code.
3016
3017
3018 Version 6.5 01-Feb-06
3019 ---------------------
3020
3021 1. When using the partial match feature with pcre_dfa_exec(), it was not
3022 anchoring the second and subsequent partial matches at the new starting
3023 point. This could lead to incorrect results. For example, with the pattern
3024 /1234/, partially matching against "123" and then "a4" gave a match.
3025
3026 2. Changes to pcregrep:
3027
3028 (a) All non-match returns from pcre_exec() were being treated as failures
3029 to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
3030 error message is output. Some extra information is given for the
3031 PCRE_ERROR_MATCHLIMIT and PCRE_ERROR_RECURSIONLIMIT errors, which are
3032 probably the only errors that are likely to be caused by users (by
3033 specifying a regex that has nested indefinite repeats, for instance).
3034 If there are more than 20 of these errors, pcregrep is abandoned.
3035
3036 (b) A binary zero was treated as data while matching, but terminated the
3037 output line if it was written out. This has been fixed: binary zeroes
3038 are now no different to any other data bytes.
3039
3040 (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
3041 used to set a locale for matching. The --locale=xxxx long option has
3042 been added (no short equivalent) to specify a locale explicitly on the
3043 pcregrep command, overriding the environment variables.
3044
3045 (d) When -B was used with -n, some line numbers in the output were one less
3046 than they should have been.
3047
3048 (e) Added the -o (--only-matching) option.
3049
3050 (f) If -A or -C was used with -c (count only), some lines of context were
3051 accidentally printed for the final match.
3052
3053 (g) Added the -H (--with-filename) option.
3054
3055 (h) The combination of options -rh failed to suppress file names for files
3056 that were found from directory arguments.
3057
3058 (i) Added the -D (--devices) and -d (--directories) options.
3059
3060 (j) Added the -F (--fixed-strings) option.
3061
3062 (k) Allow "-" to be used as a file name for -f as well as for a data file.
3063
3064 (l) Added the --colo(u)r option.
3065
3066 (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
3067 is not present by default.
3068
3069 3. A nasty bug was discovered in the handling of recursive patterns, that is,
3070 items such as (?R) or (?1), when the recursion could match a number of
3071 alternatives. If it matched one of the alternatives, but subsequently,
3072 outside the recursion, there was a failure, the code tried to back up into
3073 the recursion. However, because of the way PCRE is implemented, this is not
3074 possible, and the result was an incorrect result from the match.
3075
3076 In order to prevent this happening, the specification of recursion has
3077 been changed so that all such subpatterns are automatically treated as
3078 atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
3079
3080 4. I had overlooked the fact that, in some locales, there are characters for
3081 which isalpha() is true but neither isupper() nor islower() are true. In
3082 the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
3083 and ordfeminine) are like this. This affected the treatment of \w and \W
3084 when they appeared in character classes, but not when they appeared outside
3085 a character class. The bit map for "word" characters is now created
3086 separately from the results of isalnum() instead of just taking it from the
3087 upper, lower, and digit maps. (Plus the underscore character, of course.)
3088
3089 5. The above bug also affected the handling of POSIX character classes such as
3090 [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
3091 permanent tables. Instead, the bit maps for such a class were previously
3092 created as the appropriate unions of the upper, lower, and digit bitmaps.
3093 Now they are created by subtraction from the [[:word:]] class, which has
3094 its own bitmap.
3095
3096 6. The [[:blank:]] character class matches horizontal, but not vertical space.
3097 It is created by subtracting the vertical space characters (\x09, \x0a,
3098 \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
3099 subtraction was done in the overall bitmap for a character class, meaning
3100 that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
3101 be recognized. This bug has been fixed.
3102
3103 7. Patches from the folks at Google:
3104
3105 (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
3106 real life, but is still worth protecting against".
3107
3108 (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
3109 regular expressions".
3110
3111 (c) pcre_scanner.cc: avoid use of std::count() because not all systems
3112 have it.
3113
3114 (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
3115 "configure" and the latter not, in order to fix a problem somebody had
3116 with compiling the Arg class on HP-UX.
3117
3118 (e) Improve the error-handling of the C++ wrapper a little bit.
3119
3120 (f) New tests for checking recursion limiting.
3121
3122 8. The pcre_memmove() function, which is used only if the environment does not
3123 have a standard memmove() function (and is therefore rarely compiled),
3124 contained two bugs: (a) use of int instead of size_t, and (b) it was not
3125 returning a result (though PCRE never actually uses the result).
3126
3127 9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
3128 large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
3129 returned instead of calling malloc() with an overflowing number that would
3130 most likely cause subsequent chaos.
3131
3132 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
3133
3134 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
3135 with this option is matched, the nmatch and pmatch options of regexec() are
3136 ignored.
3137
3138 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
3139 provided in case anyone wants to the the POSIX interface with UTF-8
3140 strings.
3141
3142 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
3143 C++ linking (needed for some HP-UX environments).
3144
3145 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
3146 (unused parameter) and in the pcre_printint() function (omitted "default"
3147 switch label when the default is to do nothing).
3148
3149 15. Added some code to make it possible, when PCRE is compiled as a C++
3150 library, to replace subject pointers for pcre_exec() with a smart pointer
3151 class, thus making it possible to process discontinuous strings.
3152
3153 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
3154 much the same function. They were added by different people who were trying
3155 to make PCRE easy to compile on non-Unix systems. It has been suggested
3156 that PCRE_EXPORT be abolished now that there is more automatic apparatus
3157 for compiling on Windows systems. I have therefore replaced it with
3158 PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
3159 defaults to "extern" for C or "extern C" for C++, which works fine on
3160 Unix-like systems. It is now possible to override the value of PCRE_DATA_
3161 SCOPE with something explicit in config.h. In addition:
3162
3163 (a) pcreposix.h still had just "extern" instead of either of these macros;
3164 I have replaced it with PCRE_DATA_SCOPE.
3165
3166 (b) Functions such as _pcre_xclass(), which are internal to the library,
3167 but external in the C sense, all had PCRE_EXPORT in their definitions.
3168 This is apparently wrong for the Windows case, so I have removed it.
3169 (It makes no difference on Unix-like systems.)
3170
3171 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
3172 of recursive calls to match(). This is different to MATCH_LIMIT because
3173 that limits the total number of calls to match(), not all of which increase
3174 the depth of recursion. Limiting the recursion depth limits the amount of
3175 stack (or heap if NO_RECURSE is set) that is used. The default can be set
3176 when PCRE is compiled, and changed at run time. A patch from Google adds
3177 this functionality to the C++ interface.
3178
3179 18. Changes to the handling of Unicode character properties:
3180
3181 (a) Updated the table to Unicode 4.1.0.
3182
3183 (b) Recognize characters that are not in the table as "Cn" (undefined).
3184
3185 (c) I revised the way the table is implemented to a much improved format
3186 which includes recognition of ranges. It now supports the ranges that
3187 are defined in UnicodeData.txt, and it also amalgamates other
3188 characters into ranges. This has reduced the number of entries in the
3189 table from around 16,000 to around 3,000, thus reducing its size
3190 considerably. I realized I did not need to use a tree structure after
3191 all - a binary chop search is just as efficient. Having reduced the
3192 number of entries, I extended their size from 6 bytes to 8 bytes to
3193 allow for more data.
3194
3195 (d) Added support for Unicode script names via properties such as \p{Han}.
3196
3197 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
3198 matching that character.
3199
3200 20. When matching a repeated Unicode property with a minimum greater than zero,
3201 (for example \pL{2,}), PCRE could look past the end of the subject if it
3202 reached it while seeking the minimum number of characters. This could
3203 happen only if some of the characters were more than one byte long, because
3204 there is a check for at least the minimum number of bytes.
3205
3206 21. Refactored the implementation of \p and \P so as to be more general, to
3207 allow for more different types of property in future. This has changed the
3208 compiled form incompatibly. Anybody with saved compiled patterns that use
3209 \p or \P will have to recompile them.
3210
3211 22. Added "Any" and "L&" to the supported property types.
3212
3213 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
3214 but give a compile time error if the value is greater than 0xff.
3215
3216 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
3217 accidentally not being installed or uninstalled.
3218
3219 25. The pcre.h file was built from pcre.h.in, but the only changes that were
3220 made were to insert the current release number. This seemed silly, because
3221 it made things harder for people building PCRE on systems that don't run
3222 "configure". I have turned pcre.h into a distributed file, no longer built
3223 by "configure", with the version identification directly included. There is
3224 no longer a pcre.h.in file.
3225
3226 However, this change necessitated a change to the pcre-config script as
3227 well. It is built from pcre-config.in, and one of the substitutions was the
3228 release number. I have updated configure.ac so that ./configure now finds
3229 the release number by grepping pcre.h.
3230
3231 26. Added the ability to run the tests under valgrind.
3232
3233
3234 Version 6.4 05-Sep-05
3235 ---------------------
3236
3237 1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
3238 "--" to be printed when multiple files were scanned, even when none of the
3239 -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
3240 consider it to be a bug, and have restored the previous behaviour.
3241
3242 2. A couple of code tidies to get rid of compiler warnings.
3243
3244 3. The pcretest program used to cheat by referring to symbols in the library
3245 whose names begin with _pcre_. These are internal symbols that are not
3246 really supposed to be visible externally, and in some environments it is
3247 possible to suppress them. The cheating is now confined to including
3248 certain files from the library's source, which is a bit cleaner.
3249
3250 4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
3251 file's purpose clearer.
3252
3253 5. Reorganized pcre_ucp_findchar().
3254
3255
3256 Version 6.3 15-Aug-05
3257 ---------------------
3258
3259 1. The file libpcre.pc.in did not have general read permission in the tarball.
3260
3261 2. There were some problems when building without C++ support:
3262
3263 (a) If C++ support was not built, "make install" and "make test" still
3264 tried to test it.
3265
3266 (b) There were problems when the value of CXX was explicitly set. Some
3267 changes have been made to try to fix these, and ...
3268
3269 (c) --disable-cpp can now be used to explicitly disable C++ support.
3270
3271 (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
3272 backslash in a target when C++ was disabled. This confuses some
3273 versions of "make", apparently. Using an intermediate variable solves
3274 this. (Same for CPP_LOBJ.)
3275
3276 3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
3277 (non-Windows) now includes $(CFLAGS) because these flags are sometimes
3278 necessary on certain architectures.
3279
3280 4. Added a setting of -export-symbols-regex to the link command to remove
3281 those symbols that are exported in the C sense, but actually are local
3282 within the library, and not documented. Their names all begin with
3283 "_pcre_". This is not a perfect job, because (a) we have to except some
3284 symbols that pcretest ("illegally") uses, and (b) the facility isn't always
3285 available (and never for static libraries). I have made a note to try to
3286 find a way round (a) in the future.
3287
3288
3289 Version 6.2 01-Aug-05
3290 ---------------------
3291
3292 1. There was no test for integer overflow of quantifier values. A construction
3293 such as {1111111111111111} would give undefined results. What is worse, if
3294 a minimum quantifier for a parenthesized subpattern overflowed and became
3295 negative, the calculation of the memory size went wrong. This could have
3296 led to memory overwriting.
3297
3298 2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
3299
3300 3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
3301 operating environments where this matters.
3302
3303 4. Applied Giuseppe Maxia's patch to add additional features for controlling
3304 PCRE options from within the C++ wrapper.
3305
3306 5. Named capturing subpatterns were not being correctly counted when a pattern
3307 was compiled. This caused two problems: (a) If there were more than 100
3308 such subpatterns, the calculation of the memory needed for the whole
3309 compiled pattern went wrong, leading to an overflow error. (b) Numerical
3310 back references of the form \12, where the number was greater than 9, were
3311 not recognized as back references, even though there were sufficient
3312 previous subpatterns.
3313
3314 6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
3315 versions of gcc, e.g. 2.95.4.
3316
3317
3318 Version 6.1 21-Jun-05
3319 ---------------------
3320
3321 1. There was one reference to the variable "posix" in pcretest.c that was not
3322 surrounded by "#if !defined NOPOSIX".
3323
3324 2. Make it possible to compile pcretest without DFA support, UTF8 support, or
3325 the cross-check on the old pcre_info() function, for the benefit of the
3326 cut-down version of PCRE that is currently imported into Exim.
3327
3328 3. A (silly) pattern starting with (?i)(?-i) caused an internal space
3329 allocation error. I've done the easy fix, which wastes 2 bytes for sensible
3330 patterns that start (?i) but I don't think that matters. The use of (?i) is
3331 just an example; this all applies to the other options as well.
3332
3333 4. Since libtool seems to echo the compile commands it is issuing, the output
3334 from "make" can be reduced a bit by putting "@" in front of each libtool
3335 compile command.
3336
3337 5. Patch from the folks at Google for configure.in to be a bit more thorough
3338 in checking for a suitable C++ installation before trying to compile the
3339 C++ stuff. This should fix a reported problem when a compiler was present,
3340 but no suitable headers.
3341
3342 6. The man pages all had just "PCRE" as their title. I have changed them to
3343 be the relevant file name. I have also arranged that these names are
3344 retained in the file doc/pcre.txt, which is a concatenation in text format
3345 of all the man pages except the little individual ones for each function.
3346
3347 7. The NON-UNIX-USE file had not been updated for the different set of source
3348 files that come with release 6. I also added a few comments about the C++
3349 wrapper.
3350
3351
3352 Version 6.0 07-Jun-05
3353 ---------------------
3354
3355 1. Some minor internal re-organization to help with my DFA experiments.
3356
3357 2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
3358 didn't matter for the library itself when fully configured, but did matter
3359 when compiling without UCP support, or within Exim, where the ucp files are
3360 not imported.
3361
3362 3. Refactoring of the library code to split up the various functions into
3363 different source modules. The addition of the new DFA matching code (see
3364 below) to a single monolithic source would have made it really too
3365 unwieldy, quite apart from causing all the code to be include in a
3366 statically linked application, when only some functions are used. This is
3367 relevant even without the DFA addition now that patterns can be compiled in
3368 one application and matched in another.
3369
3370 The downside of splitting up is that there have to be some external
3371 functions and data tables that are used internally in different modules of
3372 the library but which are not part of the API. These have all had their
3373 names changed to start with "_pcre_" so that they are unlikely to clash
3374 with other external names.
3375
3376 4. Added an alternate matching function, pcre_dfa_exec(), which matches using
3377 a different (DFA) algorithm. Although it is slower than the original
3378 function, it does have some advantages for certain types of matching
3379 problem.
3380
3381 5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
3382 including restarting after a partial match.
3383
3384 6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
3385 defined when compiling for Windows was sent to me. I have put it into the
3386 code, though I have no means of testing or verifying it.
3387
3388 7. Added the pcre_refcount() auxiliary function.
3389
3390 8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
3391 match before or at the first newline in the subject string. In pcretest,
3392 the /f option on a pattern can be used to set this.
3393
3394 9. A repeated \w when used in UTF-8 mode with characters greater than 256
3395 would behave wrongly. This has been present in PCRE since release 4.0.
3396
3397 10. A number of changes to the pcregrep command:
3398
3399 (a) Refactored how -x works; insert ^(...)$ instead of setting
3400 PCRE_ANCHORED and checking the length, in preparation for adding
3401 something similar for -w.
3402
3403 (b) Added the -w (match as a word) option.
3404
3405 (c) Refactored the way lines are read and buffered so as to have more
3406 than one at a time available.
3407
3408 (d) Implemented a pcregrep test script.
3409
3410 (e) Added the -M (multiline match) option. This allows patterns to match
3411 over several lines of the subject. The buffering ensures that at least
3412 8K, or the rest of the document (whichever is the shorter) is available
3413 for matching (and similarly the previous 8K for lookbehind assertions).
3414
3415 (f) Changed the --help output so that it now says
3416
3417 -w, --word-regex(p)
3418
3419 instead of two lines, one with "regex" and the other with "regexp"
3420 because that confused at least one person since the short forms are the
3421 same. (This required a bit of code, as the output is generated
3422 automatically from a table. It wasn't just a text change.)
3423
3424 (g) -- can be used to terminate pcregrep options if the next thing isn't an
3425 option but starts with a hyphen. Could be a pattern or a path name
3426 starting with a hyphen, for instance.
3427
3428 (h) "-" can be given as a file name to represent stdin.
3429
3430 (i) When file names are being printed, "(standard input)" is used for
3431 the standard input, for compatibility with GNU grep. Previously
3432 "<stdin>" was used.
3433
3434 (j) The option --label=xxx can be used to supply a name to be used for
3435 stdin when file names are being printed. There is no short form.
3436
3437 (k) Re-factored the options decoding logic because we are going to add
3438 two more options that take data. Such options can now be given in four
3439 different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
3440
3441 (l) Added the -A, -B, and -C options for requesting that lines of context
3442 around matches be printed.
3443
3444 (m) Added the -L option to print the names of files that do not contain
3445 any matching lines, that is, the complement of -l.
3446
3447 (n) The return code is 2 if any file cannot be opened, but pcregrep does
3448 continue to scan other files.
3449
3450 (o) The -s option was incorrectly implemented. For compatibility with other
3451 greps, it now suppresses the error message for a non-existent or non-
3452 accessible file (but not the return code). There is a new option called
3453 -q that suppresses the output of matching lines, which was what -s was
3454 previously doing.
3455
3456 (p) Added --include and --exclude options to specify files for inclusion
3457 and exclusion when recursing.
3458
3459 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
3460 Hopefully, it now does.
3461
3462 12. Missing cast in pcre_study().
3463
3464 13. Added an "uninstall" target to the makefile.
3465
3466 14. Replaced "extern" in the function prototypes in Makefile.in with
3467 "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
3468 world, but is set differently for Windows.
3469
3470 15. Added a second compiling function called pcre_compile2(). The only
3471 difference is that it has an extra argument, which is a pointer to an
3472 integer error code. When there is a compile-time failure, this is set
3473 non-zero, in addition to the error test pointer being set to point to an
3474 error message. The new argument may be NULL if no error number is required
3475 (but then you may as well call pcre_compile(), which is now just a
3476 wrapper). This facility is provided because some applications need a
3477 numeric error indication, but it has also enabled me to tidy up the way
3478 compile-time errors are handled in the POSIX wrapper.
3479
3480 16. Added VPATH=.libs to the makefile; this should help when building with one
3481 prefix path and installing with another. (Or so I'm told by someone who
3482 knows more about this stuff than I do.)
3483
3484 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
3485 passes PCRE_DOTALL to the pcre_compile() function, making the "." character
3486 match everything, including newlines. This is not POSIX-compatible, but
3487 somebody wanted the feature. From pcretest it can be activated by using
3488 both the P and the s flags.
3489
3490 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
3491
3492 19. libpcre.pc was being incorrectly installed as executable.
3493
3494 20. A couple of places in pcretest check for end-of-line by looking for '\n';
3495 it now also looks for '\r' so that it will work unmodified on Windows.
3496
3497 21. Added Google's contributed C++ wrapper to the distribution.
3498
3499 22. Added some untidy missing memory free() calls in pcretest, to keep
3500 Electric Fence happy when testing.
3501
3502
3503
3504 Version 5.0 13-Sep-04
3505 ---------------------
3506
3507 1. Internal change: literal characters are no longer packed up into items
3508 containing multiple characters in a single byte-string. Each character
3509 is now matched using a separate opcode. However, there may be more than one
3510 byte in the character in UTF-8 mode.
3511
3512 2. The pcre_callout_block structure has two new fields: pattern_position and
3513 next_item_length. These contain the offset in the pattern to the next match
3514 item, and its length, respectively.
3515
3516 3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
3517 insertion of callouts before each pattern item. Added the /C option to
3518 pcretest to make use of this.
3519
3520 4. On the advice of a Windows user, the lines
3521
3522 #if defined(_WIN32) || defined(WIN32)
3523 _setmode( _fileno( stdout ), 0x8000 );
3524 #endif /* defined(_WIN32) || defined(WIN32) */
3525
3526 have been added to the source of pcretest. This apparently does useful
3527 magic in relation to line terminators.
3528
3529 5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
3530 for the benefit of those environments where the "b" makes a difference.
3531
3532 6. The icc compiler has the same options as gcc, but "configure" doesn't seem
3533 to know about it. I have put a hack into configure.in that adds in code
3534 to set GCC=yes if CC=icc. This seems to end up at a point in the
3535 generated configure script that is early enough to affect the setting of
3536 compiler options, which is what is needed, but I have no means of testing
3537 whether it really works. (The user who reported this had patched the
3538 generated configure script, which of course I cannot do.)
3539
3540 LATER: After change 22 below (new libtool files), the configure script
3541 seems to know about icc (and also ecc). Therefore, I have commented out
3542 this hack in configure.in.
3543
3544 7. Added support for pkg-config (2 patches were sent in).
3545
3546 8. Negated POSIX character classes that used a combination of internal tables
3547 were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
3548 [[:^ascii]]. Typically, they would match almost any characters. The other
3549 POSIX classes were not broken in this way.
3550
3551 9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
3552 to find the match, as PCRE was deluded into thinking that the match had to
3553 start at the start point or following a newline. The same bug applied to
3554 patterns with negative forward assertions or any backward assertions
3555 preceding ".*" at the start, unless the pattern required a fixed first
3556 character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
3557
3558 10. In UTF-8 mode, when moving forwards in the subject after a failed match
3559 starting at the last subject character, bytes beyond the end of the subject
3560 string were read.
3561
3562 11. Renamed the variable "class" as "classbits" to make life easier for C++
3563 users. (Previously there was a macro definition, but it apparently wasn't
3564 enough.)
3565
3566 12. Added the new field "tables" to the extra data so that tables can be passed
3567 in at exec time, or the internal tables can be re-selected. This allows
3568 a compiled regex to be saved and re-used at a later time by a different
3569 program that might have everything at different addresses.
3570
3571 13. Modified the pcre-config script so that, when run on Solaris, it shows a
3572 -R library as well as a -L library.
3573
3574 14. The debugging options of pcretest (-d on the command line or D on a
3575 pattern) showed incorrect output for anything following an extended class
3576 that contained multibyte characters and which was followed by a quantifier.
3577
3578 15. Added optional support for general category Unicode character properties
3579 via the \p, \P, and \X escapes. Unicode property support implies UTF-8
3580 support. It adds about 90K to the size of the library. The meanings of the
3581 inbuilt class escapes such as \d and \s have NOT been changed.
3582
3583 16. Updated pcredemo.c to include calls to free() to release the memory for the
3584 compiled pattern.
3585
3586 17. The generated file chartables.c was being created in the source directory
3587 instead of in the building directory. This caused the build to fail if the
3588 source directory was different from the building directory, and was
3589 read-only.
3590
3591 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
3592 file. No doubt somebody will tell me if they don't make sense... Also added
3593 Dan Mooney's comments about building on OpenVMS.
3594
3595 19. Added support for partial matching via the PCRE_PARTIAL option for
3596 pcre_exec() and the \P data escape in pcretest.
3597
3598 20. Extended pcretest with 3 new pattern features:
3599
3600 (i) A pattern option of the form ">rest-of-line" causes pcretest to
3601 write the compiled pattern to the file whose name is "rest-of-line".
3602 This is a straight binary dump of the data, with the saved pointer to
3603 the character tables forced to be NULL. The study data, if any, is
3604 written too. After writing, pcretest reads a new pattern.
3605
3606 (ii) If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
3607 compiled pattern from the given file. There must not be any
3608 occurrences of "<" in the file name (pretty unlikely); if there are,
3609 pcretest will instead treat the initial "<" as a pattern delimiter.
3610 After reading in the pattern, pcretest goes on to read data lines as
3611 usual.
3612
3613 (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
3614 and 16-bit fields in a compiled pattern, to simulate a pattern that
3615 was compiled on a host of opposite endianness.
3616
3617 21. The pcre-exec() function can now cope with patterns that were compiled on
3618 hosts of opposite endianness, with this restriction:
3619
3620 As for any compiled expression that is saved and used later, the tables
3621 pointer field cannot be preserved; the extra_data field in the arguments
3622 to pcre_exec() should be used to pass in a tables address if a value
3623 other than the default internal tables were used at compile time.
3624
3625 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
3626 now diagnosed as an error. Previously, most of the time, a negative number
3627 would have been treated as zero, but if in addition "ovector" was passed as
3628 NULL, a crash could occur.
3629
3630 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
3631 new versions from the libtool 1.5 distribution (the last one is a copy of
3632 a file called libtool.m4). This seems to have fixed the need to patch
3633 "configure" to support Darwin 1.3 (which I used to do). However, I still
3634 had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
3635 workstation).
3636
3637 24. Changed the PCRE licence to be the more standard "BSD" licence.
3638
3639
3640 Version 4.5 01-Dec-03
3641 ---------------------
3642
3643 1. There has been some re-arrangement of the code for the match() function so
3644 that it can be compiled in a version that does not call itself recursively.
3645 Instead, it keeps those local variables that need separate instances for
3646 each "recursion" in a frame on the heap, and gets/frees frames whenever it
3647 needs to "recurse". Keeping track of where control must go is done by means
3648 of setjmp/longjmp. The whole thing is implemented by a set of macros that
3649 hide most of the details from the main code, and operates only if
3650 NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
3651 "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
3652 operating.
3653
3654 To make it easier for callers to provide specially tailored get/free
3655 functions for this usage, two new functions, pcre_stack_malloc, and
3656 pcre_stack_free, are used. They are always called in strict stacking order,
3657 and the size of block requested is always the same.
3658
3659 The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
3660 PCRE has been compiled to use the stack or the heap for recursion. The
3661 -C option of pcretest uses this to show which version is compiled.
3662
3663 A new data escape \S, is added to pcretest; it causes the amounts of store
3664 obtained and freed by both kinds of malloc/free at match time to be added
3665 to the output.
3666
3667 2. Changed the locale test to use "fr_FR" instead of "fr" because that's
3668 what's available on my current Linux desktop machine.
3669
3670 3. When matching a UTF-8 string, the test for a valid string at the start has
3671 been extended. If start_offset is not zero, PCRE now checks that it points
3672 to a byte that is the start of a UTF-8 character. If not, it returns
3673 PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
3674 this is necessary because there may be backward assertions in the pattern.
3675 When matching the same subject several times, it may save resources to use
3676 PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
3677
3678 4. The code for checking the validity of UTF-8 strings has been tightened so
3679 that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
3680 containing "overlong sequences".
3681
3682 5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
3683 I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
3684 should have been "&", but it just so happened that all the cases this let
3685 through by mistake were picked up later in the function.
3686
3687 6. I had used a variable called "isblank" - this is a C99 function, causing
3688 some compilers to warn. To avoid this, I renamed it (as "blankclass").
3689
3690 7. Cosmetic: (a) only output another newline at the end of pcretest if it is
3691 prompting; (b) run "./pcretest /dev/null" at the start of the test script
3692 so the version is shown; (c) stop "make test" echoing "./RunTest".
3693
3694 8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
3695
3696 9. The prototype for memmove() for systems that don't have it was using
3697 size_t, but the inclusion of the header that defines size_t was later. I've
3698 moved the #includes for the C headers earlier to avoid this.
3699
3700 10. Added some adjustments to the code to make it easier to compiler on certain
3701 special systems:
3702
3703 (a) Some "const" qualifiers were missing.
3704 (b) Added the macro EXPORT before all exported functions; by default this
3705 is defined to be empty.
3706 (c) Changed the dftables auxiliary program (that builds chartables.c) so
3707 that it reads its output file name as an argument instead of writing
3708 to the standard output and assuming this can be redirected.
3709
3710 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
3711 class containing characters with values greater than 255, PCRE compilation
3712 went into a loop.
3713
3714 12. A recursive reference to a subpattern that was within another subpattern
3715 that had a minimum quantifier of zero caused PCRE to crash. For example,
3716 (x(y(?2))z)? provoked this bug with a subject that got as far as the
3717 recursion. If the recursively-called subpattern itself had a zero repeat,
3718 that was OK.
3719
3720 13. In pcretest, the buffer for reading a data line was set at 30K, but the
3721 buffer into which it was copied (for escape processing) was still set at
3722 1024, so long lines caused crashes.
3723
3724 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
3725 "internal error: code overflow...". This applied to any character class
3726 that was followed by a possessive quantifier.
3727
3728 15. Modified the Makefile to add libpcre.la as a prerequisite for
3729 libpcreposix.la because I was told this is needed for a parallel build to
3730 work.
3731
3732 16. If a pattern that contained .* following optional items at the start was
3733 studied, the wrong optimizing data was generated, leading to matching
3734 errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
3735 matching string must start with a or b or c. The correct conclusion for
3736 this pattern is that a match can start with any character.
3737
3738
3739 Version 4.4 13-Aug-03
3740 ---------------------
3741
3742 1. In UTF-8 mode, a character class containing characters with values between
3743 127 and 255 was not handled correctly if the compiled pattern was studied.
3744 In fixing this, I have also improved the studying algorithm for such
3745 classes (slightly).
3746
3747 2. Three internal functions had redundant arguments passed to them. Removal
3748 might give a very teeny performance improvement.
3749
3750 3. Documentation bug: the value of the capture_top field in a callout is *one
3751 more than* the number of the hightest numbered captured substring.
3752
3753 4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
3754 in incorrectly linking with a previously installed version. They now link
3755 explicitly with libpcre.la.
3756
3757 5. configure.in no longer needs to recognize Cygwin specially.
3758
3759 6. A problem in pcre.in for Windows platforms is fixed.
3760
3761 7. If a pattern was successfully studied, and the -d (or /D) flag was given to
3762 pcretest, it used to include the size of the study block as part of its
3763 output. Unfortunately, the structure contains a field that has a different
3764 size on different hardware architectures. This meant that the tests that
3765 showed this size failed. As the block is currently always of a fixed size,
3766 this information isn't actually particularly useful in pcretest output, so
3767 I have just removed it.
3768
3769 8. Three pre-processor statements accidentally did not start in column 1.
3770 Sadly, there are *still* compilers around that complain, even though
3771 standard C has not required this for well over a decade. Sigh.
3772
3773 9. In pcretest, the code for checking callouts passed small integers in the
3774 callout_data field, which is a void * field. However, some picky compilers
3775 complained about the casts involved for this on 64-bit systems. Now
3776 pcretest passes the address of the small integer instead, which should get
3777 rid of the warnings.
3778
3779 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
3780 both compile and run time, and gives an error if an invalid UTF-8 sequence
3781 is found. There is a option for disabling this check in cases where the
3782 string is known to be correct and/or the maximum performance is wanted.
3783
3784 11. In response to a bug report, I changed one line in Makefile.in from
3785
3786 -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
3787 to
3788 -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
3789
3790 to look similar to other lines, but I have no way of telling whether this
3791 is the right thing to do, as I do not use Windows. No doubt I'll get told
3792 if it's wrong...
3793
3794
3795 Version 4.3 21-May-03
3796 ---------------------
3797
3798 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
3799 Makefile.
3800
3801 2. Some refactoring to improve the quality of the code:
3802
3803 (i) The utf8_table... variables are now declared "const".
3804
3805 (ii) The code for \cx, which used the "case flipping" table to upper case
3806 lower case letters, now just substracts 32. This is ASCII-specific,
3807 but the whole concept of \cx is ASCII-specific, so it seems
3808 reasonable.
3809
3810 (iii) PCRE was using its character types table to recognize decimal and
3811 hexadecimal digits in the pattern. This is silly, because it handles
3812 only 0-9, a-f, and A-F, but the character types table is locale-
3813 specific, which means strange things might happen. A private
3814 table is now used for this - though it costs 256 bytes, a table is
3815 much faster than multiple explicit tests. Of course, the standard
3816 character types table is still used for matching digits in subject
3817 strings against \d.
3818
3819 (iv) Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
3820 ending in _t are). So I've renamed it as ESC_tee.
3821
3822 3. The first argument for regexec() in the POSIX wrapper should have been
3823 defined as "const".
3824
3825 4. Changed pcretest to use malloc() for its buffers so that they can be
3826 Electric Fenced for debugging.
3827
3828 5. There were several places in the code where, in UTF-8 mode, PCRE would try
3829 to read one or more bytes before the start of the subject string. Often this
3830 had no effect on PCRE's behaviour, but in some circumstances it could
3831 provoke a segmentation fault.
3832
3833 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
3834 to try to read one or more bytes before the start of the subject string.
3835
3836 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
3837 UTF-8 support could misbehave in various ways if the subject string
3838 contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
3839 area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
3840 back over UTF-8 characters.)
3841
3842
3843 Version 4.2 14-Apr-03
3844 ---------------------
3845
3846 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
3847
3848 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
3849 [ON_WINDOWS]: new variable, "#" on non-Windows platforms
3850 [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
3851 [WIN_PREFIX]: new variable, "cyg" for Cygwin
3852 * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
3853 and BUILD_EXEEXT
3854 Note: automatic setting of the BUILD variables is not yet working
3855 set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
3856 compile-time but not at link-time
3857 [LINK]: use for linking executables only
3858 make different versions for Windows and non-Windows
3859 [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
3860 libraries
3861 [LINK_FOR_BUILD]: new variable
3862 [OBJEXT]: use throughout
3863 [EXEEXT]: use throughout
3864 <winshared>: new target
3865 <wininstall>: new target
3866 <dftables.o>: use native compiler
3867 <dftables>: use native linker
3868 <install>: handle Windows platform correctly
3869 <clean>: ditto
3870 <check>: ditto
3871 copy DLL to top builddir before testing
3872
3873 As part of these changes, -no-undefined was removed again. This was reported
3874 to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
3875 in any case.
3876
3877 3. Some tidies to get rid of compiler warnings:
3878
3879 . In the match_data structure, match_limit was an unsigned long int, whereas
3880 match_call_count was an int. I've made them both unsigned long ints.
3881
3882 . In pcretest the fact that a const uschar * doesn't automatically cast to
3883 a void * provoked a warning.
3884
3885 . Turning on some more compiler warnings threw up some "shadow" variables
3886 and a few more missing casts.
3887
3888 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
3889 option, a class that contained a single character with a value between 128
3890 and 255 (e.g. /[\xFF]/) caused PCRE to crash.
3891
3892 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
3893 option, a class that contained several characters, but with at least one
3894 whose value was between 128 and 255 caused PCRE to crash.
3895
3896
3897 Version 4.1 12-Mar-03
3898 ---------------------
3899
3900 1. Compiling with gcc -pedantic found a couple of places where casts were
3901 needed, and a string in dftables.c that was longer than standard compilers are
3902 required to support.
3903
3904 2. Compiling with Sun's compiler found a few more places where the code could
3905 be tidied up in order to avoid warnings.
3906
3907 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
3908 first of these names is deprecated in the latest Autoconf in favour of the name
3909 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
3910 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
3911 analogy I have changed it to CFLAGS_FOR_BUILD.
3912
3913 4. Added -no-undefined to the linking command in the Makefile, because this is
3914 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
3915 linking step for the pcreposix library.
3916
3917 5. PCRE was failing to diagnose the case of two named groups with the same
3918 name.
3919
3920 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
3921 literal character that is needed in the subject for a match, and scans along to
3922 ensure that it is present before embarking on the full matching process. This
3923 saves time in cases of nested unlimited repeats that are never going to match.
3924 Problem: the scan can take a lot of time if the subject is very long (e.g.
3925 megabytes), thus penalizing straightforward matches. It is now done only if the
3926 amount of subject to be scanned is less than 1000 bytes.
3927
3928 7. A lesser problem with the same optimization is that it was recording the
3929 first character of an anchored pattern as "needed", thus provoking a search
3930 right along the subject, even when the first match of the pattern was going to
3931 fail. The "needed" character is now not set for anchored patterns, unless it
3932 follows something in the pattern that is of non-fixed length. Thus, it still
3933 fulfils its original purpose of finding quick non-matches in cases of nested
3934 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
3935
3936
3937 Version 4.0 17-Feb-03
3938 ---------------------
3939
3940 1. If a comment in an extended regex that started immediately after a meta-item
3941 extended to the end of string, PCRE compiled incorrect data. This could lead to
3942 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
3943
3944 2. Moved to autoconf 2.53 and libtool 1.4.2.
3945
3946 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
3947 the special perltest8 script is no longer needed - all the tests can be run
3948 from a single perltest script.
3949
3950 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
3951 by \s. It has now been removed in PCRE. This means it isn't recognized as
3952 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
3953 class [:space:] *does* include VT, thereby creating a mess.
3954
3955 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
3956 space and tab.
3957
3958 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
3959 its new features into the main test script, reducing the number of scripts.
3960
3961 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
3962 were backward compatible, and made the (?i) apply to the whole pattern, as if
3963 /i were given. Now it behaves more logically, and applies the option setting
3964 only to what follows. PCRE has been changed to follow suit. However, if it
3965 finds options settings right at the start of the pattern, it extracts them into
3966 the global options, as before. Thus, they show up in the info data.
3967
3968 8. Added support for the \Q...\E escape sequence. Characters in between are
3969 treated as literals. This is slightly different from Perl in that $ and @ are
3970 also handled as literals inside the quotes. In Perl, they will cause variable
3971 interpolation. Note the following examples:
3972
3973 Pattern PCRE matches Perl matches
3974
3975 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
3976 \Qabc\$xyz\E abc\$xyz abc\$xyz
3977 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
3978
3979 For compatibility with Perl, \Q...\E sequences are recognized inside character
3980 classes as well as outside them.
3981
3982 9. Re-organized 3 code statements in pcretest to avoid "overflow in
3983 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
3984 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
3985 signed/unsigned warnings.
3986
3987 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
3988 option for pcretest, so I've replaced it by a simple function that does just
3989 that job.
3990
3991 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
3992 "pcregrep -".
3993
3994 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
3995 Java package. This provides some syntactic sugar for simple cases of what my
3996 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
3997 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
3998 item, you can use this simplified notation. Note that only makes sense with
3999 greedy quantifiers. Consequently, the use of the possessive quantifier forces
4000 greediness, whatever the setting of the PCRE_UNGREEDY option.
4001
4002 13. A change of greediness default within a pattern was not taking effect at
4003 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
4004 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
4005 was abstracted outside.
4006
4007 14. PCRE now supports the \G assertion. It is true when the current matching
4008 position is at the start point of the match. This differs from \A when the
4009 starting offset is non-zero. Used with the /g option of pcretest (or similar
4010 code), it works in the same way as it does for Perl's /g option. If all
4011 alternatives of a regex begin with \G, the expression is anchored to the start
4012 match position, and the "anchored" flag is set in the compiled expression.
4013
4014 15. Some bugs concerning the handling of certain option changes within patterns
4015 have been fixed. These applied to options other than (?ims). For example,
4016 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
4017 been the other way round. Some of this was related to change 7 above.
4018
4019 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
4020 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
4021 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
4022 POSIX classes only within a class (e.g. /[[:alpha:]]/).
4023
4024 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
4025 mode. Unlike ".", it always matches newline, whatever the setting of
4026 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
4027 assertions. Perl allows it, but it doesn't (in general) work because it can't
4028 calculate the length of the lookbehind. At least, that's the case for Perl
4029 5.8.0 - I've been told they are going to document that it doesn't work in
4030 future.
4031
4032 18. Added an error diagnosis for escapes that PCRE does not support: these are
4033 \L, \l, \N, \P, \p, \U, \u, and \X.
4034
4035 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
4036 reading past the end of the pattern in cases such as /[abcd/.
4037
4038 20. PCRE was getting more memory than necessary for patterns with classes that
4039 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
4040
4041 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
4042 compiling PCRE for use with Virtual Pascal.
4043
4044 22. Small fix to the Makefile to make it work properly if the build is done
4045 outside the source tree.
4046
4047 23. Added a new extension: a condition to go with recursion. If a conditional
4048 subpattern starts with (?(R) the "true" branch is used if recursion has
4049 happened, whereas the "false" branch is used only at the top level.
4050
4051 24. When there was a very long string of literal characters (over 255 bytes
4052 without UTF support, over 250 bytes with UTF support), the computation of how
4053 much memory was required could be incorrect, leading to segfaults or other
4054 strange effects.
4055
4056 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
4057 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
4058 there was a subsequent back reference to those brackets. This meant that, for
4059 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
4060 possible to check for precisely this case. All we can do is abandon the
4061 optimization if .* occurs inside capturing brackets when there are any back
4062 references whatsoever. (See below for a better fix that came later.)
4063
4064 26. The handling of the optimization for finding the first character of a
4065 non-anchored pattern, and for finding a character that is required later in the
4066 match were failing in some cases. This didn't break the matching; it just
4067 failed to optimize when it could. The way this is done has been re-implemented.
4068
4069 27. Fixed typo in error message for invalid (?R item (it said "(?p").
4070
4071 28. Added a new feature that provides some of the functionality that Perl
4072 provides with (?{...}). The facility is termed a "callout". The way it is done
4073 in PCRE is for the caller to provide an optional function, by setting
4074 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
4075 global variable. By default it is unset, which disables all calling out. To get
4076 the function called, the regex must include (?C) at appropriate points. This
4077 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
4078 This provides a means of identifying different callout points. When PCRE
4079 reaches such a point in the regex, if pcre_callout has been set, the external
4080 function is called. It is provided with data in a structure called
4081 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
4082 matching continues; if it returns a non-zero value, the match at the current
4083 point fails. However, backtracking will occur if possible. [This was changed
4084 later and other features added - see item 49 below.]
4085
4086 29. pcretest is upgraded to test the callout functionality. It provides a
4087 callout function that displays information. By default, it shows the start of
4088 the match and the current position in the text. There are some new data escapes
4089 to vary what happens:
4090
4091 \C+ in addition, show current contents of captured substrings
4092 \C- do not supply a callout function
4093 \C!n return 1 when callout number n is reached
4094 \C!n!m return 1 when callout number n is reached for the mth time
4095
4096 30. If pcregrep was called with the -l option and just a single file name, it
4097 output "<stdin>" if a match was found, instead of the file name.
4098
4099 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
4100 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
4101 pcre_exec(). This saves a malloc/free per call. The default value of
4102 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
4103 when configuring.
4104
4105 32. The default maximum size of a compiled pattern is 64K. There have been a
4106 few cases of people hitting this limit. The code now uses macros to handle the
4107 storing of links as offsets within the compiled pattern. It defaults to 2-byte
4108 links, but this can be changed to 3 or 4 bytes by --with-link-size when
4109 configuring. Tests 2 and 5 work only with 2-byte links because they output
4110 debugging information about compiled patterns.
4111
4112 33. Internal code re-arrangements:
4113
4114 (a) Moved the debugging function for printing out a compiled regex into
4115 its own source file (printint.c) and used #include to pull it into
4116 pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
4117 separate copies.
4118
4119 (b) Defined the list of op-code names for debugging as a macro in
4120 internal.h so that it is next to the definition of the opcodes.
4121
4122 (c) Defined a table of op-code lengths for simpler skipping along compiled
4123 code. This is again a macro in internal.h so that it is next to the
4124 definition of the opcodes.
4125
4126 34. Added support for recursive calls to individual subpatterns, along the
4127 lines of Robin Houston's patch (but implemented somewhat differently).
4128
4129 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
4130 allow it to read and process whole directories in Win32. This code was
4131 contributed by Lionel Fourquaux; it has not been tested by me.
4132
4133 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
4134 used to name a group. Names consist of alphanumerics and underscores, and must
4135 be unique. Back references use the syntax (?P=name) and recursive calls use
4136 (?P>name) which is a PCRE extension to the Python extension. Groups still have
4137 numbers. The function pcre_fullinfo() can be used after compilation to extract
4138 a name/number map. There are three relevant calls:
4139
4140 PCRE_INFO_NAMEENTRYSIZE yields the size of each entry in the map
4141 PCRE_INFO_NAMECOUNT yields the number of entries
4142 PCRE_INFO_NAMETABLE yields a pointer to the map.
4143
4144 The map is a vector of fixed-size entries. The size of each entry depends on
4145 the length of the longest name used. The first two bytes of each entry are the
4146 group number, most significant byte first. There follows the corresponding
4147 name, zero terminated. The names are in alphabetical order.
4148
4149 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
4150 case instead of 255. Making it the same both with and without UTF-8 support
4151 means that the same test output works with both.
4152
4153 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
4154 calling malloc() with a zero argument.
4155
4156 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
4157 optimization. I've improved things by keeping a bitmap of backreferences with
4158 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
4159 fact referenced, the optimization can be applied. It is unlikely that a
4160 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
4161 the match to follow \n) will appear inside brackets with a number greater than
4162 31, but if it does, any back reference > 31 suppresses the optimization.
4163
4164 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
4165 of disabling numbered capturing parentheses. Any opening parenthesis that is
4166 not followed by ? behaves as if it were followed by ?: but named parentheses
4167 can still be used for capturing (and they will acquire numbers in the usual
4168 way).
4169
4170 41. Redesigned the return codes from the match() function into yes/no/error so
4171 that errors can be passed back from deep inside the nested calls. A malloc
4172 failure while inside a recursive subpattern call now causes the
4173 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
4174
4175 42. It is now possible to set a limit on the number of times the match()
4176 function is called in a call to pcre_exec(). This facility makes it possible to
4177 limit the amount of recursion and backtracking, though not in a directly
4178 obvious way, because the match() function is used in a number of different
4179 circumstances. The count starts from zero for each position in the subject
4180 string (for non-anchored patterns). The default limit is, for compatibility, a
4181 large number, namely 10 000 000. You can change this in two ways:
4182
4183 (a) When configuring PCRE before making, you can use --with-match-limit=n
4184 to set a default value for the compiled library.
4185
4186 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
4187 a different value is set. See 45 below.
4188
4189 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
4190
4191 43. Added a new function pcre_config(int, void *) to enable run-time extraction
4192 of things that can be changed at compile time. The first argument specifies
4193 what is wanted and the second points to where the information is to be placed.
4194 The current list of available information is:
4195
4196 PCRE_CONFIG_UTF8
4197
4198 The output is an integer that is set to one if UTF-8 support is available;
4199 otherwise it is set to zero.
4200
4201 PCRE_CONFIG_NEWLINE
4202
4203 The output is an integer that it set to the value of the code that is used for
4204 newline. It is either LF (10) or CR (13).
4205
4206 PCRE_CONFIG_LINK_SIZE
4207
4208 The output is an integer that contains the number of bytes used for internal
4209 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
4210
4211 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
4212
4213 The output is an integer that contains the threshold above which the POSIX
4214 interface uses malloc() for output vectors. See item 31 above.
4215
4216 PCRE_CONFIG_MATCH_LIMIT
4217
4218 The output is an unsigned integer that contains the default limit of the number
4219 of match() calls in a pcre_exec() execution. See 42 above.
4220
4221 44. pcretest has been upgraded by the addition of the -C option. This causes it
4222 to extract all the available output from the new pcre_config() function, and to
4223 output it. The program then exits immediately.
4224
4225 45. A need has arisen to pass over additional data with calls to pcre_exec() in
4226 order to support additional features. One way would have been to define
4227 pcre_exec2() (for example) with extra arguments, but this would not have been
4228 extensible, and would also have required all calls to the original function to
4229 be mapped to the new one. Instead, I have chosen to extend the mechanism that
4230 is used for passing in "extra" data from pcre_study().
4231
4232 The pcre_extra structure is now exposed and defined in pcre.h. It currently
4233 contains the following fields:
4234
4235 flags a bitmap indicating which of the following fields are set
4236 study_data opaque data from pcre_study()
4237 match_limit a way of specifying a limit on match() calls for a specific
4238 call to pcre_exec()
4239 callout_data data for callouts (see 49 below)
4240
4241 The flag bits are also defined in pcre.h, and are
4242
4243 PCRE_EXTRA_STUDY_DATA
4244 PCRE_EXTRA_MATCH_LIMIT
4245 PCRE_EXTRA_CALLOUT_DATA
4246
4247 The pcre_study() function now returns one of these new pcre_extra blocks, with
4248 the actual study data pointed to by the study_data field, and the
4249 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
4250 before. That is, this change is entirely upwards-compatible and requires no
4251 change to existing code.
4252
4253 If you want to pass in additional data to pcre_exec(), you can either place it
4254 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
4255 block.
4256
4257 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
4258 data string contains the escape sequence \M, pcretest calls pcre_exec() several
4259 times with different match limits, until it finds the minimum value needed for
4260 pcre_exec() to complete. The value is then output. This can be instructive; for
4261 most simple matches the number is quite small, but for pathological cases it
4262 gets very large very quickly.
4263
4264 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
4265 returns the size of the data block pointed to by the study_data field in a
4266 pcre_extra block, that is, the value that was passed as the argument to
4267 pcre_malloc() when PCRE was getting memory in which to place the information
4268 created by pcre_study(). The fourth argument should point to a size_t variable.
4269 pcretest has been extended so that this information is shown after a successful
4270 pcre_study() call when information about the compiled regex is being displayed.
4271
4272 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
4273 because what follows is always an absolute path. (Later: it turns out that this
4274 is more than cosmetic for MinGW, because it doesn't like empty path
4275 components.)
4276
4277 49. Some changes have been made to the callout feature (see 28 above):
4278
4279 (i) A callout function now has three choices for what it returns:
4280
4281 0 => success, carry on matching
4282 > 0 => failure at this point, but backtrack if possible
4283 < 0 => serious error, return this value from pcre_exec()
4284
4285 Negative values should normally be chosen from the set of PCRE_ERROR_xxx
4286 values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
4287 "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
4288 use by callout functions. It will never be used by PCRE itself.
4289
4290 (ii) The pcre_extra structure (see 45 above) has a void * field called
4291 callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
4292 pcre_callout_block structure has a field of the same name. The contents of
4293 the field passed in the pcre_extra structure are passed to the callout
4294 function in the corresponding field in the callout block. This makes it
4295 easier to use the same callout-containing regex from multiple threads. For
4296 testing, the pcretest program has a new data escape
4297
4298 \C*n pass the number n (may be negative) as callout_data
4299
4300 If the callout function in pcretest receives a non-zero value as
4301 callout_data, it returns that value.
4302
4303 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
4304 there were some redundant $(CFLAGS) in commands that are now specified as
4305 $(LINK), which already includes $(CFLAGS).
4306
4307 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
4308 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
4309 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
4310 one-byte characters throughout. Note that case-insensitive matching applies
4311 only to characters whose values are less than 256. PCRE doesn't support the
4312 notion of cases for higher-valued characters.
4313
4314 (i) A character class whose characters are all within 0-255 is handled as
4315 a bit map, and the map is inverted for negative classes. Previously, a
4316 character > 255 always failed to match such a class; however it should
4317 match if the class was a negative one (e.g. [^ab]). This has been fixed.
4318
4319 (ii) A negated character class with a single character < 255 is coded as
4320 "not this character" (OP_NOT). This wasn't working properly when the test
4321 character was multibyte, either singly or repeated.
4322
4323 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
4324 mode, for example: \x{100}{2,3}.
4325
4326 (iv) The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
4327 singly or repeated) now correctly test multibyte characters. However,
4328 PCRE doesn't recognize any characters with values greater than 255 as
4329 digits, spaces, or word characters. Such characters always match \D, \S,
4330 and \W, and never match \d, \s, or \w.
4331
4332 (v) Classes may now contain characters and character ranges with values
4333 greater than 255. For example: [ab\x{100}-\x{400}].
4334
4335 (vi) pcregrep now has a --utf-8 option (synonym -u) which makes it call
4336 PCRE in UTF-8 mode.
4337
4338 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
4339 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
4340 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
4341 value.)
4342
4343 53. The single man page has become too large. I have therefore split it up into
4344 a number of separate man pages. These also give rise to individual HTML pages;
4345 these are now put in a separate directory, and there is an index.html page that
4346 lists them all. Some hyperlinking between the pages has been installed.
4347
4348 54. Added convenience functions for handling named capturing parentheses.
4349
4350 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
4351 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
4352 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
4353 are faulted.
4354
4355 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
4356 calling configure. These values are used when compiling the dftables.c program
4357 which is run to generate the source of the default character tables. They
4358 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
4359 you will need to set these values.
4360
4361 57. Updated the building process for Windows DLL, as provided by Fred Cox.
4362
4363
4364 Version 3.9 02-Jan-02
4365 ---------------------
4366
4367 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
4368
4369 2. If --disable-static was given, the building process failed when trying to
4370 build pcretest and pcregrep. (For some reason it was using libtool to compile
4371 them, which is not right, as they aren't part of the library.)
4372
4373
4374 Version 3.8 18-Dec-01
4375 ---------------------
4376
4377 1. The experimental UTF-8 code was completely screwed up. It was packing the
4378 bytes in the wrong order. How dumb can you get?
4379
4380
4381 Version 3.7 29-Oct-01
4382 ---------------------
4383
4384 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
4385 This caused pcretest, when used on the test data, to segfault. Unfortunately,
4386 this didn't happen under Solaris 8, where I normally test things.
4387
4388 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
4389 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
4390 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
4391 this fix an hour or so after the initial 3.7 release.)
4392
4393
4394 Version 3.6 23-Oct-01
4395 ---------------------
4396
4397 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
4398 offsets passed as NULL with zero offset count.
4399
4400 2. The config.guess and config.sub files had not been updated when I moved to
4401 the latest autoconf.
4402
4403
4404 Version 3.5 15-Aug-01
4405 ---------------------
4406
4407 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
4408 had been forgotten.
4409
4410 2. By using declared but undefined structures, we can avoid using "void"
4411 definitions in pcre.h while keeping the internal definitions of the structures
4412 private.
4413
4414 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
4415 user point of view, this means that both static and shared libraries are built
4416 by default, but this can be individually controlled. More of the work of
4417 handling this static/shared cases is now inside libtool instead of PCRE's make
4418 file.
4419
4420 4. The pcretest utility is now installed along with pcregrep because it is
4421 useful for users (to test regexs) and by doing this, it automatically gets
4422 relinked by libtool. The documentation has been turned into a man page, so
4423 there are now .1, .txt, and .html versions in /doc.
4424
4425 5. Upgrades to pcregrep:
4426 (i) Added long-form option names like gnu grep.
4427 (ii) Added --help to list all options with an explanatory phrase.
4428 (iii) Added -r, --recursive to recurse into sub-directories.
4429 (iv) Added -f, --file to read patterns from a file.
4430
4431 6. pcre_exec() was referring to its "code" argument before testing that
4432 argument for NULL (and giving an error if it was NULL).
4433
4434 7. Upgraded Makefile.in to allow for compiling in a different directory from
4435 the source directory.
4436
4437 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
4438 options bits, the pointer it was passed was to an int instead of to an unsigned
4439 long int. This mattered only on 64-bit systems.
4440
4441 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
4442 generated) instead of pcre.in, which it its source. Also made the same change
4443 in several of the .c files.
4444
4445 10. A new release of gcc defines printf() as a macro, which broke pcretest
4446 because it had an ifdef in the middle of a string argument for printf(). Fixed
4447 by using separate calls to printf().
4448
4449 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
4450 script, to force use of CR or LF instead of \n in the source. On non-Unix
4451 systems, the value can be set in config.h.
4452
4453 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
4454 absolute limit. Changed the text of the error message to make this clear, and
4455 likewise updated the man page.
4456
4457 13. The limit of 99 on the number of capturing subpatterns has been removed.
4458 The new limit is 65535, which I hope will not be a "real" limit.
4459
4460
4461 Version 3.4 22-Aug-00
4462 ---------------------
4463
4464 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
4465
4466 2. Diagnose condition (?(0) as an error instead of crashing on matching.
4467
4468
4469 Version 3.3 01-Aug-00
4470 ---------------------
4471
4472 1. If an octal character was given, but the value was greater than \377, it
4473 was not getting masked to the least significant bits, as documented. This could
4474 lead to crashes in some systems.
4475
4476 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
4477 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
4478
4479 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
4480 These just pass their arguments on to (pcre_free)(), but they are provided
4481 because some uses of PCRE bind it to non-C systems that can call its functions,
4482 but cannot call free() or pcre_free() directly.
4483
4484 4. Add "make test" as a synonym for "make check". Corrected some comments in
4485 the Makefile.
4486
4487 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
4488 Makefile.
4489
4490 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
4491 command called pgrep for grepping around the active processes.
4492
4493 7. Added the beginnings of support for UTF-8 character strings.
4494
4495 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
4496 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
4497 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
4498 out for the ar command.)
4499
4500
4501 Version 3.2 12-May-00
4502 ---------------------
4503
4504 This is purely a bug fixing release.
4505
4506 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
4507 of ZA. This was just one example of several cases that could provoke this bug,
4508 which was introduced by change 9 of version 2.00. The code for breaking
4509 infinite loops after an iteration that matches an empty string was't working
4510 correctly.
4511
4512 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
4513 when matched against abbab (for example). After matching an empty string, it
4514 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
4515 caused it to match further down the string than it should.
4516
4517 3. The code contained an inclusion of sys/types.h. It isn't clear why this
4518 was there because it doesn't seem to be needed, and it causes trouble on some
4519 systems, as it is not a Standard C header. It has been removed.
4520
4521 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
4522 were reported on the Macintosh. The changes were from
4523
4524 while ((c = *(++ptr)) != 0 && c != '\n');
4525 to
4526 while ((c = *(++ptr)) != 0 && c != '\n') ;
4527
4528 Totally extraordinary, but if that's what it takes...
4529
4530 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
4531 available. Added HAVE_BCOPY and an autoconf test for it; if neither
4532 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
4533 assumes the way PCRE uses memmove() (always moving upwards).
4534
4535 6. PCRE is being used in one environment where strchr() is not available. There
4536 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
4537 faster code anyway.
4538
4539
4540 Version 3.1 09-Feb-00
4541 ---------------------
4542
4543 The only change in this release is the fixing of some bugs in Makefile.in for
4544 the "install" target:
4545
4546 (1) It was failing to install pcreposix.h.
4547
4548 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
4549
4550
4551 Version 3.0 01-Feb-00
4552 ---------------------
4553
4554 1. Add support for the /+ modifier to perltest (to output $` like it does in
4555 pcretest).
4556
4557 2. Add support for the /g modifier to perltest.
4558
4559 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
4560 matches null strings.
4561
4562 4. Fix perltest so that it doesn't do unwanted things when fed an empty
4563 pattern. Perl treats empty patterns specially - it reuses the most recent
4564 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
4565 effect.
4566
4567 5. The POSIX interface was broken in that it was just handing over the POSIX
4568 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
4569 required a bigger vector, with some working space on the end. This means that
4570 the POSIX wrapper now has to get and free some memory, and copy the results.
4571
4572 6. Added some simple autoconf support, placing the test data and the
4573 documentation in separate directories, re-organizing some of the
4574 information files, and making it build pcre-config (a GNU standard). Also added
4575 libtool support for building PCRE as a shared library, which is now the
4576 default.
4577
4578 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
4579 09 are not valid octal constants. Single digits will be used for minor values
4580 less than 10.
4581
4582 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
4583 existing programs that set these in the POSIX interface can use PCRE without
4584 modification.
4585
4586 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
4587 return all that pcre_info() returns, plus additional data. The pcre_info()
4588 function is retained for compatibility, but is considered to be obsolete.
4589
4590 10. Added experimental recursion feature (?R) to handle one common case that
4591 Perl 5.6 will be able to do with (?p{...}).
4592
4593 11. Added support for POSIX character classes like [:alpha:], which Perl is
4594 adopting.
4595
4596
4597 Version 2.08 31-Aug-99
4598 ----------------------
4599
4600 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
4601 trying to match at the startoffset position, but instead was moving forward to
4602 the next newline as if a previous match had failed.
4603
4604 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
4605 and could get into a loop if a null string was matched other than at the start
4606 of the subject.
4607
4608 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
4609 be distinguished at compile time, and for completeness also added PCRE_DATE.
4610
4611 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
4612 in GnuWin32 environments.
4613
4614
4615 Version 2.07 29-Jul-99
4616 ----------------------
4617
4618 1. The documentation is now supplied in plain text form and HTML as well as in
4619 the form of man page sources.
4620
4621 2. C++ compilers don't like assigning (void *) values to other pointer types.
4622 In particular this affects malloc(). Although there is no problem in Standard
4623 C, I've put in casts to keep C++ compilers happy.
4624
4625 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
4626 should be (const char *).
4627
4628 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
4629 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
4630 However, I haven't made this a standard facility. The documentation doesn't
4631 mention it, and the Makefile doesn't support it.
4632
4633 5. The Makefile now contains an "install" target, with editable destinations at
4634 the top of the file. The pcretest program is not installed.
4635
4636 6. pgrep -V now gives the PCRE version number and date.
4637
4638 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
4639 causing the entire string to be ignored, instead of just the last character.
4640
4641 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
4642 non-matching string, it can take a very, very long time, even for strings of
4643 quite modest length, because of the nested recursion. PCRE now does better in
4644 some of these cases. It does this by remembering the last required literal
4645 character in the pattern, and pre-searching the subject to ensure it is present
4646 before running the real match. In other words, it applies a heuristic to detect
4647 some types of certain failure quickly, and in the above example, if presented
4648 with a string that has no trailing " it gives "no match" very quickly.
4649
4650 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
4651 other alternatives are tried instead.
4652
4653
4654 Version 2.06 09-Jun-99
4655 ----------------------
4656
4657 1. Change pcretest's output for amount of store used to show just the code
4658 space, because the remainder (the data block) varies in size between 32-bit and
4659 64-bit systems.
4660
4661 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
4662 start matching at. This allows lookbehinds to work when searching for multiple
4663 occurrences in a string.
4664
4665 3. Added additional options to pcretest for testing multiple occurrences:
4666
4667 /+ outputs the rest of the string that follows a match
4668 /g loops for multiple occurrences, using the new startoffset argument
4669 /G loops for multiple occurrences by passing an incremented pointer
4670
4671 4. PCRE wasn't doing the "first character" optimization for patterns starting
4672 with \b or \B, though it was doing it for other lookbehind assertions. That is,
4673 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
4674 the letter 'x'. On long subject strings, this gives a significant speed-up.
4675
4676
4677 Version 2.05 21-Apr-99
4678 ----------------------
4679
4680 1. Changed the type of magic_number from int to long int so that it works
4681 properly on 16-bit systems.
4682
4683 2. Fixed a bug which caused patterns starting with .* not to work correctly
4684 when the subject string contained newline characters. PCRE was assuming
4685 anchoring for such patterns in all cases, which is not correct because .* will
4686 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
4687 DOTALL is set at top level; otherwise it knows that patterns starting with .*
4688 must be retried after every newline in the subject.
4689
4690
4691 Version 2.04 18-Feb-99
4692 ----------------------
4693
4694 1. For parenthesized subpatterns with repeats whose minimum was zero, the
4695 computation of the store needed to hold the pattern was incorrect (too large).
4696 If such patterns were nested a few deep, this could multiply and become a real
4697 problem.
4698
4699 2. Added /M option to pcretest to show the memory requirement of a specific
4700 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
4701
4702 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
4703 compiled in such a way that the backtracking after subsequent failure was
4704 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
4705 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
4706
4707
4708 Version 2.03 02-Feb-99
4709 ----------------------
4710
4711 1. Fixed typo and small mistake in man page.
4712
4713 2. Added 4th condition (GPL supersedes if conflict) and created separate
4714 LICENCE file containing the conditions.
4715
4716 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
4717 Perl, that is the internal \ allows the delimiter to be included in the
4718 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
4719 the final delimiter, add \ to the end of the pattern (to test the error).
4720
4721 4. Added the convenience functions for extracting substrings after a successful
4722 match. Updated pcretest to make it able to test these functions.
4723
4724
4725 Version 2.02 14-Jan-99
4726 ----------------------
4727
4728 1. Initialized the working variables associated with each extraction so that
4729 their saving and restoring doesn't refer to uninitialized store.
4730
4731 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
4732 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
4733 fix the problem.
4734
4735 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
4736 calls, and wasn't printing the correct value for compiling calls. Increased the
4737 default value of LOOPREPEAT, and the number of significant figures in the
4738 times.
4739
4740 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
4741
4742 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
4743 a building problem on Windows NT with a FAT file system.
4744
4745
4746 Version 2.01 21-Oct-98
4747 ----------------------
4748
4749 1. Changed the API for pcre_compile() to allow for the provision of a pointer
4750 to character tables built by pcre_maketables() in the current locale. If NULL
4751 is passed, the default tables are used.
4752
4753
4754 Version 2.00 24-Sep-98
4755 ----------------------
4756
4757 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
4758 it any more.
4759
4760 2. Allow quantification of (?>) groups, and make it work correctly.
4761
4762 3. The first character computation wasn't working for (?>) groups.
4763
4764 4. Correct the implementation of \Z (it is permitted to match on the \n at the
4765 end of the subject) and add 5.005's \z, which really does match only at the
4766 very end of the subject.
4767
4768 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
4769
4770 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
4771 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
4772 localized options. All options to pcre_study() were also removed.
4773
4774 7. Add other new features from 5.005:
4775
4776 $(?<= positive lookbehind
4777 $(?<! negative lookbehind
4778 (?imsx-imsx) added the unsetting capability
4779 such a setting is global if at outer level; local otherwise
4780 (?imsx-imsx:) non-capturing groups with option setting
4781 (?(cond)re|re) conditional pattern matching
4782
4783 A backreference to itself in a repeated group matches the previous
4784 captured string.
4785
4786 8. General tidying up of studying (both automatic and via "study")
4787 consequential on the addition of new assertions.
4788
4789 9. As in 5.005, unlimited repeated groups that could match an empty substring
4790 are no longer faulted at compile time. Instead, the loop is forcibly broken at
4791 runtime if any iteration does actually match an empty substring.
4792
4793 10. Include the RunTest script in the distribution.
4794
4795 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
4796 discrepancies, some of which were old and were also with respect to 5.004. They
4797 have now been fixed.
4798
4799
4800 Version 1.09 28-Apr-98
4801 ----------------------
4802
4803 1. A negated single character class followed by a quantifier with a minimum
4804 value of one (e.g. [^x]{1,6} ) was not compiled correctly. This could lead to
4805 program crashes, or just wrong answers. This did not apply to negated classes
4806 containing more than one character, or to minima other than one.
4807
4808
4809 Version 1.08 27-Mar-98
4810 ----------------------
4811
4812 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
4813
4814 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
4815 latter must appear before anything that relies on it in the pattern.
4816
4817
4818 Version 1.07 16-Feb-98
4819 ----------------------
4820
4821 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
4822 repeat of a potentially empty string).
4823
4824
4825 Version 1.06 23-Jan-98
4826 ----------------------
4827
4828 1. Added Markus Oberhumer's little patches for C++.
4829
4830 2. Literal strings longer than 255 characters were broken.
4831
4832
4833 Version 1.05 23-Dec-97
4834 ----------------------
4835
4836 1. Negated character classes containing more than one character were failing if
4837 PCRE_CASELESS was set at run time.
4838
4839
4840 Version 1.04 19-Dec-97
4841 ----------------------
4842
4843 1. Corrected the man page, where some "const" qualifiers had been omitted.
4844
4845 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
4846 input syntax.
4847
4848 3. Fixed memory leak which occurred when a regex with back references was
4849 matched with an offsets vector that wasn't big enough. The temporary memory
4850 that is used in this case wasn't being freed if the match failed.
4851
4852 4. Tidied pcretest to ensure it frees memory that it gets.
4853
4854 5. Temporary memory was being obtained in the case where the passed offsets
4855 vector was exactly big enough.
4856
4857 6. Corrected definition of offsetof() from change 5 below.
4858
4859 7. I had screwed up change 6 below and broken the rules for the use of
4860 setjmp(). Now fixed.
4861
4862
4863 Version 1.03 18-Dec-97
4864 ----------------------
4865
4866 1. A erroneous regex with a missing opening parenthesis was correctly
4867 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
4868 on some systems.
4869
4870 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
4871 it was reported that one broken compiler failed on the former because "code" is
4872 also an independent variable.
4873
4874 3. The erroneous regex a[]b caused an array overrun reference.
4875
4876 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
4877 fail on data ending with that character. (It was going on too far, and checking
4878 the next character, typically a binary zero.) This was specific to the
4879 optimized code for single-character negative classes.
4880
4881 5. Added a contributed patch from the TIN world which does the following:
4882
4883 + Add an undef for memmove, in case the the system defines a macro for it.
4884
4885 + Add a definition of offsetof(), in case there isn't one. (I don't know
4886 the reason behind this - offsetof() is part of the ANSI standard - but
4887 it does no harm).
4888
4889 + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
4890 most of the places where whitespace preceded '#'. I have given up and
4891 allowed the remaining 2 cases to be at the margin.
4892
4893 + Rename some variables in pcre to eliminate shadowing. This seems very
4894 pedantic, but does no harm, of course.
4895
4896 6. Moved the call to setjmp() into its own function, to get rid of warnings
4897 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
4898
4899 7. Constructs such as \d{8,} were compiling into the equivalent of
4900 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
4901 outcome, but in this particular case used more store than had been allocated,
4902 which caused the bug to be discovered because it threw up an internal error.
4903
4904 8. The debugging code in both pcre and pcretest for outputting the compiled
4905 form of a regex was going wrong in the case of back references followed by
4906 curly-bracketed repeats.
4907
4908
4909 Version 1.02 12-Dec-97
4910 ----------------------
4911
4912 1. Typos in pcre.3 and comments in the source fixed.
4913
4914 2. Applied a contributed patch to get rid of places where it used to remove
4915 'const' from variables, and fixed some signed/unsigned and uninitialized
4916 variable warnings.
4917
4918 3. Added the "runtest" target to Makefile.
4919
4920 4. Set default compiler flag to -O2 rather than just -O.
4921
4922
4923 Version 1.01 19-Nov-97
4924 ----------------------
4925
4926 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
4927 like /([ab]*)*/, that is, for classes with more than one character in them.
4928
4929 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
4930 as /((?>a*))*/ (a PCRE_EXTRA facility).
4931
4932
4933 Version 1.00 18-Nov-97
4934 ----------------------
4935
4936 1. Added compile-time macros to support systems such as SunOS4 which don't have
4937 memmove() or strerror() but have other things that can be used instead.
4938
4939 2. Arranged that "make clean" removes the executables.
4940
4941
4942 Version 0.99 27-Oct-97
4943 ----------------------
4944
4945 1. Fixed bug in code for optimizing classes with only one character. It was
4946 initializing a 32-byte map regardless, which could cause it to run off the end
4947 of the memory it had got.
4948
4949 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
4950
4951
4952 Version 0.98 22-Oct-97
4953 ----------------------
4954
4955 1. Fixed bug in code for handling temporary memory usage when there are more
4956 back references than supplied space in the ovector. This could cause segfaults.
4957
4958
4959 Version 0.97 21-Oct-97
4960 ----------------------
4961
4962 1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
4963
4964 2. Optimized negated single characters not to use a bit map.
4965
4966 3. Brought error texts together as macro definitions; clarified some of them;
4967 fixed one that was wrong - it said "range out of order" when it meant "invalid
4968 escape sequence".
4969
4970 4. Changed some char * arguments to const char *.
4971
4972 5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
4973
4974 6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
4975 pcretest.
4976
4977
4978 Version 0.96 16-Oct-97
4979 ----------------------
4980
4981 1. Added a simple "pgrep" utility to the distribution.
4982
4983 2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
4984 unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
4985 where "ddd" means "one or more decimal digits".
4986
4987 3. Fixed serious bug. If a pattern had a back reference, but the call to
4988 pcre_exec() didn't supply a large enough ovector to record the related
4989 identifying subpattern, the match always failed. PCRE now remembers the number
4990 of the largest back reference, and gets some temporary memory in which to save
4991 the offsets during matching if necessary, in order to ensure that
4992 backreferences always work.
4993
4994 4. Increased the compatibility with Perl in a number of ways:
4995
4996 (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
4997 to request this handling. The option can be set at compile or exec time.
4998
4999 (b) $ matches before a terminating newline by default; an option
5000 PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
5001 mode). The option can be set at compile or exec time.
5002
5003 (c) The handling of \ followed by a digit other than 0 is now supposed to be
5004 the same as Perl's. If the decimal number it represents is less than 10
5005 or there aren't that many previous left capturing parentheses, an octal
5006 escape is read. Inside a character class, it's always an octal escape,
5007 even if it is a single digit.
5008
5009 (d) An escaped but undefined alphabetic character is taken as a literal,
5010 unless PCRE_EXTRA is set. Currently this just reserves the remaining
5011 escapes.
5012
5013 (e) {0} is now permitted. (The previous item is removed from the compiled
5014 pattern).
5015
5016 5. Changed all the names of code files so that the basic parts are no longer
5017 than 10 characters, and abolished the teeny "globals.c" file.
5018
5019 6. Changed the handling of character classes; they are now done with a 32-byte
5020 bit map always.
5021
5022 7. Added the -d and /D options to pcretest to make it possible to look at the
5023 internals of compilation without having to recompile pcre.
5024
5025
5026 Version 0.95 23-Sep-97
5027 ----------------------
5028
5029 1. Fixed bug in pre-pass concerning escaped "normal" characters such as \x5c or
5030 \x20 at the start of a run of normal characters. These were being treated as
5031 real characters, instead of the source characters being re-checked.
5032
5033
5034 Version 0.94 18-Sep-97
5035 ----------------------
5036
5037 1. The functions are now thread-safe, with the caveat that the global variables
5038 containing pointers to malloc() and free() or alternative functions are the
5039 same for all threads.
5040
5041 2. Get pcre_study() to generate a bitmap of initial characters for non-
5042 anchored patterns when this is possible, and use it if passed to pcre_exec().
5043
5044
5045 Version 0.93 15-Sep-97
5046 ----------------------
5047
5048 1. /(b)|(:+)/ was computing an incorrect first character.
5049
5050 2. Add pcre_study() to the API and the passing of pcre_extra to pcre_exec(),
5051 but not actually doing anything yet.
5052
5053 3. Treat "-" characters in classes that cannot be part of ranges as literals,
5054 as Perl does (e.g. [-az] or [az-]).
5055
5056 4. Set the anchored flag if a branch starts with .* or .*? because that tests
5057 all possible positions.
5058
5059 5. Split up into different modules to avoid including unneeded functions in a
5060 compiled binary. However, compile and exec are still in one module. The "study"
5061 function is split off.
5062
5063 6. The character tables are now in a separate module whose source is generated
5064 by an auxiliary program - but can then be edited by hand if required. There are
5065 now no calls to isalnum(), isspace(), isdigit(), isxdigit(), tolower() or
5066 toupper() in the code.
5067
5068 7. Turn the malloc/free funtions variables into pcre_malloc and pcre_free and
5069 make them global. Abolish the function for setting them, as the caller can now
5070 set them directly.
5071
5072
5073 Version 0.92 11-Sep-97
5074 ----------------------
5075
5076 1. A repeat with a fixed maximum and a minimum of 1 for an ordinary character
5077 (e.g. /a{1,3}/) was broken (I mis-optimized it).
5078
5079 2. Caseless matching was not working in character classes if the characters in
5080 the pattern were in upper case.
5081
5082 3. Make ranges like [W-c] work in the same way as Perl for caseless matching.
5083
5084 4. Make PCRE_ANCHORED public and accept as a compile option.
5085
5086 5. Add an options word to pcre_exec() and accept PCRE_ANCHORED and
5087 PCRE_CASELESS at run time. Add escapes \A and \I to pcretest to cause it to
5088 pass them.
5089
5090 6. Give an error if bad option bits passed at compile or run time.
5091
5092 7. Add PCRE_MULTILINE at compile and exec time, and (?m) as well. Add \M to
5093 pcretest to cause it to pass that flag.
5094
5095 8. Add pcre_info(), to get the number of identifying subpatterns, the stored
5096 options, and the first character, if set.
5097
5098 9. Recognize C+ or C{n,m} where n >= 1 as providing a fixed starting character.
5099
5100
5101 Version 0.91 10-Sep-97
5102 ----------------------
5103
5104 1. PCRE was failing to diagnose unlimited repeats of subpatterns that could
5105 match the empty string as in /(a*)*/. It was looping and ultimately crashing.
5106
5107 2. PCRE was looping on encountering an indefinitely repeated back reference to
5108 a subpattern that had matched an empty string, e.g. /(a|)\1*/. It now does what
5109 Perl does - treats the match as successful.
5110
5111 ****

Properties

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

  ViewVC Help
Powered by ViewVC 1.1.5