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

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1299 - (show annotations)
Fri Mar 22 17:23:09 2013 UTC (6 years, 6 months ago) by ph10
File size: 245575 byte(s)
Move some tests from testinput2 to testinput1 because they are now 
Perl-compatible.
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. Experimental support of (*SKIP) backtracking verb in the JIT compiler.
108
109 29. Experimental support of (*THEN) backtracking verb in 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 to 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, except for (*ACCEPT). Again, this sometimes improves
122 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
129 Version 8.32 30-November-2012
130 -----------------------------
131
132 1. Improved JIT compiler optimizations for first character search and single
133 character iterators.
134
135 2. Supporting IBM XL C compilers for PPC architectures in the JIT compiler.
136 Patch by Daniel Richard G.
137
138 3. Single character iterator optimizations in the JIT compiler.
139
140 4. Improved JIT compiler optimizations for character ranges.
141
142 5. Rename the "leave" variable names to "quit" to improve WinCE compatibility.
143 Reported by Giuseppe D'Angelo.
144
145 6. The PCRE_STARTLINE bit, indicating that a match can occur only at the start
146 of a line, was being set incorrectly in cases where .* appeared inside
147 atomic brackets at the start of a pattern, or where there was a subsequent
148 *PRUNE or *SKIP.
149
150 7. Improved instruction cache flush for POWER/PowerPC.
151 Patch by Daniel Richard G.
152
153 8. Fixed a number of issues in pcregrep, making it more compatible with GNU
154 grep:
155
156 (a) There is now no limit to the number of patterns to be matched.
157
158 (b) An error is given if a pattern is too long.
159
160 (c) Multiple uses of --exclude, --exclude-dir, --include, and --include-dir
161 are now supported.
162
163 (d) --exclude-from and --include-from (multiple use) have been added.
164
165 (e) Exclusions and inclusions now apply to all files and directories, not
166 just to those obtained from scanning a directory recursively.
167
168 (f) Multiple uses of -f and --file-list are now supported.
169
170 (g) In a Windows environment, the default for -d has been changed from
171 "read" (the GNU grep default) to "skip", because otherwise the presence
172 of a directory in the file list provokes an error.
173
174 (h) The documentation has been revised and clarified in places.
175
176 9. Improve the matching speed of capturing brackets.
177
178 10. Changed the meaning of \X so that it now matches a Unicode extended
179 grapheme cluster.
180
181 11. Patch by Daniel Richard G to the autoconf files to add a macro for sorting
182 out POSIX threads when JIT support is configured.
183
184 12. Added support for PCRE_STUDY_EXTRA_NEEDED.
185
186 13. In the POSIX wrapper regcomp() function, setting re_nsub field in the preg
187 structure could go wrong in environments where size_t is not the same size
188 as int.
189
190 14. Applied user-supplied patch to pcrecpp.cc to allow PCRE_NO_UTF8_CHECK to be
191 set.
192
193 15. The EBCDIC support had decayed; later updates to the code had included
194 explicit references to (e.g.) \x0a instead of CHAR_LF. There has been a
195 general tidy up of EBCDIC-related issues, and the documentation was also
196 not quite right. There is now a test that can be run on ASCII systems to
197 check some of the EBCDIC-related things (but is it not a full test).
198
199 16. The new PCRE_STUDY_EXTRA_NEEDED option is now used by pcregrep, resulting
200 in a small tidy to the code.
201
202 17. Fix JIT tests when UTF is disabled and both 8 and 16 bit mode are enabled.
203
204 18. If the --only-matching (-o) option in pcregrep is specified multiple
205 times, each one causes appropriate output. For example, -o1 -o2 outputs the
206 substrings matched by the 1st and 2nd capturing parentheses. A separating
207 string can be specified by --om-separator (default empty).
208
209 19. Improving the first n character searches.
210
211 20. Turn case lists for horizontal and vertical white space into macros so that
212 they are defined only once.
213
214 21. This set of changes together give more compatible Unicode case-folding
215 behaviour for characters that have more than one other case when UCP
216 support is available.
217
218 (a) The Unicode property table now has offsets into a new table of sets of
219 three or more characters that are case-equivalent. The MultiStage2.py
220 script that generates these tables (the pcre_ucd.c file) now scans
221 CaseFolding.txt instead of UnicodeData.txt for character case
222 information.
223
224 (b) The code for adding characters or ranges of characters to a character
225 class has been abstracted into a generalized function that also handles
226 case-independence. In UTF-mode with UCP support, this uses the new data
227 to handle characters with more than one other case.
228
229 (c) A bug that is fixed as a result of (b) is that codepoints less than 256
230 whose other case is greater than 256 are now correctly matched
231 caselessly. Previously, the high codepoint matched the low one, but not
232 vice versa.
233
234 (d) The processing of \h, \H, \v, and \ in character classes now makes use
235 of the new class addition function, using character lists defined as
236 macros alongside the case definitions of 20 above.
237
238 (e) Caseless back references now work with characters that have more than
239 one other case.
240
241 (f) General caseless matching of characters with more than one other case
242 is supported.
243
244 22. Unicode character properties were updated from Unicode 6.2.0
245
246 23. Improved CMake support under Windows. Patch by Daniel Richard G.
247
248 24. Add support for 32-bit character strings, and UTF-32
249
250 25. Major JIT compiler update (code refactoring and bugfixing).
251 Experimental Sparc 32 support is added.
252
253 26. Applied a modified version of Daniel Richard G's patch to create
254 pcre.h.generic and config.h.generic by "make" instead of in the
255 PrepareRelease script.
256
257 27. Added a definition for CHAR_NULL (helpful for the z/OS port), and use it in
258 pcre_compile.c when checking for a zero character.
259
260 28. Introducing a native interface for JIT. Through this interface, the compiled
261 machine code can be directly executed. The purpose of this interface is to
262 provide fast pattern matching, so several sanity checks are not performed.
263 However, feature tests are still performed. The new interface provides
264 1.4x speedup compared to the old one.
265
266 29. If pcre_exec() or pcre_dfa_exec() was called with a negative value for
267 the subject string length, the error given was PCRE_ERROR_BADOFFSET, which
268 was confusing. There is now a new error PCRE_ERROR_BADLENGTH for this case.
269
270 30. In 8-bit UTF-8 mode, pcretest failed to give an error for data codepoints
271 greater than 0x7fffffff (which cannot be represented in UTF-8, even under
272 the "old" RFC 2279). Instead, it ended up passing a negative length to
273 pcre_exec().
274
275 31. Add support for GCC's visibility feature to hide internal functions.
276
277 32. Running "pcretest -C pcre8" or "pcretest -C pcre16" gave a spurious error
278 "unknown -C option" after outputting 0 or 1.
279
280 33. There is now support for generating a code coverage report for the test
281 suite in environments where gcc is the compiler and lcov is installed. This
282 is mainly for the benefit of the developers.
283
284 34. If PCRE is built with --enable-valgrind, certain memory regions are marked
285 unaddressable using valgrind annotations, allowing valgrind to detect
286 invalid memory accesses. This is mainly for the benefit of the developers.
287
288 25. (*UTF) can now be used to start a pattern in any of the three libraries.
289
290 26. Give configure error if --enable-cpp but no C++ compiler found.
291
292
293 Version 8.31 06-July-2012
294 -------------------------
295
296 1. Fixing a wrong JIT test case and some compiler warnings.
297
298 2. Removed a bashism from the RunTest script.
299
300 3. Add a cast to pcre_exec.c to fix the warning "unary minus operator applied
301 to unsigned type, result still unsigned" that was given by an MS compiler
302 on encountering the code "-sizeof(xxx)".
303
304 4. Partial matching support is added to the JIT compiler.
305
306 5. Fixed several bugs concerned with partial matching of items that consist
307 of more than one character:
308
309 (a) /^(..)\1/ did not partially match "aba" because checking references was
310 done on an "all or nothing" basis. This also applied to repeated
311 references.
312
313 (b) \R did not give a hard partial match if \r was found at the end of the
314 subject.
315
316 (c) \X did not give a hard partial match after matching one or more
317 characters at the end of the subject.
318
319 (d) When newline was set to CRLF, a pattern such as /a$/ did not recognize
320 a partial match for the string "\r".
321
322 (e) When newline was set to CRLF, the metacharacter "." did not recognize
323 a partial match for a CR character at the end of the subject string.
324
325 6. If JIT is requested using /S++ or -s++ (instead of just /S+ or -s+) when
326 running pcretest, the text "(JIT)" added to the output whenever JIT is
327 actually used to run the match.
328
329 7. Individual JIT compile options can be set in pcretest by following -s+[+]
330 or /S+[+] with a digit between 1 and 7.
331
332 8. OP_NOT now supports any UTF character not just single-byte ones.
333
334 9. (*MARK) control verb is now supported by the JIT compiler.
335
336 10. The command "./RunTest list" lists the available tests without actually
337 running any of them. (Because I keep forgetting what they all are.)
338
339 11. Add PCRE_INFO_MAXLOOKBEHIND.
340
341 12. Applied a (slightly modified) user-supplied patch that improves performance
342 when the heap is used for recursion (compiled with --disable-stack-for-
343 recursion). Instead of malloc and free for each heap frame each time a
344 logical recursion happens, frames are retained on a chain and re-used where
345 possible. This sometimes gives as much as 30% improvement.
346
347 13. As documented, (*COMMIT) is now confined to within a recursive subpattern
348 call.
349
350 14. As documented, (*COMMIT) is now confined to within a positive assertion.
351
352 15. It is now possible to link pcretest with libedit as an alternative to
353 libreadline.
354
355 16. (*COMMIT) control verb is now supported by the JIT compiler.
356
357 17. The Unicode data tables have been updated to Unicode 6.1.0.
358
359 18. Added --file-list option to pcregrep.
360
361 19. Added binary file support to pcregrep, including the -a, --binary-files,
362 -I, and --text options.
363
364 20. The madvise function is renamed for posix_madvise for QNX compatibility
365 reasons. Fixed by Giuseppe D'Angelo.
366
367 21. Fixed a bug for backward assertions with REVERSE 0 in the JIT compiler.
368
369 22. Changed the option for creating symbolic links for 16-bit man pages from
370 -s to -sf so that re-installing does not cause issues.
371
372 23. Support PCRE_NO_START_OPTIMIZE in JIT as (*MARK) support requires it.
373
374 24. Fixed a very old bug in pcretest that caused errors with restarted DFA
375 matches in certain environments (the workspace was not being correctly
376 retained). Also added to pcre_dfa_exec() a simple plausibility check on
377 some of the workspace data at the beginning of a restart.
378
379 25. \s*\R was auto-possessifying the \s* when it should not, whereas \S*\R
380 was not doing so when it should - probably a typo introduced by SVN 528
381 (change 8.10/14).
382
383 26. When PCRE_UCP was not set, \w+\x{c4} was incorrectly auto-possessifying the
384 \w+ when the character tables indicated that \x{c4} was a word character.
385 There were several related cases, all because the tests for doing a table
386 lookup were testing for characters less than 127 instead of 255.
387
388 27. If a pattern contains capturing parentheses that are not used in a match,
389 their slots in the ovector are set to -1. For those that are higher than
390 any matched groups, this happens at the end of processing. In the case when
391 there were back references that the ovector was too small to contain
392 (causing temporary malloc'd memory to be used during matching), and the
393 highest capturing number was not used, memory off the end of the ovector
394 was incorrectly being set to -1. (It was using the size of the temporary
395 memory instead of the true size.)
396
397 28. To catch bugs like 27 using valgrind, when pcretest is asked to specify an
398 ovector size, it uses memory at the end of the block that it has got.
399
400 29. Check for an overlong MARK name and give an error at compile time. The
401 limit is 255 for the 8-bit library and 65535 for the 16-bit library.
402
403 30. JIT compiler update.
404
405 31. JIT is now supported on jailbroken iOS devices. Thanks for Ruiger
406 Rill for the patch.
407
408 32. Put spaces around SLJIT_PRINT_D in the JIT compiler. Required by CXX11.
409
410 33. Variable renamings in the PCRE-JIT compiler. No functionality change.
411
412 34. Fixed typos in pcregrep: in two places there was SUPPORT_LIBZ2 instead of
413 SUPPORT_LIBBZ2. This caused a build problem when bzip2 but not gzip (zlib)
414 was enabled.
415
416 35. Improve JIT code generation for greedy plus quantifier.
417
418 36. When /((?:a?)*)*c/ or /((?>a?)*)*c/ was matched against "aac", it set group
419 1 to "aa" instead of to an empty string. The bug affected repeated groups
420 that could potentially match an empty string.
421
422 37. Optimizing single character iterators in JIT.
423
424 38. Wide characters specified with \uxxxx in JavaScript mode are now subject to
425 the same checks as \x{...} characters in non-JavaScript mode. Specifically,
426 codepoints that are too big for the mode are faulted, and in a UTF mode,
427 disallowed codepoints are also faulted.
428
429 39. If PCRE was compiled with UTF support, in three places in the DFA
430 matcher there was code that should only have been obeyed in UTF mode, but
431 was being obeyed unconditionally. In 8-bit mode this could cause incorrect
432 processing when bytes with values greater than 127 were present. In 16-bit
433 mode the bug would be provoked by values in the range 0xfc00 to 0xdc00. In
434 both cases the values are those that cannot be the first data item in a UTF
435 character. The three items that might have provoked this were recursions,
436 possessively repeated groups, and atomic groups.
437
438 40. Ensure that libpcre is explicitly listed in the link commands for pcretest
439 and pcregrep, because some OS require shared objects to be explicitly
440 passed to ld, causing the link step to fail if they are not.
441
442 41. There were two incorrect #ifdefs in pcre_study.c, meaning that, in 16-bit
443 mode, patterns that started with \h* or \R* might be incorrectly matched.
444
445
446 Version 8.30 04-February-2012
447 -----------------------------
448
449 1. Renamed "isnumber" as "is_a_number" because in some Mac environments this
450 name is defined in ctype.h.
451
452 2. Fixed a bug in fixed-length calculation for lookbehinds that would show up
453 only in quite long subpatterns.
454
455 3. Removed the function pcre_info(), which has been obsolete and deprecated
456 since it was replaced by pcre_fullinfo() in February 2000.
457
458 4. For a non-anchored pattern, if (*SKIP) was given with a name that did not
459 match a (*MARK), and the match failed at the start of the subject, a
460 reference to memory before the start of the subject could occur. This bug
461 was introduced by fix 17 of release 8.21.
462
463 5. A reference to an unset group with zero minimum repetition was giving
464 totally wrong answers (in non-JavaScript-compatibility mode). For example,
465 /(another)?(\1?)test/ matched against "hello world test". This bug was
466 introduced in release 8.13.
467
468 6. Add support for 16-bit character strings (a large amount of work involving
469 many changes and refactorings).
470
471 7. RunGrepTest failed on msys because \r\n was replaced by whitespace when the
472 command "pattern=`printf 'xxx\r\njkl'`" was run. The pattern is now taken
473 from a file.
474
475 8. Ovector size of 2 is also supported by JIT based pcre_exec (the ovector size
476 rounding is not applied in this particular case).
477
478 9. The invalid Unicode surrogate codepoints U+D800 to U+DFFF are now rejected
479 if they appear, or are escaped, in patterns.
480
481 10. Get rid of a number of -Wunused-but-set-variable warnings.
482
483 11. The pattern /(?=(*:x))(q|)/ matches an empty string, and returns the mark
484 "x". The similar pattern /(?=(*:x))((*:y)q|)/ did not return a mark at all.
485 Oddly, Perl behaves the same way. PCRE has been fixed so that this pattern
486 also returns the mark "x". This bug applied to capturing parentheses,
487 non-capturing parentheses, and atomic parentheses. It also applied to some
488 assertions.
489
490 12. Stephen Kelly's patch to CMakeLists.txt allows it to parse the version
491 information out of configure.ac instead of relying on pcre.h.generic, which
492 is not stored in the repository.
493
494 13. Applied Dmitry V. Levin's patch for a more portable method for linking with
495 -lreadline.
496
497 14. ZH added PCRE_CONFIG_JITTARGET; added its output to pcretest -C.
498
499 15. Applied Graycode's patch to put the top-level frame on the stack rather
500 than the heap when not using the stack for recursion. This gives a
501 performance improvement in many cases when recursion is not deep.
502
503 16. Experimental code added to "pcretest -C" to output the stack frame size.
504
505
506 Version 8.21 12-Dec-2011
507 ------------------------
508
509 1. Updating the JIT compiler.
510
511 2. JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
512 are added as well.
513
514 3. Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
515 PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
516 calling _pcre_jit_exec. Some extra comments are added.
517
518 4. (*MARK) settings inside atomic groups that do not contain any capturing
519 parentheses, for example, (?>a(*:m)), were not being passed out. This bug
520 was introduced by change 18 for 8.20.
521
522 5. Supporting of \x, \U and \u in JavaScript compatibility mode based on the
523 ECMA-262 standard.
524
525 6. Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
526 erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
527 This bug was probably introduced by change 9 of 8.13.
528
529 7. While fixing 6 above, I noticed that a number of other items were being
530 incorrectly rejected as "not fixed length". This arose partly because newer
531 opcodes had not been added to the fixed-length checking code. I have (a)
532 corrected the bug and added tests for these items, and (b) arranged for an
533 error to occur if an unknown opcode is encountered while checking for fixed
534 length instead of just assuming "not fixed length". The items that were
535 rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
536 (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
537 repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
538
539 8. A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
540 being incorrectly compiled and would have given unpredicatble results.
541
542 9. A possessively repeated subpattern with minimum repeat count greater than
543 one behaved incorrectly. For example, (A){2,}+ behaved as if it was
544 (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
545 the first (A) could occur when it should not.
546
547 10. Add a cast and remove a redundant test from the code.
548
549 11. JIT should use pcre_malloc/pcre_free for allocation.
550
551 12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
552 best practice nowadays, and helps with cross-compiling. (If the exec_prefix
553 is anything other than /usr, -L is still shown).
554
555 13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
556
557 14. Perl does not support \N without a following name in a [] class; PCRE now
558 also gives an error.
559
560 15. If a forward reference was repeated with an upper limit of around 2000,
561 it caused the error "internal error: overran compiling workspace". The
562 maximum number of forward references (including repeats) was limited by the
563 internal workspace, and dependent on the LINK_SIZE. The code has been
564 rewritten so that the workspace expands (via pcre_malloc) if necessary, and
565 the default depends on LINK_SIZE. There is a new upper limit (for safety)
566 of around 200,000 forward references. While doing this, I also speeded up
567 the filling in of repeated forward references.
568
569 16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
570 incorrectly expecting the subject to contain another "a" after the start.
571
572 17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
573 in the match, the SKIP should be ignored. This was not happening; instead
574 the SKIP was being treated as NOMATCH. For patterns such as
575 /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
576 tested.
577
578 18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
579 now much more compatible with Perl, in particular in cases where the result
580 is a non-match for a non-anchored pattern. For example, if
581 /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
582 "m", where previously it did not return a name. A side effect of this
583 change is that for partial matches, the last encountered mark name is
584 returned, as for non matches. A number of tests that were previously not
585 Perl-compatible have been moved into the Perl-compatible test files. The
586 refactoring has had the pleasing side effect of removing one argument from
587 the match() function, thus reducing its stack requirements.
588
589 19. If the /S+ option was used in pcretest to study a pattern using JIT,
590 subsequent uses of /S (without +) incorrectly behaved like /S+.
591
592 21. Retrieve executable code size support for the JIT compiler and fixing
593 some warnings.
594
595 22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
596 not work when the shorter character appeared right at the end of the
597 subject string.
598
599 23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
600 systems.
601
602 24. Added PCRE_INFO_JITSIZE to pass on the value from (21) above, and also
603 output it when the /M option is used in pcretest.
604
605 25. The CheckMan script was not being included in the distribution. Also, added
606 an explicit "perl" to run Perl scripts from the PrepareRelease script
607 because this is reportedly needed in Windows.
608
609 26. If study data was being save in a file and studying had not found a set of
610 "starts with" bytes for the pattern, the data written to the file (though
611 never used) was taken from uninitialized memory and so caused valgrind to
612 complain.
613
614 27. Updated RunTest.bat as provided by Sheri Pierce.
615
616 28. Fixed a possible uninitialized memory bug in pcre_jit_compile.c.
617
618 29. Computation of memory usage for the table of capturing group names was
619 giving an unnecessarily large value.
620
621
622 Version 8.20 21-Oct-2011
623 ------------------------
624
625 1. Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
626 a POSIX class. After further experiments with Perl, which convinced me that
627 Perl has bugs and confusions, a closing square bracket is no longer allowed
628 in a POSIX name. This bug also affected patterns with classes that started
629 with full stops.
630
631 2. If a pattern such as /(a)b|ac/ is matched against "ac", there is no
632 captured substring, but while checking the failing first alternative,
633 substring 1 is temporarily captured. If the output vector supplied to
634 pcre_exec() was not big enough for this capture, the yield of the function
635 was still zero ("insufficient space for captured substrings"). This cannot
636 be totally fixed without adding another stack variable, which seems a lot
637 of expense for a edge case. However, I have improved the situation in cases
638 such as /(a)(b)x|abc/ matched against "abc", where the return code
639 indicates that fewer than the maximum number of slots in the ovector have
640 been set.
641
642 3. Related to (2) above: when there are more back references in a pattern than
643 slots in the output vector, pcre_exec() uses temporary memory during
644 matching, and copies in the captures as far as possible afterwards. It was
645 using the entire output vector, but this conflicts with the specification
646 that only 2/3 is used for passing back captured substrings. Now it uses
647 only the first 2/3, for compatibility. This is, of course, another edge
648 case.
649
650 4. Zoltan Herczeg's just-in-time compiler support has been integrated into the
651 main code base, and can be used by building with --enable-jit. When this is
652 done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
653 runtime --no-jit option is given.
654
655 5. When the number of matches in a pcre_dfa_exec() run exactly filled the
656 ovector, the return from the function was zero, implying that there were
657 other matches that did not fit. The correct "exactly full" value is now
658 returned.
659
660 6. If a subpattern that was called recursively or as a subroutine contained
661 (*PRUNE) or any other control that caused it to give a non-standard return,
662 invalid errors such as "Error -26 (nested recursion at the same subject
663 position)" or even infinite loops could occur.
664
665 7. If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
666 computing the minimum length on reaching *ACCEPT, and so ended up with the
667 wrong value of 1 rather than 0. Further investigation indicates that
668 computing a minimum subject length in the presence of *ACCEPT is difficult
669 (think back references, subroutine calls), and so I have changed the code
670 so that no minimum is registered for a pattern that contains *ACCEPT.
671
672 8. If (*THEN) was present in the first (true) branch of a conditional group,
673 it was not handled as intended. [But see 16 below.]
674
675 9. Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
676 Sheri Pierce.
677
678 10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
679 the first byte in a match must be "a".
680
681 11. Change 17 for 8.13 increased the recursion depth for patterns like
682 /a(?:.)*?a/ drastically. I've improved things by remembering whether a
683 pattern contains any instances of (*THEN). If it does not, the old
684 optimizations are restored. It would be nice to do this on a per-group
685 basis, but at the moment that is not feasible.
686
687 12. In some environments, the output of pcretest -C is CRLF terminated. This
688 broke RunTest's code that checks for the link size. A single white space
689 character after the value is now allowed for.
690
691 13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
692 For "fr", it uses the Windows-specific input and output files.
693
694 14. If (*THEN) appeared in a group that was called recursively or as a
695 subroutine, it did not work as intended. [But see next item.]
696
697 15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
698 pattern fragments (but not containing any | characters). If A and B are
699 matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
700 was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
701 D. In other words, Perl considers parentheses that do not contain any |
702 characters to be part of a surrounding alternative, whereas PCRE was
703 treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
704 differently. PCRE now behaves in the same way as Perl, except in the case
705 of subroutine/recursion calls such as (?1) which have in any case always
706 been different (but PCRE had them first :-).
707
708 16. Related to 15 above: Perl does not treat the | in a conditional group as
709 creating alternatives. Such a group is treated in the same way as an
710 ordinary group without any | characters when processing (*THEN). PCRE has
711 been changed to match Perl's behaviour.
712
713 17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
714 RunGrepTest script failed.
715
716 18. Change 22 for version 13 caused atomic groups to use more stack. This is
717 inevitable for groups that contain captures, but it can lead to a lot of
718 stack use in large patterns. The old behaviour has been restored for atomic
719 groups that do not contain any capturing parentheses.
720
721 19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
722 suppress the check for a minimum subject length at run time. (If it was
723 given to pcre_exec() or pcre_dfa_exec() it did work.)
724
725 20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
726 fail to work when decoding hex characters in data strings in EBCDIC
727 environments.
728
729 21. It appears that in at least one Mac OS environment, the isxdigit() function
730 is implemented as a macro that evaluates to its argument more than once,
731 contravening the C 90 Standard (I haven't checked a later standard). There
732 was an instance in pcretest which caused it to go wrong when processing
733 \x{...} escapes in subject strings. The has been rewritten to avoid using
734 things like p++ in the argument of isxdigit().
735
736
737 Version 8.13 16-Aug-2011
738 ------------------------
739
740 1. The Unicode data tables have been updated to Unicode 6.0.0.
741
742 2. Two minor typos in pcre_internal.h have been fixed.
743
744 3. Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
745 pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
746 in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
747
748 4. There were a number of related bugs in the code for matching backrefences
749 caselessly in UTF-8 mode when codes for the characters concerned were
750 different numbers of bytes. For example, U+023A and U+2C65 are an upper
751 and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
752 (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
753 code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
754 2-byte code at the end of the subject (it thought there wasn't enough data
755 left).
756
757 5. Comprehensive information about what went wrong is now returned by
758 pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
759 as the output vector has at least 2 elements. The offset of the start of
760 the failing character and a reason code are placed in the vector.
761
762 6. When the UTF-8 string check fails for pcre_compile(), the offset that is
763 now returned is for the first byte of the failing character, instead of the
764 last byte inspected. This is an incompatible change, but I hope it is small
765 enough not to be a problem. It makes the returned offset consistent with
766 pcre_exec() and pcre_dfa_exec().
767
768 7. pcretest now gives a text phrase as well as the error number when
769 pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
770 failure, the offset and reason code are output.
771
772 8. When \R was used with a maximizing quantifier it failed to skip backwards
773 over a \r\n pair if the subsequent match failed. Instead, it just skipped
774 back over a single character (\n). This seems wrong (because it treated the
775 two characters as a single entity when going forwards), conflicts with the
776 documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
777 behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
778 has been changed.
779
780 9. Some internal refactoring has changed the processing so that the handling
781 of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
782 time (the PCRE_DOTALL option was changed this way some time ago: version
783 7.7 change 16). This has made it possible to abolish the OP_OPT op code,
784 which was always a bit of a fudge. It also means that there is one less
785 argument for the match() function, which reduces its stack requirements
786 slightly. This change also fixes an incompatibility with Perl: the pattern
787 (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
788
789 10. More internal refactoring has drastically reduced the number of recursive
790 calls to match() for possessively repeated groups such as (abc)++ when
791 using pcre_exec().
792
793 11. While implementing 10, a number of bugs in the handling of groups were
794 discovered and fixed:
795
796 (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
797 (a|)*(?1) gave a compile-time internal error.
798 ((a|)+)+ did not notice that the outer group could match an empty string.
799 (^a|^)+ was not marked as anchored.
800 (.*a|.*)+ was not marked as matching at start or after a newline.
801
802 12. Yet more internal refactoring has removed another argument from the match()
803 function. Special calls to this function are now indicated by setting a
804 value in a variable in the "match data" data block.
805
806 13. Be more explicit in pcre_study() instead of relying on "default" for
807 opcodes that mean there is no starting character; this means that when new
808 ones are added and accidentally left out of pcre_study(), testing should
809 pick them up.
810
811 14. The -s option of pcretest has been documented for ages as being an old
812 synonym of -m (show memory usage). I have changed it to mean "force study
813 for every regex", that is, assume /S for every regex. This is similar to -i
814 and -d etc. It's slightly incompatible, but I'm hoping nobody is still
815 using it. It makes it easier to run collections of tests with and without
816 study enabled, and thereby test pcre_study() more easily. All the standard
817 tests are now run with and without -s (but some patterns can be marked as
818 "never study" - see 20 below).
819
820 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
821 restoration of the capturing data to the outer values was not happening
822 correctly.
823
824 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
825 empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
826 pattern had matched an empty string, and so incorrectly returned a no
827 match.
828
829 17. There was optimizing code for the last branch of non-capturing parentheses,
830 and also for the obeyed branch of a conditional subexpression, which used
831 tail recursion to cut down on stack usage. Unfortunately, now that there is
832 the possibility of (*THEN) occurring in these branches, tail recursion is
833 no longer possible because the return has to be checked for (*THEN). These
834 two optimizations have therefore been removed. [But see 8.20/11 above.]
835
836 18. If a pattern containing \R was studied, it was assumed that \R always
837 matched two bytes, thus causing the minimum subject length to be
838 incorrectly computed because \R can also match just one byte.
839
840 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
841 was incorrectly computed.
842
843 20. If /S is present twice on a test pattern in pcretest input, it now
844 *disables* studying, thereby overriding the use of -s on the command line
845 (see 14 above). This is necessary for one or two tests to keep the output
846 identical in both cases.
847
848 21. When (*ACCEPT) was used in an assertion that matched an empty string and
849 PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
850
851 22. When an atomic group that contained a capturing parenthesis was
852 successfully matched, but the branch in which it appeared failed, the
853 capturing was not being forgotten if a higher numbered group was later
854 captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
855 group 1 to "a", when in fact it should be unset. This applied to multi-
856 branched capturing and non-capturing groups, repeated or not, and also to
857 positive assertions (capturing in negative assertions does not happen
858 in PCRE) and also to nested atomic groups.
859
860 23. Add the ++ qualifier feature to pcretest, to show the remainder of the
861 subject after a captured substring, to make it easier to tell which of a
862 number of identical substrings has been captured.
863
864 24. The way atomic groups are processed by pcre_exec() has been changed so that
865 if they are repeated, backtracking one repetition now resets captured
866 values correctly. For example, if ((?>(a+)b)+aabab) is matched against
867 "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
868 "aaa". Previously, it would have been "a". As part of this code
869 refactoring, the way recursive calls are handled has also been changed.
870
871 25. If an assertion condition captured any substrings, they were not passed
872 back unless some other capturing happened later. For example, if
873 (?(?=(a))a) was matched against "a", no capturing was returned.
874
875 26. When studying a pattern that contained subroutine calls or assertions,
876 the code for finding the minimum length of a possible match was handling
877 direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
878 group 1 called group 2 while simultaneously a separate group 2 called group
879 1). A stack overflow occurred in this case. I have fixed this by limiting
880 the recursion depth to 10.
881
882 27. Updated RunTest.bat in the distribution to the version supplied by Tom
883 Fortmann. This supports explicit test numbers on the command line, and has
884 argument validation and error reporting.
885
886 28. An instance of \X with an unlimited repeat could fail if at any point the
887 first character it looked at was a mark character.
888
889 29. Some minor code refactoring concerning Unicode properties and scripts
890 should reduce the stack requirement of match() slightly.
891
892 30. Added the '=' option to pcretest to check the setting of unused capturing
893 slots at the end of the pattern, which are documented as being -1, but are
894 not included in the return count.
895
896 31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
897 compiled something random. Now it gives a compile-time error (as does
898 Perl).
899
900 32. A *MARK encountered during the processing of a positive assertion is now
901 recorded and passed back (compatible with Perl).
902
903 33. If --only-matching or --colour was set on a pcregrep call whose pattern
904 had alternative anchored branches, the search for a second match in a line
905 was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
906 matched the line "0102" twice. The same bug affected patterns that started
907 with a backwards assertion. For example /\b01|\b02/ also matched "0102"
908 twice.
909
910 34. Previously, PCRE did not allow quantification of assertions. However, Perl
911 does, and because of capturing effects, quantifying parenthesized
912 assertions may at times be useful. Quantifiers are now allowed for
913 parenthesized assertions.
914
915 35. A minor code tidy in pcre_compile() when checking options for \R usage.
916
917 36. \g was being checked for fancy things in a character class, when it should
918 just be a literal "g".
919
920 37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
921 appearance of a nested POSIX class supersedes an apparent external class.
922 For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
923 unescaped square brackets may also appear as part of class names. For
924 example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
925 more like Perl. (But see 8.20/1 above.)
926
927 38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
928 was because it thought it was \N{name}, which is not supported).
929
930 39. Add minix to OS list not supporting the -S option in pcretest.
931
932 40. PCRE tries to detect cases of infinite recursion at compile time, but it
933 cannot analyze patterns in sufficient detail to catch mutual recursions
934 such as ((?1))((?2)). There is now a runtime test that gives an error if a
935 subgroup is called recursively as a subpattern for a second time at the
936 same position in the subject string. In previous releases this might have
937 been caught by the recursion limit, or it might have run out of stack.
938
939 41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
940 happen only once. PCRE was, however incorrectly giving a compile time error
941 "recursive call could loop indefinitely" because it cannot analyze the
942 pattern in sufficient detail. The compile time test no longer happens when
943 PCRE is compiling a conditional subpattern, but actual runaway loops are
944 now caught at runtime (see 40 above).
945
946 42. It seems that Perl allows any characters other than a closing parenthesis
947 to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
948 has been changed to be the same.
949
950 43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
951 as not to get warnings when autogen.sh is called. Also changed
952 AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
953
954 44. To help people who use pcregrep to scan files containing exceedingly long
955 lines, the following changes have been made:
956
957 (a) The default value of the buffer size parameter has been increased from
958 8K to 20K. (The actual buffer used is three times this size.)
959
960 (b) The default can be changed by ./configure --with-pcregrep-bufsize when
961 PCRE is built.
962
963 (c) A --buffer-size=n option has been added to pcregrep, to allow the size
964 to be set at run time.
965
966 (d) Numerical values in pcregrep options can be followed by K or M, for
967 example --buffer-size=50K.
968
969 (e) If a line being scanned overflows pcregrep's buffer, an error is now
970 given and the return code is set to 2.
971
972 45. Add a pointer to the latest mark to the callout data block.
973
974 46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
975 partial match of an empty string instead of no match. This was specific to
976 the use of ".".
977
978 47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
979 complete match instead of a partial match. This bug was dependent on both
980 the PCRE_UTF8 and PCRE_DOTALL options being set.
981
982 48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
983 starting byte set, because \b was not being ignored.
984
985
986 Version 8.12 15-Jan-2011
987 ------------------------
988
989 1. Fixed some typos in the markup of the man pages, and wrote a script that
990 checks for such things as part of the documentation building process.
991
992 2. On a big-endian 64-bit system, pcregrep did not correctly process the
993 --match-limit and --recursion-limit options (added for 8.11). In
994 particular, this made one of the standard tests fail. (The integer value
995 went into the wrong half of a long int.)
996
997 3. If the --colour option was given to pcregrep with -v (invert match), it
998 did strange things, either producing crazy output, or crashing. It should,
999 of course, ignore a request for colour when reporting lines that do not
1000 match.
1001
1002 4. Another pcregrep bug caused similar problems if --colour was specified with
1003 -M (multiline) and the pattern match finished with a line ending.
1004
1005 5. In pcregrep, when a pattern that ended with a literal newline sequence was
1006 matched in multiline mode, the following line was shown as part of the
1007 match. This seems wrong, so I have changed it.
1008
1009 6. Another pcregrep bug in multiline mode, when --colour was specified, caused
1010 the check for further matches in the same line (so they could be coloured)
1011 to overrun the end of the current line. If another match was found, it was
1012 incorrectly shown (and then shown again when found in the next line).
1013
1014 7. If pcregrep was compiled under Windows, there was a reference to the
1015 function pcregrep_exit() before it was defined. I am assuming this was
1016 the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
1017 reported by a user. I've moved the definition above the reference.
1018
1019
1020 Version 8.11 10-Dec-2010
1021 ------------------------
1022
1023 1. (*THEN) was not working properly if there were untried alternatives prior
1024 to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
1025 backtracked to try for "b" instead of moving to the next alternative branch
1026 at the same level (in this case, to look for "c"). The Perl documentation
1027 is clear that when (*THEN) is backtracked onto, it goes to the "next
1028 alternative in the innermost enclosing group".
1029
1030 2. (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
1031 such as (A(*COMMIT)B(*THEN)C|D) any failure after matching A should
1032 result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
1033 (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
1034 (*THEN).
1035
1036 3. If \s appeared in a character class, it removed the VT character from
1037 the class, even if it had been included by some previous item, for example
1038 in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
1039 of \s, but is part of the POSIX "space" class.)
1040
1041 4. A partial match never returns an empty string (because you can always
1042 match an empty string at the end of the subject); however the checking for
1043 an empty string was starting at the "start of match" point. This has been
1044 changed to the "earliest inspected character" point, because the returned
1045 data for a partial match starts at this character. This means that, for
1046 example, /(?<=abc)def/ gives a partial match for the subject "abc"
1047 (previously it gave "no match").
1048
1049 5. Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
1050 of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
1051 previously a full match would be given. However, setting PCRE_PARTIAL_HARD
1052 has an implication that the given string is incomplete (because a partial
1053 match is preferred over a full match). For this reason, these items now
1054 give a partial match in this situation. [Aside: previously, the one case
1055 /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
1056 match rather than a full match, which was wrong by the old rules, but is
1057 now correct.]
1058
1059 6. There was a bug in the handling of #-introduced comments, recognized when
1060 PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
1061 If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
1062 UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
1063 scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
1064 but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
1065 places in pcre_compile().
1066
1067 7. Related to (6) above, when pcre_compile() was skipping #-introduced
1068 comments when looking ahead for named forward references to subpatterns,
1069 the only newline sequence it recognized was NL. It now handles newlines
1070 according to the set newline convention.
1071
1072 8. SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
1073 former, but used strtoul(), whereas pcretest avoided strtoul() but did not
1074 cater for a lack of strerror(). These oversights have been fixed.
1075
1076 9. Added --match-limit and --recursion-limit to pcregrep.
1077
1078 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
1079
1080 11. When the -o option was used, pcregrep was setting a return code of 1, even
1081 when matches were found, and --line-buffered was not being honoured.
1082
1083 12. Added an optional parentheses number to the -o and --only-matching options
1084 of pcregrep.
1085
1086 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
1087 can match an empty string. The code to do it in pcretest and pcredemo
1088 needed fixing:
1089
1090 (a) When the newline convention was "crlf", pcretest got it wrong, skipping
1091 only one byte after an empty string match just before CRLF (this case
1092 just got forgotten; "any" and "anycrlf" were OK).
1093
1094 (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
1095 mode when an empty string match preceded an ASCII character followed by
1096 a non-ASCII character. (The code for advancing by one character rather
1097 than one byte was nonsense.)
1098
1099 (c) The pcredemo.c sample program did not have any code at all to handle
1100 the cases when CRLF is a valid newline sequence.
1101
1102 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
1103 as a starting offset was within the subject string. There is now a new
1104 error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
1105 negative or greater than the length of the string. In order to test this,
1106 pcretest is extended to allow the setting of negative starting offsets.
1107
1108 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
1109 starting offset points to the beginning of a UTF-8 character was
1110 unnecessarily clumsy. I tidied it up.
1111
1112 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
1113 bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
1114
1115 17. Nobody had reported that the --include_dir option, which was added in
1116 release 7.7 should have been called --include-dir (hyphen, not underscore)
1117 for compatibility with GNU grep. I have changed it to --include-dir, but
1118 left --include_dir as an undocumented synonym, and the same for
1119 --exclude-dir, though that is not available in GNU grep, at least as of
1120 release 2.5.4.
1121
1122 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
1123 characters from a string of bytes) have been redefined so as not to use
1124 loops, in order to improve performance in some environments. At the same
1125 time, I abstracted some of the common code into auxiliary macros to save
1126 repetition (this should not affect the compiled code).
1127
1128 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
1129 compile-time error is now given if \c is not followed by an ASCII
1130 character, that is, a byte less than 128. (In EBCDIC mode, the code is
1131 different, and any byte value is allowed.)
1132
1133 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
1134 START_OPTIMIZE option, which is now allowed at compile time - but just
1135 passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
1136 to pcregrep and other applications that have no direct access to PCRE
1137 options. The new /Y option in pcretest sets this option when calling
1138 pcre_compile().
1139
1140 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
1141 back references. Groups containing recursive back references were forced to
1142 be atomic by that change, but in the case of named groups, the amount of
1143 memory required was incorrectly computed, leading to "Failed: internal
1144 error: code overflow". This has been fixed.
1145
1146 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
1147 pcretest.c, to avoid build problems in some Borland environments.
1148
1149
1150 Version 8.10 25-Jun-2010
1151 ------------------------
1152
1153 1. Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
1154 THEN.
1155
1156 2. (*ACCEPT) was not working when inside an atomic group.
1157
1158 3. Inside a character class, \B is treated as a literal by default, but
1159 faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
1160 causes the error). The code is unchanged, but I tidied the documentation.
1161
1162 4. Inside a character class, PCRE always treated \R and \X as literals,
1163 whereas Perl faults them if its -w option is set. I have changed PCRE so
1164 that it faults them when PCRE_EXTRA is set.
1165
1166 5. Added support for \N, which always matches any character other than
1167 newline. (It is the same as "." when PCRE_DOTALL is not set.)
1168
1169 6. When compiling pcregrep with newer versions of gcc which may have
1170 FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
1171 declared with attribute warn_unused_result" were given. Just casting the
1172 result to (void) does not stop the warnings; a more elaborate fudge is
1173 needed. I've used a macro to implement this.
1174
1175 7. Minor change to pcretest.c to avoid a compiler warning.
1176
1177 8. Added four artifical Unicode properties to help with an option to make
1178 \s etc use properties (see next item). The new properties are: Xan
1179 (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
1180
1181 9. Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
1182 use Unicode properties. (*UCP) at the start of a pattern can be used to set
1183 this option. Modified pcretest to add /W to test this facility. Added
1184 REG_UCP to make it available via the POSIX interface.
1185
1186 10. Added --line-buffered to pcregrep.
1187
1188 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
1189 studied, and the match started with a letter with a code point greater than
1190 127 whose first byte was different to the first byte of the other case of
1191 the letter, the other case of this starting letter was not recognized
1192 (#976).
1193
1194 12. If a pattern that was studied started with a repeated Unicode property
1195 test, for example, \p{Nd}+, there was the theoretical possibility of
1196 setting up an incorrect bitmap of starting bytes, but fortunately it could
1197 not have actually happened in practice until change 8 above was made (it
1198 added property types that matched character-matching opcodes).
1199
1200 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
1201 possible starting bytes for non-anchored patterns.
1202
1203 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
1204 \R, and also a number of cases that involve Unicode properties, both
1205 explicit and implicit when PCRE_UCP is set.
1206
1207 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
1208 input, it could crash or give wrong results if characters with values
1209 greater than 0xc0 were present in the subject string. (Detail: it assumed
1210 UTF-8 input when processing these items.)
1211
1212 16. Added a lot of (int) casts to avoid compiler warnings in systems where
1213 size_t is 64-bit (#991).
1214
1215 17. Added a check for running out of memory when PCRE is compiled with
1216 --disable-stack-for-recursion (#990).
1217
1218 18. If the last data line in a file for pcretest does not have a newline on
1219 the end, a newline was missing in the output.
1220
1221 19. The default pcre_chartables.c file recognizes only ASCII characters (values
1222 less than 128) in its various bitmaps. However, there is a facility for
1223 generating tables according to the current locale when PCRE is compiled. It
1224 turns out that in some environments, 0x85 and 0xa0, which are Unicode space
1225 characters, are recognized by isspace() and therefore were getting set in
1226 these tables, and indeed these tables seem to approximate to ISO 8859. This
1227 caused a problem in UTF-8 mode when pcre_study() was used to create a list
1228 of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
1229 which of course cannot start UTF-8 characters. I have changed the code so
1230 that only real ASCII characters (less than 128) and the correct starting
1231 bytes for UTF-8 encodings are set for characters greater than 127 when in
1232 UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
1233 altogether.)
1234
1235 20. Added the /T option to pcretest so as to be able to run tests with non-
1236 standard character tables, thus making it possible to include the tests
1237 used for 19 above in the standard set of tests.
1238
1239 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
1240 reference to a subpattern the other side of a comment that contains an
1241 opening parenthesis caused either an internal compiling error, or a
1242 reference to the wrong subpattern.
1243
1244
1245 Version 8.02 19-Mar-2010
1246 ------------------------
1247
1248 1. The Unicode data tables have been updated to Unicode 5.2.0.
1249
1250 2. Added the option --libs-cpp to pcre-config, but only when C++ support is
1251 configured.
1252
1253 3. Updated the licensing terms in the pcregexp.pas file, as agreed with the
1254 original author of that file, following a query about its status.
1255
1256 4. On systems that do not have stdint.h (e.g. Solaris), check for and include
1257 inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
1258
1259 5. A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
1260 quantifier applied to a forward-referencing subroutine call, could compile
1261 incorrect code or give the error "internal error: previously-checked
1262 referenced subpattern not found".
1263
1264 6. Both MS Visual Studio and Symbian OS have problems with initializing
1265 variables to point to external functions. For these systems, therefore,
1266 pcre_malloc etc. are now initialized to local functions that call the
1267 relevant global functions.
1268
1269 7. There were two entries missing in the vectors called coptable and poptable
1270 in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
1271 I've fixed the data, and added a kludgy way of testing at compile time that
1272 the lengths are correct (equal to the number of opcodes).
1273
1274 8. Following on from 7, I added a similar kludge to check the length of the
1275 eint vector in pcreposix.c.
1276
1277 9. Error texts for pcre_compile() are held as one long string to avoid too
1278 much relocation at load time. To find a text, the string is searched,
1279 counting zeros. There was no check for running off the end of the string,
1280 which could happen if a new error number was added without updating the
1281 string.
1282
1283 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
1284
1285 11. \K was not working if it appeared in an atomic group or in a group that
1286 was called as a "subroutine", or in an assertion. Perl 5.11 documents that
1287 \K is "not well defined" if used in an assertion. PCRE now accepts it if
1288 the assertion is positive, but not if it is negative.
1289
1290 12. Change 11 fortuitously reduced the size of the stack frame used in the
1291 "match()" function of pcre_exec.c by one pointer. Forthcoming
1292 implementation of support for (*MARK) will need an extra pointer on the
1293 stack; I have reserved it now, so that the stack frame size does not
1294 decrease.
1295
1296 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
1297 item in branch that calls a recursion is a subroutine call - as in the
1298 second branch in the above example - was incorrectly given the compile-
1299 time error "recursive call could loop indefinitely" because pcre_compile()
1300 was not correctly checking the subroutine for matching a non-empty string.
1301
1302 14. The checks for overrunning compiling workspace could trigger after an
1303 overrun had occurred. This is a "should never occur" error, but it can be
1304 triggered by pathological patterns such as hundreds of nested parentheses.
1305 The checks now trigger 100 bytes before the end of the workspace.
1306
1307 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
1308
1309
1310 Version 8.01 19-Jan-2010
1311 ------------------------
1312
1313 1. If a pattern contained a conditional subpattern with only one branch (in
1314 particular, this includes all (*DEFINE) patterns), a call to pcre_study()
1315 computed the wrong minimum data length (which is of course zero for such
1316 subpatterns). This could cause incorrect "no match" results.
1317
1318 2. For patterns such as (?i)a(?-i)b|c where an option setting at the start of
1319 the pattern is reset in the first branch, pcre_compile() failed with
1320 "internal error: code overflow at offset...". This happened only when
1321 the reset was to the original external option setting. (An optimization
1322 abstracts leading options settings into an external setting, which was the
1323 cause of this.)
1324
1325 3. A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
1326 of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
1327 assertion pattern did not match (meaning that the assertion was true), it
1328 was incorrectly treated as false if the SKIP had been reached during the
1329 matching. This also applied to assertions used as conditions.
1330
1331 4. If an item that is not supported by pcre_dfa_exec() was encountered in an
1332 assertion subpattern, including such a pattern used as a condition,
1333 unpredictable results occurred, instead of the error return
1334 PCRE_ERROR_DFA_UITEM.
1335
1336 5. The C++ GlobalReplace function was not working like Perl for the special
1337 situation when an empty string is matched. It now does the fancy magic
1338 stuff that is necessary.
1339
1340 6. In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
1341 removed. (These were left over from very, very early versions of PCRE.)
1342
1343 7. Some cosmetic changes to the code to make life easier when compiling it
1344 as part of something else:
1345
1346 (a) Change DEBUG to PCRE_DEBUG.
1347
1348 (b) In pcre_compile(), rename the member of the "branch_chain" structure
1349 called "current" as "current_branch", to prevent a collision with the
1350 Linux macro when compiled as a kernel module.
1351
1352 (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
1353 prevent a collision with the Linux macro when compiled as a kernel
1354 module.
1355
1356 8. In pcre_compile() there are some checks for integer overflows that used to
1357 cast potentially large values to (double). This has been changed to that
1358 when building, a check for int64_t is made, and if it is found, it is used
1359 instead, thus avoiding the use of floating point arithmetic. (There is no
1360 other use of FP in PCRE.) If int64_t is not found, the fallback is to
1361 double.
1362
1363 9. Added two casts to avoid signed/unsigned warnings from VS Studio Express
1364 2005 (difference between two addresses compared to an unsigned value).
1365
1366 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
1367 custom one, because of the following reported problem in Windows:
1368
1369 - libbz2 uses the Pascal calling convention (WINAPI) for the functions
1370 under Win32.
1371 - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
1372 therefore missing the function definition.
1373 - The compiler thus generates a "C" signature for the test function.
1374 - The linker fails to find the "C" function.
1375 - PCRE fails to configure if asked to do so against libbz2.
1376
1377 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
1378 messages were output:
1379
1380 Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
1381 rerunning libtoolize, to keep the correct libtool macros in-tree.
1382 Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
1383
1384 I have done both of these things.
1385
1386 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
1387 most of the time, it *can* run out if it is given a pattern that contains a
1388 runaway infinite recursion. I updated the discussion in the pcrestack man
1389 page.
1390
1391 13. Now that we have gone to the x.xx style of version numbers, the minor
1392 version may start with zero. Using 08 or 09 is a bad idea because users
1393 might check the value of PCRE_MINOR in their code, and 08 or 09 may be
1394 interpreted as invalid octal numbers. I've updated the previous comment in
1395 configure.ac, and also added a check that gives an error if 08 or 09 are
1396 used.
1397
1398 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
1399 causing partial matching to fail when the end of the subject matched \W
1400 in a UTF-8 pattern where \W was quantified with a minimum of 3.
1401
1402 15. There were some discrepancies between the declarations in pcre_internal.h
1403 of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
1404 their definitions. The declarations used "const uschar *" and the
1405 definitions used USPTR. Even though USPTR is normally defined as "const
1406 unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
1407 reported that: "This difference in casting confuses some C++ compilers, for
1408 example, SunCC recognizes above declarations as different functions and
1409 generates broken code for hbpcre." I have changed the declarations to use
1410 USPTR.
1411
1412 16. GNU libtool is named differently on some systems. The autogen.sh script now
1413 tries several variants such as glibtoolize (MacOSX) and libtoolize1x
1414 (FreeBSD).
1415
1416 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
1417 (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
1418 comment: "Figure out how to create a longlong from a string: strtoll and
1419 equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
1420 instance, but it only takes 2 args instead of 3!"
1421
1422 18. A subtle bug concerned with back references has been fixed by a change of
1423 specification, with a corresponding code fix. A pattern such as
1424 ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
1425 refers, was giving matches when it shouldn't. For example, xa=xaaa would
1426 match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
1427 same bug. Such groups have to be quantified to be useful, or contained
1428 inside another quantified group. (If there's no repetition, the reference
1429 can never match.) The problem arises because, having left the group and
1430 moved on to the rest of the pattern, a later failure that backtracks into
1431 the group uses the captured value from the final iteration of the group
1432 rather than the correct earlier one. I have fixed this in PCRE by forcing
1433 any group that contains a reference to itself to be an atomic group; that
1434 is, there cannot be any backtracking into it once it has completed. This is
1435 similar to recursive and subroutine calls.
1436
1437
1438 Version 8.00 19-Oct-09
1439 ----------------------
1440
1441 1. The table for translating pcre_compile() error codes into POSIX error codes
1442 was out-of-date, and there was no check on the pcre_compile() error code
1443 being within the table. This could lead to an OK return being given in
1444 error.
1445
1446 2. Changed the call to open a subject file in pcregrep from fopen(pathname,
1447 "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
1448 in a Windows environment.
1449
1450 3. The pcregrep --count option prints the count for each file even when it is
1451 zero, as does GNU grep. However, pcregrep was also printing all files when
1452 --files-with-matches was added. Now, when both options are given, it prints
1453 counts only for those files that have at least one match. (GNU grep just
1454 prints the file name in this circumstance, but including the count seems
1455 more useful - otherwise, why use --count?) Also ensured that the
1456 combination -clh just lists non-zero counts, with no names.
1457
1458 4. The long form of the pcregrep -F option was incorrectly implemented as
1459 --fixed_strings instead of --fixed-strings. This is an incompatible change,
1460 but it seems right to fix it, and I didn't think it was worth preserving
1461 the old behaviour.
1462
1463 5. The command line items --regex=pattern and --regexp=pattern were not
1464 recognized by pcregrep, which required --regex pattern or --regexp pattern
1465 (with a space rather than an '='). The man page documented the '=' forms,
1466 which are compatible with GNU grep; these now work.
1467
1468 6. No libpcreposix.pc file was created for pkg-config; there was just
1469 libpcre.pc and libpcrecpp.pc. The omission has been rectified.
1470
1471 7. Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
1472 when UCP support is not needed, by modifying the Python script that
1473 generates it from Unicode data files. This should not matter if the module
1474 is correctly used as a library, but I received one complaint about 50K of
1475 unwanted data. My guess is that the person linked everything into his
1476 program rather than using a library. Anyway, it does no harm.
1477
1478 8. A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
1479 was a minimum greater than 1 for a wide character in a possessive
1480 repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
1481 which had an unlimited repeat of a nested, fixed maximum repeat of a wide
1482 character. Chaos in the form of incorrect output or a compiling loop could
1483 result.
1484
1485 9. The restrictions on what a pattern can contain when partial matching is
1486 requested for pcre_exec() have been removed. All patterns can now be
1487 partially matched by this function. In addition, if there are at least two
1488 slots in the offset vector, the offset of the earliest inspected character
1489 for the match and the offset of the end of the subject are set in them when
1490 PCRE_ERROR_PARTIAL is returned.
1491
1492 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
1493 synonymous with PCRE_PARTIAL, for backwards compatibility, and
1494 PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
1495 and may be more useful for multi-segment matching.
1496
1497 11. Partial matching with pcre_exec() is now more intuitive. A partial match
1498 used to be given if ever the end of the subject was reached; now it is
1499 given only if matching could not proceed because another character was
1500 needed. This makes a difference in some odd cases such as Z(*FAIL) with the
1501 string "Z", which now yields "no match" instead of "partial match". In the
1502 case of pcre_dfa_exec(), "no match" is given if every matching path for the
1503 final character ended with (*FAIL).
1504
1505 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
1506 if the pattern had a "must contain" character that was already found in the
1507 earlier partial match, unless partial matching was again requested. For
1508 example, with the pattern /dog.(body)?/, the "must contain" character is
1509 "g". If the first part-match was for the string "dog", restarting with
1510 "sbody" failed. This bug has been fixed.
1511
1512 13. The string returned by pcre_dfa_exec() after a partial match has been
1513 changed so that it starts at the first inspected character rather than the
1514 first character of the match. This makes a difference only if the pattern
1515 starts with a lookbehind assertion or \b or \B (\K is not supported by
1516 pcre_dfa_exec()). It's an incompatible change, but it makes the two
1517 matching functions compatible, and I think it's the right thing to do.
1518
1519 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
1520 so that the demonstration program is easily available in environments where
1521 PCRE has not been installed from source.
1522
1523 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
1524 libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
1525 library.
1526
1527 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
1528 It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
1529 is not the first non-POSIX option to be added. Clearly some people find
1530 these options useful.
1531
1532 17. If a caller to the POSIX matching function regexec() passes a non-zero
1533 value for nmatch with a NULL value for pmatch, the value of
1534 nmatch is forced to zero.
1535
1536 18. RunGrepTest did not have a test for the availability of the -u option of
1537 the diff command, as RunTest does. It now checks in the same way as
1538 RunTest, and also checks for the -b option.
1539
1540 19. If an odd number of negated classes containing just a single character
1541 interposed, within parentheses, between a forward reference to a named
1542 subpattern and the definition of the subpattern, compilation crashed with
1543 an internal error, complaining that it could not find the referenced
1544 subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
1545 [The bug was that it was starting one character too far in when skipping
1546 over the character class, thus treating the ] as data rather than
1547 terminating the class. This meant it could skip too much.]
1548
1549 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
1550 /g option in pcretest when the pattern contains \K, which makes it possible
1551 to have an empty string match not at the start, even when the pattern is
1552 anchored. Updated pcretest and pcredemo to use this option.
1553
1554 21. If the maximum number of capturing subpatterns in a recursion was greater
1555 than the maximum at the outer level, the higher number was returned, but
1556 with unset values at the outer level. The correct (outer level) value is
1557 now given.
1558
1559 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
1560 PCRE did not set those parentheses (unlike Perl). I have now found a way to
1561 make it do so. The string so far is captured, making this feature
1562 compatible with Perl.
1563
1564 23. The tests have been re-organized, adding tests 11 and 12, to make it
1565 possible to check the Perl 5.10 features against Perl 5.10.
1566
1567 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
1568 pattern matches a fixed length string. PCRE did not allow this; now it
1569 does. Neither allows recursion.
1570
1571 25. I finally figured out how to implement a request to provide the minimum
1572 length of subject string that was needed in order to match a given pattern.
1573 (It was back references and recursion that I had previously got hung up
1574 on.) This code has now been added to pcre_study(); it finds a lower bound
1575 to the length of subject needed. It is not necessarily the greatest lower
1576 bound, but using it to avoid searching strings that are too short does give
1577 some useful speed-ups. The value is available to calling programs via
1578 pcre_fullinfo().
1579
1580 26. While implementing 25, I discovered to my embarrassment that pcretest had
1581 not been passing the result of pcre_study() to pcre_dfa_exec(), so the
1582 study optimizations had never been tested with that matching function.
1583 Oops. What is worse, even when it was passed study data, there was a bug in
1584 pcre_dfa_exec() that meant it never actually used it. Double oops. There
1585 were also very few tests of studied patterns with pcre_dfa_exec().
1586
1587 27. If (?| is used to create subpatterns with duplicate numbers, they are now
1588 allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
1589 on the other side of the coin, they are no longer allowed to have different
1590 names, because these cannot be distinguished in PCRE, and this has caused
1591 confusion. (This is a difference from Perl.)
1592
1593 28. When duplicate subpattern names are present (necessarily with different
1594 numbers, as required by 27 above), and a test is made by name in a
1595 conditional pattern, either for a subpattern having been matched, or for
1596 recursion in such a pattern, all the associated numbered subpatterns are
1597 tested, and the overall condition is true if the condition is true for any
1598 one of them. This is the way Perl works, and is also more like the way
1599 testing by number works.
1600
1601
1602 Version 7.9 11-Apr-09
1603 ---------------------
1604
1605 1. When building with support for bzlib/zlib (pcregrep) and/or readline
1606 (pcretest), all targets were linked against these libraries. This included
1607 libpcre, libpcreposix, and libpcrecpp, even though they do not use these
1608 libraries. This caused unwanted dependencies to be created. This problem
1609 has been fixed, and now only pcregrep is linked with bzlib/zlib and only
1610 pcretest is linked with readline.
1611
1612 2. The "typedef int BOOL" in pcre_internal.h that was included inside the
1613 "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
1614 moved outside it again, because FALSE and TRUE are already defined in AIX,
1615 but BOOL is not.
1616
1617 3. The pcre_config() function was treating the PCRE_MATCH_LIMIT and
1618 PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
1619
1620 4. The pcregrep documentation said spaces were inserted as well as colons (or
1621 hyphens) following file names and line numbers when outputting matching
1622 lines. This is not true; no spaces are inserted. I have also clarified the
1623 wording for the --colour (or --color) option.
1624
1625 5. In pcregrep, when --colour was used with -o, the list of matching strings
1626 was not coloured; this is different to GNU grep, so I have changed it to be
1627 the same.
1628
1629 6. When --colo(u)r was used in pcregrep, only the first matching substring in
1630 each matching line was coloured. Now it goes on to look for further matches
1631 of any of the test patterns, which is the same behaviour as GNU grep.
1632
1633 7. A pattern that could match an empty string could cause pcregrep to loop; it
1634 doesn't make sense to accept an empty string match in pcregrep, so I have
1635 locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
1636 seems to be how GNU grep behaves.
1637
1638 8. The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
1639 start or after a newline", because the conditional assertion was not being
1640 correctly handled. The rule now is that both the assertion and what follows
1641 in the first alternative must satisfy the test.
1642
1643 9. If auto-callout was enabled in a pattern with a conditional group whose
1644 condition was an assertion, PCRE could crash during matching, both with
1645 pcre_exec() and pcre_dfa_exec().
1646
1647 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
1648 used for matching.
1649
1650 11. Unicode property support in character classes was not working for
1651 characters (bytes) greater than 127 when not in UTF-8 mode.
1652
1653 12. Added the -M command line option to pcretest.
1654
1655 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
1656
1657 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
1658
1659 16. Added comments and documentation about mis-use of no_arg in the C++
1660 wrapper.
1661
1662 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
1663 from Martin Jerabek that uses macro names for all relevant character and
1664 string constants.
1665
1666 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
1667 SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
1668 SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
1669 these, but not everybody uses configure.
1670
1671 19. A conditional group that had only one branch was not being correctly
1672 recognized as an item that could match an empty string. This meant that an
1673 enclosing group might also not be so recognized, causing infinite looping
1674 (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
1675 with the subject "ab", where knowledge that the repeated group can match
1676 nothing is needed in order to break the loop.
1677
1678 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
1679 exec(), but without supplying a callout function, matching went wrong.
1680
1681 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
1682 leak if the size of the offset vector was greater than 30. When the vector
1683 is smaller, the saved offsets during recursion go onto a local stack
1684 vector, but for larger vectors malloc() is used. It was failing to free
1685 when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
1686 error, in fact).
1687
1688 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
1689 heapframe that is used only when UTF-8 support is enabled. This caused no
1690 problem, but was untidy.
1691
1692 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
1693 CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
1694 included within another project.
1695
1696 24. Steven Van Ingelgem's patches to add more options to the CMake support,
1697 slightly modified by me:
1698
1699 (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
1700 not building pcregrep.
1701
1702 (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
1703 if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
1704
1705 25. Forward references, both numeric and by name, in patterns that made use of
1706 duplicate group numbers, could behave incorrectly or give incorrect errors,
1707 because when scanning forward to find the reference group, PCRE was not
1708 taking into account the duplicate group numbers. A pattern such as
1709 ^X(?3)(a)(?|(b)|(q))(Y) is an example.
1710
1711 26. Changed a few more instances of "const unsigned char *" to USPTR, making
1712 the feature of a custom pointer more persuasive (as requested by a user).
1713
1714 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
1715 pcretest.c, inside #ifndefs, because it seems they are sometimes already
1716 pre-defined.
1717
1718 28. Added support for (*UTF8) at the start of a pattern.
1719
1720 29. Arrange for flags added by the "release type" setting in CMake to be shown
1721 in the configuration summary.
1722
1723
1724 Version 7.8 05-Sep-08
1725 ---------------------
1726
1727 1. Replaced UCP searching code with optimized version as implemented for Ad
1728 Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
1729 stage table and inline lookup instead of a function, giving speed ups of 2
1730 to 5 times on some simple patterns that I tested. Permission was given to
1731 distribute the MultiStage2.py script that generates the tables (it's not in
1732 the tarball, but is in the Subversion repository).
1733
1734 2. Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
1735 scripts.
1736
1737 3. Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
1738 a group with a zero qualifier. The result of the study could be incorrect,
1739 or the function might crash, depending on the pattern.
1740
1741 4. Caseless matching was not working for non-ASCII characters in back
1742 references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
1743 It now works when Unicode Property Support is available.
1744
1745 5. In pcretest, an escape such as \x{de} in the data was always generating
1746 a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
1747 non-UTF-8 mode. If the value is greater than 255, it gives a warning about
1748 truncation.
1749
1750 6. Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
1751
1752 7. Added two (int) casts to pcregrep when printing the difference of two
1753 pointers, in case they are 64-bit values.
1754
1755 8. Added comments about Mac OS X stack usage to the pcrestack man page and to
1756 test 2 if it fails.
1757
1758 9. Added PCRE_CALL_CONVENTION just before the names of all exported functions,
1759 and a #define of that name to empty if it is not externally set. This is to
1760 allow users of MSVC to set it if necessary.
1761
1762 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
1763 the convenience functions in the pcre_get.c source file.
1764
1765 11. An option change at the start of a pattern that had top-level alternatives
1766 could cause overwriting and/or a crash. This command provoked a crash in
1767 some environments:
1768
1769 printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
1770
1771 This potential security problem was recorded as CVE-2008-2371.
1772
1773 12. For a pattern where the match had to start at the beginning or immediately
1774 after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
1775 pcre_dfa_exec() could read past the end of the passed subject if there was
1776 no match. To help with detecting such bugs (e.g. with valgrind), I modified
1777 pcretest so that it places the subject at the end of its malloc-ed buffer.
1778
1779 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
1780 exec() might read past the end of the data buffer in UTF-8 mode.
1781
1782 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
1783 the data contained the byte 0x85 as part of a UTF-8 character within its
1784 first line. This applied both to normal and DFA matching.
1785
1786 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
1787 /^[^d]*?$/8 failed to match "abc".
1788
1789 16. Added a missing copyright notice to pcrecpp_internal.h.
1790
1791 17. Make it more clear in the documentation that values returned from
1792 pcre_exec() in ovector are byte offsets, not character counts.
1793
1794 18. Tidied a few places to stop certain compilers from issuing warnings.
1795
1796 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
1797 supplied by Stefan Weber. I made a further small update for 7.8 because
1798 there is a change of source arrangements: the pcre_searchfuncs.c module is
1799 replaced by pcre_ucd.c.
1800
1801
1802 Version 7.7 07-May-08
1803 ---------------------
1804
1805 1. Applied Craig's patch to sort out a long long problem: "If we can't convert
1806 a string to a long long, pretend we don't even have a long long." This is
1807 done by checking for the strtoq, strtoll, and _strtoi64 functions.
1808
1809 2. Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
1810 pre-7.6 versions, which defined a global no_arg variable instead of putting
1811 it in the RE class. (See also #8 below.)
1812
1813 3. Remove a line of dead code, identified by coverity and reported by Nuno
1814 Lopes.
1815
1816 4. Fixed two related pcregrep bugs involving -r with --include or --exclude:
1817
1818 (1) The include/exclude patterns were being applied to the whole pathnames
1819 of files, instead of just to the final components.
1820
1821 (2) If there was more than one level of directory, the subdirectories were
1822 skipped unless they satisfied the include/exclude conditions. This is
1823 inconsistent with GNU grep (and could even be seen as contrary to the
1824 pcregrep specification - which I improved to make it absolutely clear).
1825 The action now is always to scan all levels of directory, and just
1826 apply the include/exclude patterns to regular files.
1827
1828 5. Added the --include_dir and --exclude_dir patterns to pcregrep, and used
1829 --exclude_dir in the tests to avoid scanning .svn directories.
1830
1831 6. Applied Craig's patch to the QuoteMeta function so that it escapes the
1832 NUL character as backslash + 0 rather than backslash + NUL, because PCRE
1833 doesn't support NULs in patterns.
1834
1835 7. Added some missing "const"s to declarations of static tables in
1836 pcre_compile.c and pcre_dfa_exec.c.
1837
1838 8. Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
1839 caused by fix #2 above. (Subsequently also a second patch to fix the
1840 first patch. And a third patch - this was a messy problem.)
1841
1842 9. Applied Craig's patch to remove the use of push_back().
1843
1844 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
1845 matching function regexec().
1846
1847 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
1848 which, however, unlike Perl's \g{...}, are subroutine calls, not back
1849 references. PCRE supports relative numbers with this syntax (I don't think
1850 Oniguruma does).
1851
1852 12. Previously, a group with a zero repeat such as (...){0} was completely
1853 omitted from the compiled regex. However, this means that if the group
1854 was called as a subroutine from elsewhere in the pattern, things went wrong
1855 (an internal error was given). Such groups are now left in the compiled
1856 pattern, with a new opcode that causes them to be skipped at execution
1857 time.
1858
1859 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
1860 to the way PCRE behaves:
1861
1862 (a) A lone ] character is dis-allowed (Perl treats it as data).
1863
1864 (b) A back reference to an unmatched subpattern matches an empty string
1865 (Perl fails the current match path).
1866
1867 (c) A data ] in a character class must be notated as \] because if the
1868 first data character in a class is ], it defines an empty class. (In
1869 Perl it is not possible to have an empty class.) The empty class []
1870 never matches; it forces failure and is equivalent to (*FAIL) or (?!).
1871 The negative empty class [^] matches any one character, independently
1872 of the DOTALL setting.
1873
1874 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
1875 non-existent subpattern following a character class starting with ']' and
1876 containing () gave an internal compiling error instead of "reference to
1877 non-existent subpattern". Fortunately, when the pattern did exist, the
1878 compiled code was correct. (When scanning forwards to check for the
1879 existencd of the subpattern, it was treating the data ']' as terminating
1880 the class, so got the count wrong. When actually compiling, the reference
1881 was subsequently set up correctly.)
1882
1883 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
1884 it was being rejected as not supported by pcre_dfa_exec(), even though
1885 other assertions are supported. I have made pcre_dfa_exec() support
1886 (*FAIL).
1887
1888 16. The implementation of 13c above involved the invention of a new opcode,
1889 OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
1890 cannot be changed at match time, I realized I could make a small
1891 improvement to matching performance by compiling OP_ALLANY instead of
1892 OP_ANY for "." when DOTALL was set, and then removing the runtime tests
1893 on the OP_ANY path.
1894
1895 17. Compiling pcretest on Windows with readline support failed without the
1896 following two fixes: (1) Make the unistd.h include conditional on
1897 HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
1898
1899 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
1900 ncurses library to be included for pcretest when ReadLine support is
1901 requested, but also to allow for it to be overridden. This patch came from
1902 Daniel Bergström.
1903
1904 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
1905 as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
1906 any errors with the current Unicode tables. Thanks to Peter Kankowski for
1907 spotting this.
1908
1909
1910 Version 7.6 28-Jan-08
1911 ---------------------
1912
1913 1. A character class containing a very large number of characters with
1914 codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
1915 overflow.
1916
1917 2. Patch to cut out the "long long" test in pcrecpp_unittest when
1918 HAVE_LONG_LONG is not defined.
1919
1920 3. Applied Christian Ehrlicher's patch to update the CMake build files to
1921 bring them up to date and include new features. This patch includes:
1922
1923 - Fixed PH's badly added libz and libbz2 support.
1924 - Fixed a problem with static linking.
1925 - Added pcredemo. [But later removed - see 7 below.]
1926 - Fixed dftables problem and added an option.
1927 - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
1928 HAVE_LONG_LONG.
1929 - Added readline support for pcretest.
1930 - Added an listing of the option settings after cmake has run.
1931
1932 4. A user submitted a patch to Makefile that makes it easy to create
1933 "pcre.dll" under mingw when using Configure/Make. I added stuff to
1934 Makefile.am that cause it to include this special target, without
1935 affecting anything else. Note that the same mingw target plus all
1936 the other distribution libraries and programs are now supported
1937 when configuring with CMake (see 6 below) instead of with
1938 Configure/Make.
1939
1940 5. Applied Craig's patch that moves no_arg into the RE class in the C++ code.
1941 This is an attempt to solve the reported problem "pcrecpp::no_arg is not
1942 exported in the Windows port". It has not yet been confirmed that the patch
1943 solves the problem, but it does no harm.
1944
1945 6. Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
1946 NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
1947 with CMake, and also correct the comment about stack recursion.
1948
1949 7. Remove the automatic building of pcredemo from the ./configure system and
1950 from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
1951 of a program that users should build themselves after PCRE is installed, so
1952 building it automatically is not really right. What is more, it gave
1953 trouble in some build environments.
1954
1955 8. Further tidies to CMakeLists.txt from Sheri and Christian.
1956
1957
1958 Version 7.5 10-Jan-08
1959 ---------------------
1960
1961 1. Applied a patch from Craig: "This patch makes it possible to 'ignore'
1962 values in parens when parsing an RE using the C++ wrapper."
1963
1964 2. Negative specials like \S did not work in character classes in UTF-8 mode.
1965 Characters greater than 255 were excluded from the class instead of being
1966 included.
1967
1968 3. The same bug as (2) above applied to negated POSIX classes such as
1969 [:^space:].
1970
1971 4. PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
1972 defined or documented. It seems to have been a typo for PCRE_STATIC, so
1973 I have changed it.
1974
1975 5. The construct (?&) was not diagnosed as a syntax error (it referenced the
1976 first named subpattern) and a construct such as (?&a) would reference the
1977 first named subpattern whose name started with "a" (in other words, the
1978 length check was missing). Both these problems are fixed. "Subpattern name
1979 expected" is now given for (?&) (a zero-length name), and this patch also
1980 makes it give the same error for \k'' (previously it complained that that
1981 was a reference to a non-existent subpattern).
1982
1983 6. The erroneous patterns (?+-a) and (?-+a) give different error messages;
1984 this is right because (?- can be followed by option settings as well as by
1985 digits. I have, however, made the messages clearer.
1986
1987 7. Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
1988 than the number used in the conditional) now cause a compile-time error.
1989 This is actually not compatible with Perl, which accepts such patterns, but
1990 treats the conditional as always being FALSE (as PCRE used to), but it
1991 seems to me that giving a diagnostic is better.
1992
1993 8. Change "alphameric" to the more common word "alphanumeric" in comments
1994 and messages.
1995
1996 9. Fix two occurrences of "backslash" in comments that should have been
1997 "backspace".
1998
1999 10. Remove two redundant lines of code that can never be obeyed (their function
2000 was moved elsewhere).
2001
2002 11. The program that makes PCRE's Unicode character property table had a bug
2003 which caused it to generate incorrect table entries for sequences of
2004 characters that have the same character type, but are in different scripts.
2005 It amalgamated them into a single range, with the script of the first of
2006 them. In other words, some characters were in the wrong script. There were
2007 thirteen such cases, affecting characters in the following ranges:
2008
2009 U+002b0 - U+002c1
2010 U+0060c - U+0060d
2011 U+0061e - U+00612
2012 U+0064b - U+0065e
2013 U+0074d - U+0076d
2014 U+01800 - U+01805
2015 U+01d00 - U+01d77
2016 U+01d9b - U+01dbf
2017 U+0200b - U+0200f
2018 U+030fc - U+030fe
2019 U+03260 - U+0327f
2020 U+0fb46 - U+0fbb1
2021 U+10450 - U+1049d
2022
2023 12. The -o option (show only the matching part of a line) for pcregrep was not
2024 compatible with GNU grep in that, if there was more than one match in a
2025 line, it showed only the first of them. It now behaves in the same way as
2026 GNU grep.
2027
2028 13. If the -o and -v options were combined for pcregrep, it printed a blank
2029 line for every non-matching line. GNU grep prints nothing, and pcregrep now
2030 does the same. The return code can be used to tell if there were any
2031 non-matching lines.
2032
2033 14. Added --file-offsets and --line-offsets to pcregrep.
2034
2035 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
2036 infinitely looping recursion. The bug was that positive lookaheads were not
2037 being skipped when checking for a possible empty match (negative lookaheads
2038 and both kinds of lookbehind were skipped).
2039
2040 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
2041 inclusion of <windows.h> to before rather than after the definition of
2042 INVALID_FILE_ATTRIBUTES (patch from David Byron).
2043
2044 17. Specifying a possessive quantifier with a specific limit for a Unicode
2045 character property caused pcre_compile() to compile bad code, which led at
2046 runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
2047 are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
2048 caused the error; without that there was no problem.
2049
2050 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
2051
2052 19. Added --enable-pcretest-libreadline.
2053
2054 20. In pcrecpp.cc, the variable 'count' was incremented twice in
2055 RE::GlobalReplace(). As a result, the number of replacements returned was
2056 double what it should be. I removed one of the increments, but Craig sent a
2057 later patch that removed the other one (the right fix) and added unit tests
2058 that check the return values (which was not done before).
2059
2060 21. Several CMake things:
2061
2062 (1) Arranged that, when cmake is used on Unix, the libraries end up with
2063 the names libpcre and libpcreposix, not just pcre and pcreposix.
2064
2065 (2) The above change means that pcretest and pcregrep are now correctly
2066 linked with the newly-built libraries, not previously installed ones.
2067
2068 (3) Added PCRE_SUPPORT_LIBREADLINE, PCRE_SUPPORT_LIBZ, PCRE_SUPPORT_LIBBZ2.
2069
2070 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
2071 crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
2072 UTF-8 newline character). The key issue is that the pattern starts .*;
2073 this means that the match must be either at the beginning, or after a
2074 newline. The bug was in the code for advancing after a failed match and
2075 checking that the new position followed a newline. It was not taking
2076 account of UTF-8 characters correctly.
2077
2078 23. PCRE was behaving differently from Perl in the way it recognized POSIX
2079 character classes. PCRE was not treating the sequence [:...:] as a
2080 character class unless the ... were all letters. Perl, however, seems to
2081 allow any characters between [: and :], though of course it rejects as
2082 unknown any "names" that contain non-letters, because all the known class
2083 names consist only of letters. Thus, Perl gives an error for [[:1234:]],
2084 for example, whereas PCRE did not - it did not recognize a POSIX character
2085 class. This seemed a bit dangerous, so the code has been changed to be
2086 closer to Perl. The behaviour is not identical to Perl, because PCRE will
2087 diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
2088 treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
2089 Perl does, and where it didn't before.
2090
2091 24. Rewrite so as to remove the single use of %n from pcregrep because in some
2092 Windows environments %n is disabled by default.
2093
2094
2095 Version 7.4 21-Sep-07
2096 ---------------------
2097
2098 1. Change 7.3/28 was implemented for classes by looking at the bitmap. This
2099 means that a class such as [\s] counted as "explicit reference to CR or
2100 LF". That isn't really right - the whole point of the change was to try to
2101 help when there was an actual mention of one of the two characters. So now
2102 the change happens only if \r or \n (or a literal CR or LF) character is
2103 encountered.
2104
2105 2. The 32-bit options word was also used for 6 internal flags, but the numbers
2106 of both had grown to the point where there were only 3 bits left.
2107 Fortunately, there was spare space in the data structure, and so I have
2108 moved the internal flags into a new 16-bit field to free up more option
2109 bits.
2110
2111 3. The appearance of (?J) at the start of a pattern set the DUPNAMES option,
2112 but did not set the internal JCHANGED flag - either of these is enough to
2113 control the way the "get" function works - but the PCRE_INFO_JCHANGED
2114 facility is supposed to tell if (?J) was ever used, so now (?J) at the
2115 start sets both bits.
2116
2117 4. Added options (at build time, compile time, exec time) to change \R from
2118 matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
2119
2120 5. doc/pcresyntax.html was missing from the distribution.
2121
2122 6. Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
2123 compatibility, even though it is no longer used.
2124
2125 7. Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
2126 strtoull to pcrecpp.cc to select the available functions in WIN32 when the
2127 windows.h file is present (where different names are used). [This was
2128 reversed later after testing - see 16 below.]
2129
2130 8. Changed all #include <config.h> to #include "config.h". There were also
2131 some further <pcre.h> cases that I changed to "pcre.h".
2132
2133 9. When pcregrep was used with the --colour option, it missed the line ending
2134 sequence off the lines that it output.
2135
2136 10. It was pointed out to me that arrays of string pointers cause lots of
2137 relocations when a shared library is dynamically loaded. A technique of
2138 using a single long string with a table of offsets can drastically reduce
2139 these. I have refactored PCRE in four places to do this. The result is
2140 dramatic:
2141
2142 Originally: 290
2143 After changing UCP table: 187
2144 After changing error message table: 43
2145 After changing table of "verbs" 36
2146 After changing table of Posix names 22
2147
2148 Thanks to the folks working on Gregex for glib for this insight.
2149
2150 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
2151 unicode-properties was also set.
2152
2153 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
2154
2155 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
2156 checked only for CRLF.
2157
2158 14. Added casts to pcretest.c to avoid compiler warnings.
2159
2160 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
2161
2162 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
2163 and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
2164 entirely. This removes changes made in 7 above.
2165
2166 17. The CMake files have been updated, and there is now more information about
2167 building with CMake in the NON-UNIX-USE document.
2168
2169
2170 Version 7.3 28-Aug-07
2171 ---------------------
2172
2173 1. In the rejigging of the build system that eventually resulted in 7.1, the
2174 line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
2175 brackets there is not right, since it causes compilers to look for an
2176 installed pcre.h, not the version that is in the source that is being
2177 compiled (which of course may be different). I have changed it back to:
2178
2179 #include "pcre.h"
2180
2181 I have a vague recollection that the change was concerned with compiling in
2182 different directories, but in the new build system, that is taken care of
2183 by the VPATH setting the Makefile.
2184
2185 2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
2186 when the subject happened to end in the byte 0x85 (e.g. if the last
2187 character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
2188 characters but of course it shouldn't be taken as a newline when it is part
2189 of another character. The bug was that, for an unlimited repeat of . in
2190 not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
2191 characters when looking for a newline.
2192
2193 3. A small performance improvement in the DOTALL UTF-8 mode .* case.
2194
2195 4. Debugging: adjusted the names of opcodes for different kinds of parentheses
2196 in debug output.
2197
2198 5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
2199 long printing in the pcrecpp unittest when running under MinGW.
2200
2201 6. ESC_K was left out of the EBCDIC table.
2202
2203 7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
2204 parentheses; I made it 1000, which seemed large enough. Unfortunately, the
2205 limit also applies to "virtual nesting" when a pattern is recursive, and in
2206 this case 1000 isn't so big. I have been able to remove this limit at the
2207 expense of backing off one optimization in certain circumstances. Normally,
2208 when pcre_exec() would call its internal match() function recursively and
2209 immediately return the result unconditionally, it uses a "tail recursion"
2210 feature to save stack. However, when a subpattern that can match an empty
2211 string has an unlimited repetition quantifier, it no longer makes this
2212 optimization. That gives it a stack frame in which to save the data for
2213 checking that an empty string has been matched. Previously this was taken
2214 from the 1000-entry workspace that had been reserved. So now there is no
2215 explicit limit, but more stack is used.
2216
2217 8. Applied Daniel's patches to solve problems with the import/export magic
2218 syntax that is required for Windows, and which was going wrong for the
2219 pcreposix and pcrecpp parts of the library. These were overlooked when this
2220 problem was solved for the main library.
2221
2222 9. There were some crude static tests to avoid integer overflow when computing
2223 the size of patterns that contain repeated groups with explicit upper
2224 limits. As the maximum quantifier is 65535, the maximum group length was
2225 set at 30,000 so that the product of these two numbers did not overflow a
2226 32-bit integer. However, it turns out that people want to use groups that
2227 are longer than 30,000 bytes (though not repeat them that many times).
2228 Change 7.0/17 (the refactoring of the way the pattern size is computed) has
2229 made it possible to implement the integer overflow checks in a much more
2230 dynamic way, which I have now done. The artificial limitation on group
2231 length has been removed - we now have only the limit on the total length of
2232 the compiled pattern, which depends on the LINK_SIZE setting.
2233
2234 10. Fixed a bug in the documentation for get/copy named substring when
2235 duplicate names are permitted. If none of the named substrings are set, the
2236 functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
2237 empty string.
2238
2239 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
2240 instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
2241 because the ] is interpreted as the first data character and the
2242 terminating ] is not found. PCRE has been made compatible with Perl in this
2243 regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
2244 cause memory overwriting.
2245
2246 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
2247 string has been matched (to stop an infinite loop). It was not recognizing
2248 a conditional subpattern that could match an empty string if that
2249 subpattern was within another subpattern. For example, it looped when
2250 trying to match (((?(1)X|))*) but it was OK with ((?(1)X|)*) where the
2251 condition was not nested. This bug has been fixed.
2252
2253 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
2254 past the start of the subject in the presence of bytes with the top bit
2255 set, for example "\x8aBCD".
2256
2257 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
2258 (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
2259
2260 14. Optimized (?!) to (*FAIL).
2261
2262 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
2263 This restricts code points to be within the range 0 to 0x10FFFF, excluding
2264 the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
2265 full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
2266 does: it's just the validity check that is more restrictive.
2267
2268 16. Inserted checks for integer overflows during escape sequence (backslash)
2269 processing, and also fixed erroneous offset values for syntax errors during
2270 backslash processing.
2271
2272 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
2273 for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
2274
2275 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
2276 caused an overrun.
2277
2278 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
2279 something other than just ASCII characters) inside a group that had an
2280 unlimited repeat caused a loop at compile time (while checking to see
2281 whether the group could match an empty string).
2282
2283 20. Debugging a pattern containing \p or \P could cause a crash. For example,
2284 [\P{Any}] did so. (Error in the code for printing property names.)
2285
2286 21. An orphan \E inside a character class could cause a crash.
2287
2288 22. A repeated capturing bracket such as (A)? could cause a wild memory
2289 reference during compilation.
2290
2291 23. There are several functions in pcre_compile() that scan along a compiled
2292 expression for various reasons (e.g. to see if it's fixed length for look
2293 behind). There were bugs in these functions when a repeated \p or \P was
2294 present in the pattern. These operators have additional parameters compared
2295 with \d, etc, and these were not being taken into account when moving along
2296 the compiled data. Specifically:
2297
2298 (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
2299 length.
2300
2301 (b) An item such as \pL+ within a repeated group could cause crashes or
2302 loops.
2303
2304 (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
2305 "reference to non-existent subpattern" error.
2306
2307 (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
2308
2309 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
2310 characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
2311
2312 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
2313
2314 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
2315 character were causing crashes (broken optimization).
2316
2317 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
2318 \p or \P) caused a compile-time loop.
2319
2320 28. More problems have arisen in unanchored patterns when CRLF is a valid line
2321 break. For example, the unstudied pattern [\r\n]A does not match the string
2322 "\r\nA" because change 7.0/46 below moves the current point on by two
2323 characters after failing to match at the start. However, the pattern \nA
2324 *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
2325 the same is true. There doesn't seem any very clean way out of this, but
2326 what I have chosen to do makes the common cases work: PCRE now takes note
2327 of whether there can be an explicit match for \r or \n anywhere in the
2328 pattern, and if so, 7.0/46 no longer applies. As part of this change,
2329 there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
2330 pattern has explicit CR or LF references.
2331
2332 29. Added (*CR) etc for changing newline setting at start of pattern.
2333
2334
2335 Version 7.2 19-Jun-07
2336 ---------------------
2337
2338 1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
2339 which is apparently normally available under Windows.
2340
2341 2. Re-jig the pcregrep tests with different newline settings in an attempt
2342 to make them independent of the local environment's newline setting.
2343
2344 3. Add code to configure.ac to remove -g from the CFLAGS default settings.
2345
2346 4. Some of the "internals" tests were previously cut out when the link size
2347 was not 2, because the output contained actual offsets. The recent new
2348 "Z" feature of pcretest means that these can be cut out, making the tests
2349 usable with all link sizes.
2350
2351 5. Implemented Stan Switzer's goto replacement for longjmp() when not using
2352 stack recursion. This gives a massive performance boost under BSD, but just
2353 a small improvement under Linux. However, it saves one field in the frame
2354 in all cases.
2355
2356 6. Added more features from the forthcoming Perl 5.10:
2357
2358 (a) (?-n) (where n is a string of digits) is a relative subroutine or
2359 recursion call. It refers to the nth most recently opened parentheses.
2360
2361 (b) (?+n) is also a relative subroutine call; it refers to the nth next
2362 to be opened parentheses.
2363
2364 (c) Conditions that refer to capturing parentheses can be specified
2365 relatively, for example, (?(-2)... or (?(+3)...
2366
2367 (d) \K resets the start of the current match so that everything before
2368 is not part of it.
2369
2370 (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
2371
2372 (f) \g{name} is another synonym - part of Perl 5.10's unification of
2373 reference syntax.
2374
2375 (g) (?| introduces a group in which the numbering of parentheses in each
2376 alternative starts with the same number.
2377
2378 (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
2379
2380 7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
2381 PCRE_INFO_JCHANGED.
2382
2383 8. A pattern such as (.*(.)?)* caused pcre_exec() to fail by either not
2384 terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
2385 for detecting groups that can match an empty string.
2386
2387 9. A pattern with a very large number of alternatives (more than several
2388 hundred) was running out of internal workspace during the pre-compile
2389 phase, where pcre_compile() figures out how much memory will be needed. A
2390 bit of new cunning has reduced the workspace needed for groups with
2391 alternatives. The 1000-alternative test pattern now uses 12 bytes of
2392 workspace instead of running out of the 4096 that are available.
2393
2394 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
2395
2396 11. Applied patch from Google to remove an optimization that didn't quite work.
2397 The report of the bug said:
2398
2399 pcrecpp::RE("a*").FullMatch("aaa") matches, while
2400 pcrecpp::RE("a*?").FullMatch("aaa") does not, and
2401 pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
2402
2403 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
2404 it matched the wrong number of bytes.
2405
2406
2407 Version 7.1 24-Apr-07
2408 ---------------------
2409
2410 1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
2411 that is more "standard", making use of automake and other Autotools. There
2412 is some re-arrangement of the files and adjustment of comments consequent
2413 on this.
2414
2415 2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
2416 for recursive directory scanning broke on some systems because the files
2417 are not scanned in any specific order and on different systems the order
2418 was different. A call to "sort" has been inserted into RunGrepTest for the
2419 approprate test as a short-term fix. In the longer term there may be an
2420 alternative.
2421
2422 3. I had an email from Eric Raymond about problems translating some of PCRE's
2423 man pages to HTML (despite the fact that I distribute HTML pages, some
2424 people do their own conversions for various reasons). The problems
2425 concerned the use of low-level troff macros .br and .in. I have therefore
2426 removed all such uses from the man pages (some were redundant, some could
2427 be replaced by .nf/.fi pairs). The 132html script that I use to generate
2428 HTML has been updated to handle .nf/.fi and to complain if it encounters
2429 .br or .in.
2430
2431 4. Updated comments in configure.ac that get placed in config.h.in and also
2432 arranged for config.h to be included in the distribution, with the name
2433 config.h.generic, for the benefit of those who have to compile without
2434 Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
2435
2436 5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
2437 Weber: (1) pcre_internal.h was missing some function renames; (2) updated
2438 makevp.bat for the current PCRE, using the additional files
2439 makevp_c.txt, makevp_l.txt, and pcregexp.pas.
2440
2441 6. A Windows user reported a minor discrepancy with test 2, which turned out
2442 to be caused by a trailing space on an input line that had got lost in his
2443 copy. The trailing space was an accident, so I've just removed it.
2444
2445 7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
2446 that is needed.
2447
2448 8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
2449 as "const" (a) because they are and (b) because it helps the PHP
2450 maintainers who have recently made a script to detect big data structures
2451 in the php code that should be moved to the .rodata section. I remembered
2452 to update Builducptable as well, so it won't revert if ucptable.h is ever
2453 re-created.
2454
2455 9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
2456 pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
2457 order to be able to cut out the UTF-8 tables in the latter when UTF-8
2458 support is not required. This saves 1.5-2K of code, which is important in
2459 some applications.
2460
2461 Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
2462 so as not to refer to the tables, even though these functions will never be
2463 called when UTF-8 support is disabled. Otherwise there are problems with a
2464 shared library.
2465
2466 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
2467
2468 (a) It was defining its arguments as char * instead of void *.
2469
2470 (b) It was assuming that all moves were upwards in memory; this was true
2471 a long time ago when I wrote it, but is no longer the case.
2472
2473 The emulated memove() is provided for those environments that have neither
2474 memmove() nor bcopy(). I didn't think anyone used it these days, but that
2475 is clearly not the case, as these two bugs were recently reported.
2476
2477 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
2478 and Detrail to create the HTML documentation, the .txt form of the man
2479 pages, and it removes trailing spaces from listed files. It also creates
2480 pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
2481 case, it wraps all the #defines with #ifndefs. This script should be run
2482 before "make dist".
2483
2484 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
2485 with Unicode property support.
2486
2487 (a) For a maximizing quantifier, if the two different cases of the
2488 character were of different lengths in their UTF-8 codings (there are
2489 some cases like this - I found 11), and the matching function had to
2490 back up over a mixture of the two cases, it incorrectly assumed they
2491 were both the same length.
2492
2493 (b) When PCRE was configured to use the heap rather than the stack for
2494 recursion during matching, it was not correctly preserving the data for
2495 the other case of a UTF-8 character when checking ahead for a match
2496 while processing a minimizing repeat. If the check also involved
2497 matching a wide character, but failed, corruption could cause an
2498 erroneous result when trying to check for a repeat of the original
2499 character.
2500
2501 13. Some tidying changes to the testing mechanism:
2502
2503 (a) The RunTest script now detects the internal link size and whether there
2504 is UTF-8 and UCP support by running ./pcretest -C instead of relying on
2505 values substituted by "configure". (The RunGrepTest script already did
2506 this for UTF-8.) The configure.ac script no longer substitutes the
2507 relevant variables.
2508
2509 (b) The debugging options /B and /D in pcretest show the compiled bytecode
2510 with length and offset values. This means that the output is different
2511 for different internal link sizes. Test 2 is skipped for link sizes
2512 other than 2 because of this, bypassing the problem. Unfortunately,
2513 there was also a test in test 3 (the locale tests) that used /B and
2514 failed for link sizes other than 2. Rather than cut the whole test out,
2515 I have added a new /Z option to pcretest that replaces the length and
2516 offset values with spaces. This is now used to make test 3 independent
2517 of link size. (Test 2 will be tidied up later.)
2518
2519 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
2520 segmentation fault instead of returning the appropriate error message.
2521
2522 15. In multiline mode when the newline sequence was set to "any", the pattern
2523 ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
2524 This doesn't seem right; it now treats the CRLF combination as the line
2525 ending, and so does not match in that case. It's only a pattern such as ^$
2526 that would hit this one: something like ^ABC$ would have failed after \r
2527 and then tried again after \r\n.
2528
2529 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
2530 in an attempt to make files that differ only in their line terminators
2531 compare equal. This works on Linux.
2532
2533 17. Under certain error circumstances pcregrep might try to free random memory
2534 as it exited. This is now fixed, thanks to valgrind.
2535
2536 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
2537 "abc\r\n\r\n", it found an unwanted second match after the second \r. This
2538 was because its rules for how to advance for /g after matching an empty
2539 string at the end of a line did not allow for this case. They now check for
2540 it specially.
2541
2542 20. pcretest is supposed to handle patterns and data of any length, by
2543 extending its buffers when necessary. It was getting this wrong when the
2544 buffer for a data line had to be extended.
2545
2546 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
2547 CRLF as a newline sequence.
2548
2549 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
2550 out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
2551 I have nevertheless tidied it up.
2552
2553 23. Added some casts to kill warnings from HP-UX ia64 compiler.
2554
2555 24. Added a man page for pcre-config.
2556
2557
2558 Version 7.0 19-Dec-06
2559 ---------------------
2560
2561 1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
2562 moving to gcc 4.1.1.
2563
2564 2. The -S option for pcretest uses setrlimit(); I had omitted to #include
2565 sys/time.h, which is documented as needed for this function. It doesn't
2566 seem to matter on Linux, but it showed up on some releases of OS X.
2567
2568 3. It seems that there are systems where bytes whose values are greater than
2569 127 match isprint() in the "C" locale. The "C" locale should be the
2570 default when a C program starts up. In most systems, only ASCII printing
2571 characters match isprint(). This difference caused the output from pcretest
2572 to vary, making some of the tests fail. I have changed pcretest so that:
2573
2574 (a) When it is outputting text in the compiled version of a pattern, bytes
2575 other than 32-126 are always shown as hex escapes.
2576
2577 (b) When it is outputting text that is a matched part of a subject string,
2578 it does the same, unless a different locale has been set for the match
2579 (using the /L modifier). In this case, it uses isprint() to decide.
2580
2581 4. Fixed a major bug that caused incorrect computation of the amount of memory
2582 required for a compiled pattern when options that changed within the
2583 pattern affected the logic of the preliminary scan that determines the
2584 length. The relevant options are -x, and -i in UTF-8 mode. The result was
2585 that the computed length was too small. The symptoms of this bug were
2586 either the PCRE error "internal error: code overflow" from pcre_compile(),
2587 or a glibc crash with a message such as "pcretest: free(): invalid next
2588 size (fast)". Examples of patterns that provoked this bug (shown in
2589 pcretest format) are:
2590
2591 /(?-x: )/x
2592 /(?x)(?-x: \s*#\s*)/
2593 /((?i)[\x{c0}])/8
2594 /(?i:[\x{c0}])/8
2595
2596 HOWEVER: Change 17 below makes this fix obsolete as the memory computation
2597 is now done differently.
2598
2599 5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
2600 wrapper classes; (b) implement a new function in the C++ scanner that is
2601 more efficient than the old way of doing things because it avoids levels of
2602 recursion in the regex matching; (c) add a paragraph to the documentation
2603 for the FullMatch() function.
2604
2605 6. The escape sequence \n was being treated as whatever was defined as
2606 "newline". Not only was this contrary to the documentation, which states
2607 that \n is character 10 (hex 0A), but it also went horribly wrong when
2608 "newline" was defined as CRLF. This has been fixed.
2609
2610 7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
2611 was being set to -1 for the "end of line" case (supposedly a value that no
2612 character can have). Though this value is never used (the check for end of
2613 line is "zero bytes in current character"), it caused compiler complaints.
2614 I've changed it to 0xffffffff.
2615
2616 8. In pcre_version.c, the version string was being built by a sequence of
2617 C macros that, in the event of PCRE_PRERELEASE being defined as an empty
2618 string (as it is for production releases) called a macro with an empty
2619 argument. The C standard says the result of this is undefined. The gcc
2620 compiler treats it as an empty string (which was what was wanted) but it is
2621 reported that Visual C gives an error. The source has been hacked around to
2622 avoid this problem.
2623
2624 9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
2625 builds of pcretest, and changed the call to _setmode() to use _O_BINARY
2626 instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
2627 of them did).
2628
2629 10. Originally, pcretest opened its input and output without "b"; then I was
2630 told that "b" was needed in some environments, so it was added for release
2631 5.0 to both the input and output. (It makes no difference on Unix-like
2632 systems.) Later I was told that it is wrong for the input on Windows. I've
2633 now abstracted the modes into two macros, to make it easier to fiddle with
2634 them, and removed "b" from the input mode under Windows.
2635
2636 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
2637
2638 12. Added -help and --help to pcretest as an official way of being reminded
2639 of the options.
2640
2641 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
2642 and pcrecpp.cc because they annoy compilers at high warning levels.
2643
2644 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
2645
2646 15. Fixed an occurrence of == in configure.ac that should have been = (shell
2647 scripts are not C programs :-) and which was not noticed because it works
2648 on Linux.
2649
2650 16. pcretest is supposed to handle any length of pattern and data line (as one
2651 line or as a continued sequence of lines) by extending its input buffer if
2652 necessary. This feature was broken for very long pattern lines, leading to
2653 a string of junk being passed to pcre_compile() if the pattern was longer
2654 than about 50K.
2655
2656 17. I have done a major re-factoring of the way pcre_compile() computes the
2657 amount of memory needed for a compiled pattern. Previously, there was code
2658 that made a preliminary scan of the pattern in order to do this. That was
2659 OK when PCRE was new, but as the facilities have expanded, it has become
2660 harder and harder to keep it in step with the real compile phase, and there
2661 have been a number of bugs (see for example, 4 above). I have now found a
2662 cunning way of running the real compile function in a "fake" mode that
2663 enables it to compute how much memory it would need, while actually only
2664 ever using a few hundred bytes of working memory and without too many
2665 tests of the mode. This should make future maintenance and development
2666 easier. A side effect of this work is that the limit of 200 on the nesting
2667 depth of parentheses has been removed (though this was never a serious
2668 limitation, I suspect). However, there is a downside: pcre_compile() now
2669 runs more slowly than before (30% or more, depending on the pattern). I
2670 hope this isn't a big issue. There is no effect on runtime performance.
2671
2672 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
2673 newline (only possible for the last line of a file) and it was a
2674 pattern that set a locale (followed by /Lsomething), pcretest crashed.
2675
2676 19. Added additional timing features to pcretest. (1) The -tm option now times
2677 matching only, not compiling. (2) Both -t and -tm can be followed, as a
2678 separate command line item, by a number that specifies the number of
2679 repeats to use when timing. The default is 50000; this gives better
2680 precision, but takes uncomfortably long for very large patterns.
2681
2682 20. Extended pcre_study() to be more clever in cases where a branch of a
2683 subpattern has no definite first character. For example, (a*|b*)[cd] would
2684 previously give no result from pcre_study(). Now it recognizes that the
2685 first character must be a, b, c, or d.
2686
2687 21. There was an incorrect error "recursive call could loop indefinitely" if
2688 a subpattern (or the entire pattern) that was being tested for matching an
2689 empty string contained only one non-empty item after a nested subpattern.
2690 For example, the pattern (?>\x{100}*)\d(?R) provoked this error
2691 incorrectly, because the \d was being skipped in the check.
2692
2693 22. The pcretest program now has a new pattern option /B and a command line
2694 option -b, which is equivalent to adding /B to every pattern. This causes
2695 it to show the compiled bytecode, without the additional information that
2696 -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
2697 is the same as /B/I).
2698
2699 23. A new optimization is now able automatically to treat some sequences such
2700 as a*b as a*+b. More specifically, if something simple (such as a character
2701 or a simple class like \d) has an unlimited quantifier, and is followed by
2702 something that cannot possibly match the quantified thing, the quantifier
2703 is automatically "possessified".
2704
2705 24. A recursive reference to a subpattern whose number was greater than 39
2706 went wrong under certain circumstances in UTF-8 mode. This bug could also
2707 have affected the operation of pcre_study().
2708
2709 25. Realized that a little bit of performance could be had by replacing
2710 (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
2711
2712 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
2713
2714 27. Possessive quantifiers such as a++ were previously implemented by turning
2715 them into atomic groups such as ($>a+). Now they have their own opcodes,
2716 which improves performance. This includes the automatically created ones
2717 from 23 above.
2718
2719 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
2720 lookahead was broken if it was not anchored. PCRE was mistakenly expecting
2721 the first matched character to be a colon. This applied both to named and
2722 numbered groups.
2723
2724 29. The ucpinternal.h header file was missing its idempotency #ifdef.
2725
2726 30. I was sent a "project" file called libpcre.a.dev which I understand makes
2727 building PCRE on Windows easier, so I have included it in the distribution.
2728
2729 31. There is now a check in pcretest against a ridiculously large number being
2730 returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
2731 loop, the loop is abandoned.
2732
2733 32. Forward references to subpatterns in conditions such as (?(2)...) where
2734 subpattern 2 is defined later cause pcre_compile() to search forwards in
2735 the pattern for the relevant set of parentheses. This search went wrong
2736 when there were unescaped parentheses in a character class, parentheses
2737 escaped with \Q...\E, or parentheses in a #-comment in /x mode.
2738
2739 33. "Subroutine" calls and backreferences were previously restricted to
2740 referencing subpatterns earlier in the regex. This restriction has now
2741 been removed.
2742
2743 34. Added a number of extra features that are going to be in Perl 5.10. On the
2744 whole, these are just syntactic alternatives for features that PCRE had
2745 previously implemented using the Python syntax or my own invention. The
2746 other formats are all retained for compatibility.
2747
2748 (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
2749 as (?P<name>...). The new forms, as well as being in Perl 5.10, are
2750 also .NET compatible.
2751
2752 (b) A recursion or subroutine call to a named group can now be defined as
2753 (?&name) as well as (?P>name).
2754
2755 (c) A backreference to a named group can now be defined as \k<name> or
2756 \k'name' as well as (?P=name). The new forms, as well as being in Perl
2757 5.10, are also .NET compatible.
2758
2759 (d) A conditional reference to a named group can now use the syntax
2760 (?(<name>) or (?('name') as well as (?(name).
2761
2762 (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
2763 groups (named and numbered) that are never evaluated inline, but can be
2764 called as "subroutines" from elsewhere. In effect, the DEFINE condition
2765 is always false. There may be only one alternative in such a group.
2766
2767 (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
2768 as the simple (?(R). The condition is true only if the most recent
2769 recursion is that of the given number or name. It does not search out
2770 through the entire recursion stack.
2771
2772 (g) The escape \gN or \g{N} has been added, where N is a positive or
2773 negative number, specifying an absolute or relative reference.
2774
2775 35. Tidied to get rid of some further signed/unsigned compiler warnings and
2776 some "unreachable code" warnings.
2777
2778 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
2779 things, this adds five new scripts.
2780
2781 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
2782 There were also incompatibilities regarding the handling of \Q..\E inside
2783 character classes, for example with patterns like [\Qa\E-\Qz\E] where the
2784 hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
2785
2786 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
2787 matches an empty string, and forcibly breaks the loop. There were bugs in
2788 this code in non-simple cases. For a pattern such as ^(a()*)* matched
2789 against aaaa the result was just "a" rather than "aaaa", for example. Two
2790 separate and independent bugs (that affected different cases) have been
2791 fixed.
2792
2793 39. Refactored the code to abolish the use of different opcodes for small
2794 capturing bracket numbers. This is a tidy that I avoided doing when I
2795 removed the limit on the number of capturing brackets for 3.5 back in 2001.
2796 The new approach is not only tidier, it makes it possible to reduce the
2797 memory needed to fix the previous bug (38).
2798
2799 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
2800 sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
2801 processing dot, circumflex, or dollar metacharacters, or #-comments in /x
2802 mode.
2803
2804 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
2805 report.
2806
2807 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
2808 copy construction and assignment in the C++ wrapper.
2809
2810 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
2811 couple of bugs that could have given wrong results in the "--newline=crlf"
2812 case.
2813
2814 44. Added a number of casts and did some reorganization of signed/unsigned int
2815 variables following suggestions from Dair Grant. Also renamed the variable
2816 "this" as "item" because it is a C++ keyword.
2817
2818 45. Arranged for dftables to add
2819
2820 #include "pcre_internal.h"
2821
2822 to pcre_chartables.c because without it, gcc 4.x may remove the array
2823 definition from the final binary if PCRE is built into a static library and
2824 dead code stripping is activated.
2825
2826 46. For an unanchored pattern, if a match attempt fails at the start of a
2827 newline sequence, and the newline setting is CRLF or ANY, and the next two
2828 characters are CRLF, advance by two characters instead of one.
2829
2830
2831 Version 6.7 04-Jul-06
2832 ---------------------
2833
2834 1. In order to handle tests when input lines are enormously long, pcretest has
2835 been re-factored so that it automatically extends its buffers when
2836 necessary. The code is crude, but this _is_ just a test program. The
2837 default size has been increased from 32K to 50K.
2838
2839 2. The code in pcre_study() was using the value of the re argument before
2840 testing it for NULL. (Of course, in any sensible call of the function, it
2841 won't be NULL.)
2842
2843 3. The memmove() emulation function in pcre_internal.h, which is used on
2844 systems that lack both memmove() and bcopy() - that is, hardly ever -
2845 was missing a "static" storage class specifier.
2846
2847 4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
2848 containing an extended class (one that cannot be represented by a bitmap
2849 because it contains high-valued characters or Unicode property items, e.g.
2850 [\pZ]). Almost always one would set UTF-8 mode when processing such a
2851 pattern, but PCRE should not loop if you do not (it no longer does).
2852 [Detail: two cases were found: (a) a repeated subpattern containing an
2853 extended class; (b) a recursive reference to a subpattern that followed a
2854 previous extended class. It wasn't skipping over the extended class
2855 correctly when UTF-8 mode was not set.]
2856
2857 5. A negated single-character class was not being recognized as fixed-length
2858 in lookbehind assertions such as (?<=[^f]), leading to an incorrect
2859 compile error "lookbehind assertion is not fixed length".
2860
2861 6. The RunPerlTest auxiliary script was showing an unexpected difference
2862 between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
2863 write a Perl script that can interpret lines of an input file either as
2864 byte characters or as UTF-8, which is what "perltest" was being required to
2865 do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
2866 can't do is switch easily at run time between having the "use utf8;" pragma
2867 or not. In the end, I fudged it by using the RunPerlTest script to insert
2868 "use utf8;" explicitly for the UTF-8 tests.
2869
2870 7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
2871 the end of the subject string, contrary to the documentation and to what
2872 Perl does. This was true of both matching functions. Now it matches only at
2873 the start of the subject and immediately after *internal* newlines.
2874
2875 8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
2876 a pointer to an int instead of a pointer to an unsigned long int. This
2877 caused problems on 64-bit systems.
2878
2879 9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
2880 instance of the 'standard' template library not being so standard".
2881
2882 10. There was no check on the number of named subpatterns nor the maximum
2883 length of a subpattern name. The product of these values is used to compute
2884 the size of the memory block for a compiled pattern. By supplying a very
2885 long subpattern name and a large number of named subpatterns, the size
2886 computation could be caused to overflow. This is now prevented by limiting
2887 the length of names to 32 characters, and the number of named subpatterns
2888 to 10,000.
2889
2890 11. Subpatterns that are repeated with specific counts have to be replicated in
2891 the compiled pattern. The size of memory for this was computed from the
2892 length of the subpattern and the repeat count. The latter is limited to
2893 65535, but there was no limit on the former, meaning that integer overflow
2894 could in principle occur. The compiled length of a repeated subpattern is
2895 now limited to 30,000 bytes in order to prevent this.
2896
2897 12. Added the optional facility to have named substrings with the same name.
2898
2899 13. Added the ability to use a named substring as a condition, using the
2900 Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
2901 are numbers (not recommended). Forward references are permitted.
2902
2903 14. Added forward references in named backreferences (if you see what I mean).
2904
2905 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
2906 pattern could run off the end of the subject. For example, the pattern
2907 "(?s)(.{1,5})"8 did this with the subject "ab".
2908
2909 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
2910 PCRE_CASELESS was set when matching characters that were quantified with ?
2911 or *.
2912
2913 17. A character class other than a single negated character that had a minimum
2914 but no maximum quantifier - for example [ab]{6,} - was not handled
2915 correctly by pce_dfa_exec(). It would match only one character.
2916
2917 18. A valid (though odd) pattern that looked like a POSIX character
2918 class but used an invalid character after [ (for example [[,abc,]]) caused
2919 pcre_compile() to give the error "Failed: internal error: code overflow" or
2920 in some cases to crash with a glibc free() error. This could even happen if
2921 the pattern terminated after [[ but there just happened to be a sequence of
2922 letters, a binary zero, and a closing ] in the memory that followed.
2923
2924 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
2925 over the years. Originally (before any Unicode support), just the bottom 8
2926 bits were taken. Thus, for example, \500 really meant \100. Nowadays the
2927 output from "man perlunicode" includes this:
2928
2929 The regular expression compiler produces polymorphic opcodes. That
2930 is, the pattern adapts to the data and automatically switches to
2931 the Unicode character scheme when presented with Unicode data--or
2932 instead uses a traditional byte scheme when presented with byte
2933 data.
2934
2935 Sadly, a wide octal escape does not cause a switch, and in a string with
2936 no other multibyte characters, these octal escapes are treated as before.
2937 Thus, in Perl, the pattern /\500/ actually matches \100 but the pattern
2938 /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
2939 Unicode string.
2940
2941 I have not perpetrated such confusion in PCRE. Up till now, it took just
2942 the bottom 8 bits, as in old Perl. I have now made octal escapes with
2943 values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
2944 translate to the appropriate multibyte character.
2945
2946 29. Applied some refactoring to reduce the number of warnings from Microsoft
2947 and Borland compilers. This has included removing the fudge introduced
2948 seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
2949 a warning about an unused variable.
2950
2951 21. PCRE has not included VT (character 0x0b) in the set of whitespace
2952 characters since release 4.0, because Perl (from release 5.004) does not.
2953 [Or at least, is documented not to: some releases seem to be in conflict
2954 with the documentation.] However, when a pattern was studied with
2955 pcre_study() and all its branches started with \s, PCRE still included VT
2956 as a possible starting character. Of course, this did no harm; it just
2957 caused an unnecessary match attempt.
2958
2959 22. Removed a now-redundant internal flag bit that recorded the fact that case
2960 dependency changed within the pattern. This was once needed for "required
2961 byte" processing, but is no longer used. This recovers a now-scarce options
2962 bit. Also moved the least significant internal flag bit to the most-
2963 significant bit of the word, which was not previously used (hangover from
2964 the days when it was an int rather than a uint) to free up another bit for
2965 the future.
2966
2967 23. Added support for CRLF line endings as well as CR and LF. As well as the
2968 default being selectable at build time, it can now be changed at runtime
2969 via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
2970 specify that it is scanning data with non-default line endings.
2971
2972 24. Changed the definition of CXXLINK to make it agree with the definition of
2973 LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
2974
2975 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
2976 recursions. This makes a big different to stack usage for some patterns.
2977
2978 26. If a subpattern containing a named recursion or subroutine reference such
2979 as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
2980 the space required for the compiled pattern went wrong and gave too small a
2981 value. Depending on the environment, this could lead to "Failed: internal
2982 error: code overflow at offset 49" or "glibc detected double free or
2983 corruption" errors.
2984
2985 27. Applied patches from Google (a) to support the new newline modes and (b) to
2986 advance over multibyte UTF-8 characters in GlobalReplace.
2987
2988 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
2989 difference for some implementation of PCRE in some Windows version.
2990
2991 29. Added some extra testing facilities to pcretest:
2992
2993 \q<number> in a data line sets the "match limit" value
2994 \Q<number> in a data line sets the "match recursion limt" value
2995 -S <number> sets the stack size, where <number> is in megabytes
2996
2997 The -S option isn't available for Windows.
2998
2999
3000 Version 6.6 06-Feb-06
3001 ---------------------
3002
3003 1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
3004 in pcreposix.h. I have copied the definition from pcre.h.
3005
3006 2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
3007 because pcre.h is no longer a built file.
3008
3009 3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
3010 not normally included in the compiled code.
3011
3012
3013 Version 6.5 01-Feb-06
3014 ---------------------
3015
3016 1. When using the partial match feature with pcre_dfa_exec(), it was not
3017 anchoring the second and subsequent partial matches at the new starting
3018 point. This could lead to incorrect results. For example, with the pattern
3019 /1234/, partially matching against "123" and then "a4" gave a match.
3020
3021 2. Changes to pcregrep:
3022
3023 (a) All non-match returns from pcre_exec() were being treated as failures
3024 to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
3025 error message is output. Some extra information is given for the
3026 PCRE_ERROR_MATCHLIMIT and PCRE_ERROR_RECURSIONLIMIT errors, which are
3027 probably the only errors that are likely to be caused by users (by
3028 specifying a regex that has nested indefinite repeats, for instance).
3029 If there are more than 20 of these errors, pcregrep is abandoned.
3030
3031 (b) A binary zero was treated as data while matching, but terminated the
3032 output line if it was written out. This has been fixed: binary zeroes
3033 are now no different to any other data bytes.
3034
3035 (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
3036 used to set a locale for matching. The --locale=xxxx long option has
3037 been added (no short equivalent) to specify a locale explicitly on the
3038 pcregrep command, overriding the environment variables.
3039
3040 (d) When -B was used with -n, some line numbers in the output were one less
3041 than they should have been.
3042
3043 (e) Added the -o (--only-matching) option.
3044
3045 (f) If -A or -C was used with -c (count only), some lines of context were
3046 accidentally printed for the final match.
3047
3048 (g) Added the -H (--with-filename) option.
3049
3050 (h) The combination of options -rh failed to suppress file names for files
3051 that were found from directory arguments.
3052
3053 (i) Added the -D (--devices) and -d (--directories) options.
3054
3055 (j) Added the -F (--fixed-strings) option.
3056
3057 (k) Allow "-" to be used as a file name for -f as well as for a data file.
3058
3059 (l) Added the --colo(u)r option.
3060
3061 (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
3062 is not present by default.
3063
3064 3. A nasty bug was discovered in the handling of recursive patterns, that is,
3065 items such as (?R) or (?1), when the recursion could match a number of
3066 alternatives. If it matched one of the alternatives, but subsequently,
3067 outside the recursion, there was a failure, the code tried to back up into
3068 the recursion. However, because of the way PCRE is implemented, this is not
3069 possible, and the result was an incorrect result from the match.
3070
3071 In order to prevent this happening, the specification of recursion has
3072 been changed so that all such subpatterns are automatically treated as
3073 atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
3074
3075 4. I had overlooked the fact that, in some locales, there are characters for
3076 which isalpha() is true but neither isupper() nor islower() are true. In
3077 the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
3078 and ordfeminine) are like this. This affected the treatment of \w and \W
3079 when they appeared in character classes, but not when they appeared outside
3080 a character class. The bit map for "word" characters is now created
3081 separately from the results of isalnum() instead of just taking it from the
3082 upper, lower, and digit maps. (Plus the underscore character, of course.)
3083
3084 5. The above bug also affected the handling of POSIX character classes such as
3085 [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
3086 permanent tables. Instead, the bit maps for such a class were previously
3087 created as the appropriate unions of the upper, lower, and digit bitmaps.
3088 Now they are created by subtraction from the [[:word:]] class, which has
3089 its own bitmap.
3090
3091 6. The [[:blank:]] character class matches horizontal, but not vertical space.
3092 It is created by subtracting the vertical space characters (\x09, \x0a,
3093 \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
3094 subtraction was done in the overall bitmap for a character class, meaning
3095 that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
3096 be recognized. This bug has been fixed.
3097
3098 7. Patches from the folks at Google:
3099
3100 (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
3101 real life, but is still worth protecting against".
3102
3103 (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
3104 regular expressions".
3105
3106 (c) pcre_scanner.cc: avoid use of std::count() because not all systems
3107 have it.
3108
3109 (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
3110 "configure" and the latter not, in order to fix a problem somebody had
3111 with compiling the Arg class on HP-UX.
3112
3113 (e) Improve the error-handling of the C++ wrapper a little bit.
3114
3115 (f) New tests for checking recursion limiting.
3116
3117 8. The pcre_memmove() function, which is used only if the environment does not
3118 have a standard memmove() function (and is therefore rarely compiled),
3119 contained two bugs: (a) use of int instead of size_t, and (b) it was not
3120 returning a result (though PCRE never actually uses the result).
3121
3122 9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
3123 large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
3124 returned instead of calling malloc() with an overflowing number that would
3125 most likely cause subsequent chaos.
3126
3127 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
3128
3129 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
3130 with this option is matched, the nmatch and pmatch options of regexec() are
3131 ignored.
3132
3133 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
3134 provided in case anyone wants to the the POSIX interface with UTF-8
3135 strings.
3136
3137 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
3138 C++ linking (needed for some HP-UX environments).
3139
3140 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
3141 (unused parameter) and in the pcre_printint() function (omitted "default"
3142 switch label when the default is to do nothing).
3143
3144 15. Added some code to make it possible, when PCRE is compiled as a C++
3145 library, to replace subject pointers for pcre_exec() with a smart pointer
3146 class, thus making it possible to process discontinuous strings.
3147
3148 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
3149 much the same function. They were added by different people who were trying
3150 to make PCRE easy to compile on non-Unix systems. It has been suggested
3151 that PCRE_EXPORT be abolished now that there is more automatic apparatus
3152 for compiling on Windows systems. I have therefore replaced it with
3153 PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
3154 defaults to "extern" for C or "extern C" for C++, which works fine on
3155 Unix-like systems. It is now possible to override the value of PCRE_DATA_
3156 SCOPE with something explicit in config.h. In addition:
3157
3158 (a) pcreposix.h still had just "extern" instead of either of these macros;
3159 I have replaced it with PCRE_DATA_SCOPE.
3160
3161 (b) Functions such as _pcre_xclass(), which are internal to the library,
3162 but external in the C sense, all had PCRE_EXPORT in their definitions.
3163 This is apparently wrong for the Windows case, so I have removed it.
3164 (It makes no difference on Unix-like systems.)
3165
3166 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
3167 of recursive calls to match(). This is different to MATCH_LIMIT because
3168 that limits the total number of calls to match(), not all of which increase
3169 the depth of recursion. Limiting the recursion depth limits the amount of
3170 stack (or heap if NO_RECURSE is set) that is used. The default can be set
3171 when PCRE is compiled, and changed at run time. A patch from Google adds
3172 this functionality to the C++ interface.
3173
3174 18. Changes to the handling of Unicode character properties:
3175
3176 (a) Updated the table to Unicode 4.1.0.
3177
3178 (b) Recognize characters that are not in the table as "Cn" (undefined).
3179
3180 (c) I revised the way the table is implemented to a much improved format
3181 which includes recognition of ranges. It now supports the ranges that
3182 are defined in UnicodeData.txt, and it also amalgamates other
3183 characters into ranges. This has reduced the number of entries in the
3184 table from around 16,000 to around 3,000, thus reducing its size
3185 considerably. I realized I did not need to use a tree structure after
3186 all - a binary chop search is just as efficient. Having reduced the
3187 number of entries, I extended their size from 6 bytes to 8 bytes to
3188 allow for more data.
3189
3190 (d) Added support for Unicode script names via properties such as \p{Han}.
3191
3192 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
3193 matching that character.
3194
3195 20. When matching a repeated Unicode property with a minimum greater than zero,
3196 (for example \pL{2,}), PCRE could look past the end of the subject if it
3197 reached it while seeking the minimum number of characters. This could
3198 happen only if some of the characters were more than one byte long, because
3199 there is a check for at least the minimum number of bytes.
3200
3201 21. Refactored the implementation of \p and \P so as to be more general, to
3202 allow for more different types of property in future. This has changed the
3203 compiled form incompatibly. Anybody with saved compiled patterns that use
3204 \p or \P will have to recompile them.
3205
3206 22. Added "Any" and "L&" to the supported property types.
3207
3208 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
3209 but give a compile time error if the value is greater than 0xff.
3210
3211 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
3212 accidentally not being installed or uninstalled.
3213
3214 25. The pcre.h file was built from pcre.h.in, but the only changes that were
3215 made were to insert the current release number. This seemed silly, because
3216 it made things harder for people building PCRE on systems that don't run
3217 "configure". I have turned pcre.h into a distributed file, no longer built
3218 by "configure", with the version identification directly included. There is
3219 no longer a pcre.h.in file.
3220
3221 However, this change necessitated a change to the pcre-config script as
3222 well. It is built from pcre-config.in, and one of the substitutions was the
3223 release number. I have updated configure.ac so that ./configure now finds
3224 the release number by grepping pcre.h.
3225
3226 26. Added the ability to run the tests under valgrind.
3227
3228
3229 Version 6.4 05-Sep-05
3230 ---------------------
3231
3232 1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
3233 "--" to be printed when multiple files were scanned, even when none of the
3234 -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
3235 consider it to be a bug, and have restored the previous behaviour.
3236
3237 2. A couple of code tidies to get rid of compiler warnings.
3238
3239 3. The pcretest program used to cheat by referring to symbols in the library
3240 whose names begin with _pcre_. These are internal symbols that are not
3241 really supposed to be visible externally, and in some environments it is
3242 possible to suppress them. The cheating is now confined to including
3243 certain files from the library's source, which is a bit cleaner.
3244
3245 4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
3246 file's purpose clearer.
3247
3248 5. Reorganized pcre_ucp_findchar().
3249
3250
3251 Version 6.3 15-Aug-05
3252 ---------------------
3253
3254 1. The file libpcre.pc.in did not have general read permission in the tarball.
3255
3256 2. There were some problems when building without C++ support:
3257
3258 (a) If C++ support was not built, "make install" and "make test" still
3259 tried to test it.
3260
3261 (b) There were problems when the value of CXX was explicitly set. Some
3262 changes have been made to try to fix these, and ...
3263
3264 (c) --disable-cpp can now be used to explicitly disable C++ support.
3265
3266 (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
3267 backslash in a target when C++ was disabled. This confuses some
3268 versions of "make", apparently. Using an intermediate variable solves
3269 this. (Same for CPP_LOBJ.)
3270
3271 3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
3272 (non-Windows) now includes $(CFLAGS) because these flags are sometimes
3273 necessary on certain architectures.
3274
3275 4. Added a setting of -export-symbols-regex to the link command to remove
3276 those symbols that are exported in the C sense, but actually are local
3277 within the library, and not documented. Their names all begin with
3278 "_pcre_". This is not a perfect job, because (a) we have to except some
3279 symbols that pcretest ("illegally") uses, and (b) the facility isn't always
3280 available (and never for static libraries). I have made a note to try to
3281 find a way round (a) in the future.
3282
3283
3284 Version 6.2 01-Aug-05
3285 ---------------------
3286
3287 1. There was no test for integer overflow of quantifier values. A construction
3288 such as {1111111111111111} would give undefined results. What is worse, if
3289 a minimum quantifier for a parenthesized subpattern overflowed and became
3290 negative, the calculation of the memory size went wrong. This could have
3291 led to memory overwriting.
3292
3293 2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
3294
3295 3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
3296 operating environments where this matters.
3297
3298 4. Applied Giuseppe Maxia's patch to add additional features for controlling
3299 PCRE options from within the C++ wrapper.
3300
3301 5. Named capturing subpatterns were not being correctly counted when a pattern
3302 was compiled. This caused two problems: (a) If there were more than 100
3303 such subpatterns, the calculation of the memory needed for the whole
3304 compiled pattern went wrong, leading to an overflow error. (b) Numerical
3305 back references of the form \12, where the number was greater than 9, were
3306 not recognized as back references, even though there were sufficient
3307 previous subpatterns.
3308
3309 6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
3310 versions of gcc, e.g. 2.95.4.
3311
3312
3313 Version 6.1 21-Jun-05
3314 ---------------------
3315
3316 1. There was one reference to the variable "posix" in pcretest.c that was not
3317 surrounded by "#if !defined NOPOSIX".
3318
3319 2. Make it possible to compile pcretest without DFA support, UTF8 support, or
3320 the cross-check on the old pcre_info() function, for the benefit of the
3321 cut-down version of PCRE that is currently imported into Exim.
3322
3323 3. A (silly) pattern starting with (?i)(?-i) caused an internal space
3324 allocation error. I've done the easy fix, which wastes 2 bytes for sensible
3325 patterns that start (?i) but I don't think that matters. The use of (?i) is
3326 just an example; this all applies to the other options as well.
3327
3328 4. Since libtool seems to echo the compile commands it is issuing, the output
3329 from "make" can be reduced a bit by putting "@" in front of each libtool
3330 compile command.
3331
3332 5. Patch from the folks at Google for configure.in to be a bit more thorough
3333 in checking for a suitable C++ installation before trying to compile the
3334 C++ stuff. This should fix a reported problem when a compiler was present,
3335 but no suitable headers.
3336
3337 6. The man pages all had just "PCRE" as their title. I have changed them to
3338 be the relevant file name. I have also arranged that these names are
3339 retained in the file doc/pcre.txt, which is a concatenation in text format
3340 of all the man pages except the little individual ones for each function.
3341
3342 7. The NON-UNIX-USE file had not been updated for the different set of source
3343 files that come with release 6. I also added a few comments about the C++
3344 wrapper.
3345
3346
3347 Version 6.0 07-Jun-05
3348 ---------------------
3349
3350 1. Some minor internal re-organization to help with my DFA experiments.
3351
3352 2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
3353 didn't matter for the library itself when fully configured, but did matter
3354 when compiling without UCP support, or within Exim, where the ucp files are
3355 not imported.
3356
3357 3. Refactoring of the library code to split up the various functions into
3358 different source modules. The addition of the new DFA matching code (see
3359 below) to a single monolithic source would have made it really too
3360 unwieldy, quite apart from causing all the code to be include in a
3361 statically linked application, when only some functions are used. This is
3362 relevant even without the DFA addition now that patterns can be compiled in
3363 one application and matched in another.
3364
3365 The downside of splitting up is that there have to be some external
3366 functions and data tables that are used internally in different modules of
3367 the library but which are not part of the API. These have all had their
3368 names changed to start with "_pcre_" so that they are unlikely to clash
3369 with other external names.
3370
3371 4. Added an alternate matching function, pcre_dfa_exec(), which matches using
3372 a different (DFA) algorithm. Although it is slower than the original
3373 function, it does have some advantages for certain types of matching
3374 problem.
3375
3376 5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
3377 including restarting after a partial match.
3378
3379 6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
3380 defined when compiling for Windows was sent to me. I have put it into the
3381 code, though I have no means of testing or verifying it.
3382
3383 7. Added the pcre_refcount() auxiliary function.
3384
3385 8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
3386 match before or at the first newline in the subject string. In pcretest,
3387 the /f option on a pattern can be used to set this.
3388
3389 9. A repeated \w when used in UTF-8 mode with characters greater than 256
3390 would behave wrongly. This has been present in PCRE since release 4.0.
3391
3392 10. A number of changes to the pcregrep command:
3393
3394 (a) Refactored how -x works; insert ^(...)$ instead of setting
3395 PCRE_ANCHORED and checking the length, in preparation for adding
3396 something similar for -w.
3397
3398 (b) Added the -w (match as a word) option.
3399
3400 (c) Refactored the way lines are read and buffered so as to have more
3401 than one at a time available.
3402
3403 (d) Implemented a pcregrep test script.
3404
3405 (e) Added the -M (multiline match) option. This allows patterns to match
3406 over several lines of the subject. The buffering ensures that at least
3407 8K, or the rest of the document (whichever is the shorter) is available
3408 for matching (and similarly the previous 8K for lookbehind assertions).
3409
3410 (f) Changed the --help output so that it now says
3411
3412 -w, --word-regex(p)
3413
3414 instead of two lines, one with "regex" and the other with "regexp"
3415 because that confused at least one person since the short forms are the
3416 same. (This required a bit of code, as the output is generated
3417 automatically from a table. It wasn't just a text change.)
3418
3419 (g) -- can be used to terminate pcregrep options if the next thing isn't an
3420 option but starts with a hyphen. Could be a pattern or a path name
3421 starting with a hyphen, for instance.
3422
3423 (h) "-" can be given as a file name to represent stdin.
3424
3425 (i) When file names are being printed, "(standard input)" is used for
3426 the standard input, for compatibility with GNU grep. Previously
3427 "<stdin>" was used.
3428
3429 (j) The option --label=xxx can be used to supply a name to be used for
3430 stdin when file names are being printed. There is no short form.
3431
3432 (k) Re-factored the options decoding logic because we are going to add
3433 two more options that take data. Such options can now be given in four
3434 different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
3435
3436 (l) Added the -A, -B, and -C options for requesting that lines of context
3437 around matches be printed.
3438
3439 (m) Added the -L option to print the names of files that do not contain
3440 any matching lines, that is, the complement of -l.
3441
3442 (n) The return code is 2 if any file cannot be opened, but pcregrep does
3443 continue to scan other files.
3444
3445 (o) The -s option was incorrectly implemented. For compatibility with other
3446 greps, it now suppresses the error message for a non-existent or non-
3447 accessible file (but not the return code). There is a new option called
3448 -q that suppresses the output of matching lines, which was what -s was
3449 previously doing.
3450
3451 (p) Added --include and --exclude options to specify files for inclusion
3452 and exclusion when recursing.
3453
3454 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
3455 Hopefully, it now does.
3456
3457 12. Missing cast in pcre_study().
3458
3459 13. Added an "uninstall" target to the makefile.
3460
3461 14. Replaced "extern" in the function prototypes in Makefile.in with
3462 "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
3463 world, but is set differently for Windows.
3464
3465 15. Added a second compiling function called pcre_compile2(). The only
3466 difference is that it has an extra argument, which is a pointer to an
3467 integer error code. When there is a compile-time failure, this is set
3468 non-zero, in addition to the error test pointer being set to point to an
3469 error message. The new argument may be NULL if no error number is required
3470 (but then you may as well call pcre_compile(), which is now just a
3471 wrapper). This facility is provided because some applications need a
3472 numeric error indication, but it has also enabled me to tidy up the way
3473 compile-time errors are handled in the POSIX wrapper.
3474
3475 16. Added VPATH=.libs to the makefile; this should help when building with one
3476 prefix path and installing with another. (Or so I'm told by someone who
3477 knows more about this stuff than I do.)
3478
3479 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
3480 passes PCRE_DOTALL to the pcre_compile() function, making the "." character
3481 match everything, including newlines. This is not POSIX-compatible, but
3482 somebody wanted the feature. From pcretest it can be activated by using
3483 both the P and the s flags.
3484
3485 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
3486
3487 19. libpcre.pc was being incorrectly installed as executable.
3488
3489 20. A couple of places in pcretest check for end-of-line by looking for '\n';
3490 it now also looks for '\r' so that it will work unmodified on Windows.
3491
3492 21. Added Google's contributed C++ wrapper to the distribution.
3493
3494 22. Added some untidy missing memory free() calls in pcretest, to keep
3495 Electric Fence happy when testing.
3496
3497
3498
3499 Version 5.0 13-Sep-04
3500 ---------------------
3501
3502 1. Internal change: literal characters are no longer packed up into items
3503 containing multiple characters in a single byte-string. Each character
3504 is now matched using a separate opcode. However, there may be more than one
3505 byte in the character in UTF-8 mode.
3506
3507 2. The pcre_callout_block structure has two new fields: pattern_position and
3508 next_item_length. These contain the offset in the pattern to the next match
3509 item, and its length, respectively.
3510
3511 3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
3512 insertion of callouts before each pattern item. Added the /C option to
3513 pcretest to make use of this.
3514
3515 4. On the advice of a Windows user, the lines
3516
3517 #if defined(_WIN32) || defined(WIN32)
3518 _setmode( _fileno( stdout ), 0x8000 );
3519 #endif /* defined(_WIN32) || defined(WIN32) */
3520
3521 have been added to the source of pcretest. This apparently does useful
3522 magic in relation to line terminators.
3523
3524 5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
3525 for the benefit of those environments where the "b" makes a difference.
3526
3527 6. The icc compiler has the same options as gcc, but "configure" doesn't seem
3528 to know about it. I have put a hack into configure.in that adds in code
3529 to set GCC=yes if CC=icc. This seems to end up at a point in the
3530 generated configure script that is early enough to affect the setting of
3531 compiler options, which is what is needed, but I have no means of testing
3532 whether it really works. (The user who reported this had patched the
3533 generated configure script, which of course I cannot do.)
3534
3535 LATER: After change 22 below (new libtool files), the configure script
3536 seems to know about icc (and also ecc). Therefore, I have commented out
3537 this hack in configure.in.
3538
3539 7. Added support for pkg-config (2 patches were sent in).
3540
3541 8. Negated POSIX character classes that used a combination of internal tables
3542 were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
3543 [[:^ascii]]. Typically, they would match almost any characters. The other
3544 POSIX classes were not broken in this way.
3545
3546 9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
3547 to find the match, as PCRE was deluded into thinking that the match had to
3548 start at the start point or following a newline. The same bug applied to
3549 patterns with negative forward assertions or any backward assertions
3550 preceding ".*" at the start, unless the pattern required a fixed first
3551 character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
3552
3553 10. In UTF-8 mode, when moving forwards in the subject after a failed match
3554 starting at the last subject character, bytes beyond the end of the subject
3555 string were read.
3556
3557 11. Renamed the variable "class" as "classbits" to make life easier for C++
3558 users. (Previously there was a macro definition, but it apparently wasn't
3559 enough.)
3560
3561 12. Added the new field "tables" to the extra data so that tables can be passed
3562 in at exec time, or the internal tables can be re-selected. This allows
3563 a compiled regex to be saved and re-used at a later time by a different
3564 program that might have everything at different addresses.
3565
3566 13. Modified the pcre-config script so that, when run on Solaris, it shows a
3567 -R library as well as a -L library.
3568
3569 14. The debugging options of pcretest (-d on the command line or D on a
3570 pattern) showed incorrect output for anything following an extended class
3571 that contained multibyte characters and which was followed by a quantifier.
3572
3573 15. Added optional support for general category Unicode character properties
3574 via the \p, \P, and \X escapes. Unicode property support implies UTF-8
3575 support. It adds about 90K to the size of the library. The meanings of the
3576 inbuilt class escapes such as \d and \s have NOT been changed.
3577
3578 16. Updated pcredemo.c to include calls to free() to release the memory for the
3579 compiled pattern.
3580
3581 17. The generated file chartables.c was being created in the source directory
3582 instead of in the building directory. This caused the build to fail if the
3583 source directory was different from the building directory, and was
3584 read-only.
3585
3586 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
3587 file. No doubt somebody will tell me if they don't make sense... Also added
3588 Dan Mooney's comments about building on OpenVMS.
3589
3590 19. Added support for partial matching via the PCRE_PARTIAL option for
3591 pcre_exec() and the \P data escape in pcretest.
3592
3593 20. Extended pcretest with 3 new pattern features:
3594
3595 (i) A pattern option of the form ">rest-of-line" causes pcretest to
3596 write the compiled pattern to the file whose name is "rest-of-line".
3597 This is a straight binary dump of the data, with the saved pointer to
3598 the character tables forced to be NULL. The study data, if any, is
3599 written too. After writing, pcretest reads a new pattern.
3600
3601 (ii) If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
3602 compiled pattern from the given file. There must not be any
3603 occurrences of "<" in the file name (pretty unlikely); if there are,
3604 pcretest will instead treat the initial "<" as a pattern delimiter.
3605 After reading in the pattern, pcretest goes on to read data lines as
3606 usual.
3607
3608 (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
3609 and 16-bit fields in a compiled pattern, to simulate a pattern that
3610 was compiled on a host of opposite endianness.
3611
3612 21. The pcre-exec() function can now cope with patterns that were compiled on
3613 hosts of opposite endianness, with this restriction:
3614
3615 As for any compiled expression that is saved and used later, the tables
3616 pointer field cannot be preserved; the extra_data field in the arguments
3617 to pcre_exec() should be used to pass in a tables address if a value
3618 other than the default internal tables were used at compile time.
3619
3620 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
3621 now diagnosed as an error. Previously, most of the time, a negative number
3622 would have been treated as zero, but if in addition "ovector" was passed as
3623 NULL, a crash could occur.
3624
3625 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
3626 new versions from the libtool 1.5 distribution (the last one is a copy of
3627 a file called libtool.m4). This seems to have fixed the need to patch
3628 "configure" to support Darwin 1.3 (which I used to do). However, I still
3629 had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
3630 workstation).
3631
3632 24. Changed the PCRE licence to be the more standard "BSD" licence.
3633
3634
3635 Version 4.5 01-Dec-03
3636 ---------------------
3637
3638 1. There has been some re-arrangement of the code for the match() function so
3639 that it can be compiled in a version that does not call itself recursively.
3640 Instead, it keeps those local variables that need separate instances for
3641 each "recursion" in a frame on the heap, and gets/frees frames whenever it
3642 needs to "recurse". Keeping track of where control must go is done by means
3643 of setjmp/longjmp. The whole thing is implemented by a set of macros that
3644 hide most of the details from the main code, and operates only if
3645 NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
3646 "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
3647 operating.
3648
3649 To make it easier for callers to provide specially tailored get/free
3650 functions for this usage, two new functions, pcre_stack_malloc, and
3651 pcre_stack_free, are used. They are always called in strict stacking order,
3652 and the size of block requested is always the same.
3653
3654 The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
3655 PCRE has been compiled to use the stack or the heap for recursion. The
3656 -C option of pcretest uses this to show which version is compiled.
3657
3658 A new data escape \S, is added to pcretest; it causes the amounts of store
3659 obtained and freed by both kinds of malloc/free at match time to be added
3660 to the output.
3661
3662 2. Changed the locale test to use "fr_FR" instead of "fr" because that's
3663 what's available on my current Linux desktop machine.
3664
3665 3. When matching a UTF-8 string, the test for a valid string at the start has
3666 been extended. If start_offset is not zero, PCRE now checks that it points
3667 to a byte that is the start of a UTF-8 character. If not, it returns
3668 PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
3669 this is necessary because there may be backward assertions in the pattern.
3670 When matching the same subject several times, it may save resources to use
3671 PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
3672
3673 4. The code for checking the validity of UTF-8 strings has been tightened so
3674 that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
3675 containing "overlong sequences".
3676
3677 5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
3678 I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
3679 should have been "&", but it just so happened that all the cases this let
3680 through by mistake were picked up later in the function.
3681
3682 6. I had used a variable called "isblank" - this is a C99 function, causing
3683 some compilers to warn. To avoid this, I renamed it (as "blankclass").
3684
3685 7. Cosmetic: (a) only output another newline at the end of pcretest if it is
3686 prompting; (b) run "./pcretest /dev/null" at the start of the test script
3687 so the version is shown; (c) stop "make test" echoing "./RunTest".
3688
3689 8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
3690
3691 9. The prototype for memmove() for systems that don't have it was using
3692 size_t, but the inclusion of the header that defines size_t was later. I've
3693 moved the #includes for the C headers earlier to avoid this.
3694
3695 10. Added some adjustments to the code to make it easier to compiler on certain
3696 special systems:
3697
3698 (a) Some "const" qualifiers were missing.
3699 (b) Added the macro EXPORT before all exported functions; by default this
3700 is defined to be empty.
3701 (c) Changed the dftables auxiliary program (that builds chartables.c) so
3702 that it reads its output file name as an argument instead of writing
3703 to the standard output and assuming this can be redirected.
3704
3705 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
3706 class containing characters with values greater than 255, PCRE compilation
3707 went into a loop.
3708
3709 12. A recursive reference to a subpattern that was within another subpattern
3710 that had a minimum quantifier of zero caused PCRE to crash. For example,
3711 (x(y(?2))z)? provoked this bug with a subject that got as far as the
3712 recursion. If the recursively-called subpattern itself had a zero repeat,
3713 that was OK.
3714
3715 13. In pcretest, the buffer for reading a data line was set at 30K, but the
3716 buffer into which it was copied (for escape processing) was still set at
3717 1024, so long lines caused crashes.
3718
3719 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
3720 "internal error: code overflow...". This applied to any character class
3721 that was followed by a possessive quantifier.
3722
3723 15. Modified the Makefile to add libpcre.la as a prerequisite for
3724 libpcreposix.la because I was told this is needed for a parallel build to
3725 work.
3726
3727 16. If a pattern that contained .* following optional items at the start was
3728 studied, the wrong optimizing data was generated, leading to matching
3729 errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
3730 matching string must start with a or b or c. The correct conclusion for
3731 this pattern is that a match can start with any character.
3732
3733
3734 Version 4.4 13-Aug-03
3735 ---------------------
3736
3737 1. In UTF-8 mode, a character class containing characters with values between
3738 127 and 255 was not handled correctly if the compiled pattern was studied.
3739 In fixing this, I have also improved the studying algorithm for such
3740 classes (slightly).
3741
3742 2. Three internal functions had redundant arguments passed to them. Removal
3743 might give a very teeny performance improvement.
3744
3745 3. Documentation bug: the value of the capture_top field in a callout is *one
3746 more than* the number of the hightest numbered captured substring.
3747
3748 4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
3749 in incorrectly linking with a previously installed version. They now link
3750 explicitly with libpcre.la.
3751
3752 5. configure.in no longer needs to recognize Cygwin specially.
3753
3754 6. A problem in pcre.in for Windows platforms is fixed.
3755
3756 7. If a pattern was successfully studied, and the -d (or /D) flag was given to
3757 pcretest, it used to include the size of the study block as part of its
3758 output. Unfortunately, the structure contains a field that has a different
3759 size on different hardware architectures. This meant that the tests that
3760 showed this size failed. As the block is currently always of a fixed size,
3761 this information isn't actually particularly useful in pcretest output, so
3762 I have just removed it.
3763
3764 8. Three pre-processor statements accidentally did not start in column 1.
3765 Sadly, there are *still* compilers around that complain, even though
3766 standard C has not required this for well over a decade. Sigh.
3767
3768 9. In pcretest, the code for checking callouts passed small integers in the
3769 callout_data field, which is a void * field. However, some picky compilers
3770 complained about the casts involved for this on 64-bit systems. Now
3771 pcretest passes the address of the small integer instead, which should get
3772 rid of the warnings.
3773
3774 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
3775 both compile and run time, and gives an error if an invalid UTF-8 sequence
3776 is found. There is a option for disabling this check in cases where the
3777 string is known to be correct and/or the maximum performance is wanted.
3778
3779 11. In response to a bug report, I changed one line in Makefile.in from
3780
3781 -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
3782 to
3783 -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
3784
3785 to look similar to other lines, but I have no way of telling whether this
3786 is the right thing to do, as I do not use Windows. No doubt I'll get told
3787 if it's wrong...
3788
3789
3790 Version 4.3 21-May-03
3791 ---------------------
3792
3793 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
3794 Makefile.
3795
3796 2. Some refactoring to improve the quality of the code:
3797
3798 (i) The utf8_table... variables are now declared "const".
3799
3800 (ii) The code for \cx, which used the "case flipping" table to upper case
3801 lower case letters, now just substracts 32. This is ASCII-specific,
3802 but the whole concept of \cx is ASCII-specific, so it seems
3803 reasonable.
3804
3805 (iii) PCRE was using its character types table to recognize decimal and
3806 hexadecimal digits in the pattern. This is silly, because it handles
3807 only 0-9, a-f, and A-F, but the character types table is locale-
3808 specific, which means strange things might happen. A private
3809 table is now used for this - though it costs 256 bytes, a table is
3810 much faster than multiple explicit tests. Of course, the standard
3811 character types table is still used for matching digits in subject
3812 strings against \d.
3813
3814 (iv) Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
3815 ending in _t are). So I've renamed it as ESC_tee.
3816
3817 3. The first argument for regexec() in the POSIX wrapper should have been
3818 defined as "const".
3819
3820 4. Changed pcretest to use malloc() for its buffers so that they can be
3821 Electric Fenced for debugging.
3822
3823 5. There were several places in the code where, in UTF-8 mode, PCRE would try
3824 to read one or more bytes before the start of the subject string. Often this
3825 had no effect on PCRE's behaviour, but in some circumstances it could
3826 provoke a segmentation fault.
3827
3828 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
3829 to try to read one or more bytes before the start of the subject string.
3830
3831 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
3832 UTF-8 support could misbehave in various ways if the subject string
3833 contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
3834 area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
3835 back over UTF-8 characters.)
3836
3837
3838 Version 4.2 14-Apr-03
3839 ---------------------
3840
3841 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
3842
3843 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
3844 [ON_WINDOWS]: new variable, "#" on non-Windows platforms
3845 [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
3846 [WIN_PREFIX]: new variable, "cyg" for Cygwin
3847 * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
3848 and BUILD_EXEEXT
3849 Note: automatic setting of the BUILD variables is not yet working
3850 set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
3851 compile-time but not at link-time
3852 [LINK]: use for linking executables only
3853 make different versions for Windows and non-Windows
3854 [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
3855 libraries
3856 [LINK_FOR_BUILD]: new variable
3857 [OBJEXT]: use throughout
3858 [EXEEXT]: use throughout
3859 <winshared>: new target
3860 <wininstall>: new target
3861 <dftables.o>: use native compiler
3862 <dftables>: use native linker
3863 <install>: handle Windows platform correctly
3864 <clean>: ditto
3865 <check>: ditto
3866 copy DLL to top builddir before testing
3867
3868 As part of these changes, -no-undefined was removed again. This was reported
3869 to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
3870 in any case.
3871
3872 3. Some tidies to get rid of compiler warnings:
3873
3874 . In the match_data structure, match_limit was an unsigned long int, whereas
3875 match_call_count was an int. I've made them both unsigned long ints.
3876
3877 . In pcretest the fact that a const uschar * doesn't automatically cast to
3878 a void * provoked a warning.
3879
3880 . Turning on some more compiler warnings threw up some "shadow" variables
3881 and a few more missing casts.
3882
3883 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
3884 option, a class that contained a single character with a value between 128
3885 and 255 (e.g. /[\xFF]/) caused PCRE to crash.
3886
3887 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
3888 option, a class that contained several characters, but with at least one
3889 whose value was between 128 and 255 caused PCRE to crash.
3890
3891
3892 Version 4.1 12-Mar-03
3893 ---------------------
3894
3895 1. Compiling with gcc -pedantic found a couple of places where casts were
3896 needed, and a string in dftables.c that was longer than standard compilers are
3897 required to support.
3898
3899 2. Compiling with Sun's compiler found a few more places where the code could
3900 be tidied up in order to avoid warnings.
3901
3902 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
3903 first of these names is deprecated in the latest Autoconf in favour of the name
3904 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
3905 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
3906 analogy I have changed it to CFLAGS_FOR_BUILD.
3907
3908 4. Added -no-undefined to the linking command in the Makefile, because this is
3909 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
3910 linking step for the pcreposix library.
3911
3912 5. PCRE was failing to diagnose the case of two named groups with the same
3913 name.
3914
3915 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
3916 literal character that is needed in the subject for a match, and scans along to
3917 ensure that it is present before embarking on the full matching process. This
3918 saves time in cases of nested unlimited repeats that are never going to match.
3919 Problem: the scan can take a lot of time if the subject is very long (e.g.
3920 megabytes), thus penalizing straightforward matches. It is now done only if the
3921 amount of subject to be scanned is less than 1000 bytes.
3922
3923 7. A lesser problem with the same optimization is that it was recording the
3924 first character of an anchored pattern as "needed", thus provoking a search
3925 right along the subject, even when the first match of the pattern was going to
3926 fail. The "needed" character is now not set for anchored patterns, unless it
3927 follows something in the pattern that is of non-fixed length. Thus, it still
3928 fulfils its original purpose of finding quick non-matches in cases of nested
3929 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
3930
3931
3932 Version 4.0 17-Feb-03
3933 ---------------------
3934
3935 1. If a comment in an extended regex that started immediately after a meta-item
3936 extended to the end of string, PCRE compiled incorrect data. This could lead to
3937 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
3938
3939 2. Moved to autoconf 2.53 and libtool 1.4.2.
3940
3941 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
3942 the special perltest8 script is no longer needed - all the tests can be run
3943 from a single perltest script.
3944
3945 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
3946 by \s. It has now been removed in PCRE. This means it isn't recognized as
3947 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
3948 class [:space:] *does* include VT, thereby creating a mess.
3949
3950 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
3951 space and tab.
3952
3953 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
3954 its new features into the main test script, reducing the number of scripts.
3955
3956 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
3957 were backward compatible, and made the (?i) apply to the whole pattern, as if
3958 /i were given. Now it behaves more logically, and applies the option setting
3959 only to what follows. PCRE has been changed to follow suit. However, if it
3960 finds options settings right at the start of the pattern, it extracts them into
3961 the global options, as before. Thus, they show up in the info data.
3962
3963 8. Added support for the \Q...\E escape sequence. Characters in between are
3964 treated as literals. This is slightly different from Perl in that $ and @ are
3965 also handled as literals inside the quotes. In Perl, they will cause variable
3966 interpolation. Note the following examples:
3967
3968 Pattern PCRE matches Perl matches
3969
3970 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
3971 \Qabc\$xyz\E abc\$xyz abc\$xyz
3972 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
3973
3974 For compatibility with Perl, \Q...\E sequences are recognized inside character
3975 classes as well as outside them.
3976
3977 9. Re-organized 3 code statements in pcretest to avoid "overflow in
3978 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
3979 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
3980 signed/unsigned warnings.
3981
3982 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
3983 option for pcretest, so I've replaced it by a simple function that does just
3984 that job.
3985
3986 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
3987 "pcregrep -".
3988
3989 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
3990 Java package. This provides some syntactic sugar for simple cases of what my
3991 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
3992 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
3993 item, you can use this simplified notation. Note that only makes sense with
3994 greedy quantifiers. Consequently, the use of the possessive quantifier forces
3995 greediness, whatever the setting of the PCRE_UNGREEDY option.
3996
3997 13. A change of greediness default within a pattern was not taking effect at
3998 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
3999 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
4000 was abstracted outside.
4001
4002 14. PCRE now supports the \G assertion. It is true when the current matching
4003 position is at the start point of the match. This differs from \A when the
4004 starting offset is non-zero. Used with the /g option of pcretest (or similar
4005 code), it works in the same way as it does for Perl's /g option. If all
4006 alternatives of a regex begin with \G, the expression is anchored to the start
4007 match position, and the "anchored" flag is set in the compiled expression.
4008
4009 15. Some bugs concerning the handling of certain option changes within patterns
4010 have been fixed. These applied to options other than (?ims). For example,
4011 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
4012 been the other way round. Some of this was related to change 7 above.
4013
4014 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
4015 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
4016 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
4017 POSIX classes only within a class (e.g. /[[:alpha:]]/).
4018
4019 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
4020 mode. Unlike ".", it always matches newline, whatever the setting of
4021 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
4022 assertions. Perl allows it, but it doesn't (in general) work because it can't
4023 calculate the length of the lookbehind. At least, that's the case for Perl
4024 5.8.0 - I've been told they are going to document that it doesn't work in
4025 future.
4026
4027 18. Added an error diagnosis for escapes that PCRE does not support: these are
4028 \L, \l, \N, \P, \p, \U, \u, and \X.
4029
4030 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
4031 reading past the end of the pattern in cases such as /[abcd/.
4032
4033 20. PCRE was getting more memory than necessary for patterns with classes that
4034 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
4035
4036 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
4037 compiling PCRE for use with Virtual Pascal.
4038
4039 22. Small fix to the Makefile to make it work properly if the build is done
4040 outside the source tree.
4041
4042 23. Added a new extension: a condition to go with recursion. If a conditional
4043 subpattern starts with (?(R) the "true" branch is used if recursion has
4044 happened, whereas the "false" branch is used only at the top level.
4045
4046 24. When there was a very long string of literal characters (over 255 bytes
4047 without UTF support, over 250 bytes with UTF support), the computation of how
4048 much memory was required could be incorrect, leading to segfaults or other
4049 strange effects.
4050
4051 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
4052 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
4053 there was a subsequent back reference to those brackets. This meant that, for
4054 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
4055 possible to check for precisely this case. All we can do is abandon the
4056 optimization if .* occurs inside capturing brackets when there are any back
4057 references whatsoever. (See below for a better fix that came later.)
4058
4059 26. The handling of the optimization for finding the first character of a
4060 non-anchored pattern, and for finding a character that is required later in the
4061 match were failing in some cases. This didn't break the matching; it just
4062 failed to optimize when it could. The way this is done has been re-implemented.
4063
4064 27. Fixed typo in error message for invalid (?R item (it said "(?p").
4065
4066 28. Added a new feature that provides some of the functionality that Perl
4067 provides with (?{...}). The facility is termed a "callout". The way it is done
4068 in PCRE is for the caller to provide an optional function, by setting
4069 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
4070 global variable. By default it is unset, which disables all calling out. To get
4071 the function called, the regex must include (?C) at appropriate points. This
4072 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
4073 This provides a means of identifying different callout points. When PCRE
4074 reaches such a point in the regex, if pcre_callout has been set, the external
4075 function is called. It is provided with data in a structure called
4076 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
4077 matching continues; if it returns a non-zero value, the match at the current
4078 point fails. However, backtracking will occur if possible. [This was changed
4079 later and other features added - see item 49 below.]
4080
4081 29. pcretest is upgraded to test the callout functionality. It provides a
4082 callout function that displays information. By default, it shows the start of
4083 the match and the current position in the text. There are some new data escapes
4084 to vary what happens:
4085
4086 \C+ in addition, show current contents of captured substrings
4087 \C- do not supply a callout function
4088 \C!n return 1 when callout number n is reached
4089 \C!n!m return 1 when callout number n is reached for the mth time
4090
4091 30. If pcregrep was called with the -l option and just a single file name, it
4092 output "<stdin>" if a match was found, instead of the file name.
4093
4094 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
4095 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
4096 pcre_exec(). This saves a malloc/free per call. The default value of
4097 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
4098 when configuring.
4099
4100 32. The default maximum size of a compiled pattern is 64K. There have been a
4101 few cases of people hitting this limit. The code now uses macros to handle the
4102 storing of links as offsets within the compiled pattern. It defaults to 2-byte
4103 links, but this can be changed to 3 or 4 bytes by --with-link-size when
4104 configuring. Tests 2 and 5 work only with 2-byte links because they output
4105 debugging information about compiled patterns.
4106
4107 33. Internal code re-arrangements:
4108
4109 (a) Moved the debugging function for printing out a compiled regex into
4110 its own source file (printint.c) and used #include to pull it into
4111 pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
4112 separate copies.
4113
4114 (b) Defined the list of op-code names for debugging as a macro in
4115 internal.h so that it is next to the definition of the opcodes.
4116
4117 (c) Defined a table of op-code lengths for simpler skipping along compiled
4118 code. This is again a macro in internal.h so that it is next to the
4119 definition of the opcodes.
4120
4121 34. Added support for recursive calls to individual subpatterns, along the
4122 lines of Robin Houston's patch (but implemented somewhat differently).
4123
4124 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
4125 allow it to read and process whole directories in Win32. This code was
4126 contributed by Lionel Fourquaux; it has not been tested by me.
4127
4128 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
4129 used to name a group. Names consist of alphanumerics and underscores, and must
4130 be unique. Back references use the syntax (?P=name) and recursive calls use
4131 (?P>name) which is a PCRE extension to the Python extension. Groups still have
4132 numbers. The function pcre_fullinfo() can be used after compilation to extract
4133 a name/number map. There are three relevant calls:
4134
4135 PCRE_INFO_NAMEENTRYSIZE yields the size of each entry in the map
4136 PCRE_INFO_NAMECOUNT yields the number of entries
4137 PCRE_INFO_NAMETABLE yields a pointer to the map.
4138
4139 The map is a vector of fixed-size entries. The size of each entry depends on
4140 the length of the longest name used. The first two bytes of each entry are the
4141 group number, most significant byte first. There follows the corresponding
4142 name, zero terminated. The names are in alphabetical order.
4143
4144 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
4145 case instead of 255. Making it the same both with and without UTF-8 support
4146 means that the same test output works with both.
4147
4148 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
4149 calling malloc() with a zero argument.
4150
4151 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
4152 optimization. I've improved things by keeping a bitmap of backreferences with
4153 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
4154 fact referenced, the optimization can be applied. It is unlikely that a
4155 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
4156 the match to follow \n) will appear inside brackets with a number greater than
4157 31, but if it does, any back reference > 31 suppresses the optimization.
4158
4159 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
4160 of disabling numbered capturing parentheses. Any opening parenthesis that is
4161 not followed by ? behaves as if it were followed by ?: but named parentheses
4162 can still be used for capturing (and they will acquire numbers in the usual
4163 way).
4164
4165 41. Redesigned the return codes from the match() function into yes/no/error so
4166 that errors can be passed back from deep inside the nested calls. A malloc
4167 failure while inside a recursive subpattern call now causes the
4168 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
4169
4170 42. It is now possible to set a limit on the number of times the match()
4171 function is called in a call to pcre_exec(). This facility makes it possible to
4172 limit the amount of recursion and backtracking, though not in a directly
4173 obvious way, because the match() function is used in a number of different
4174 circumstances. The count starts from zero for each position in the subject
4175 string (for non-anchored patterns). The default limit is, for compatibility, a
4176 large number, namely 10 000 000. You can change this in two ways:
4177
4178 (a) When configuring PCRE before making, you can use --with-match-limit=n
4179 to set a default value for the compiled library.
4180
4181 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
4182 a different value is set. See 45 below.
4183
4184 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
4185
4186 43. Added a new function pcre_config(int, void *) to enable run-time extraction
4187 of things that can be changed at compile time. The first argument specifies
4188 what is wanted and the second points to where the information is to be placed.
4189 The current list of available information is:
4190
4191 PCRE_CONFIG_UTF8
4192
4193 The output is an integer that is set to one if UTF-8 support is available;
4194 otherwise it is set to zero.
4195
4196 PCRE_CONFIG_NEWLINE
4197
4198 The output is an integer that it set to the value of the code that is used for
4199 newline. It is either LF (10) or CR (13).
4200
4201 PCRE_CONFIG_LINK_SIZE
4202
4203 The output is an integer that contains the number of bytes used for internal
4204 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
4205
4206 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
4207
4208 The output is an integer that contains the threshold above which the POSIX
4209 interface uses malloc() for output vectors. See item 31 above.
4210
4211 PCRE_CONFIG_MATCH_LIMIT
4212
4213 The output is an unsigned integer that contains the default limit of the number
4214 of match() calls in a pcre_exec() execution. See 42 above.
4215
4216 44. pcretest has been upgraded by the addition of the -C option. This causes it
4217 to extract all the available output from the new pcre_config() function, and to
4218 output it. The program then exits immediately.
4219
4220 45. A need has arisen to pass over additional data with calls to pcre_exec() in
4221 order to support additional features. One way would have been to define
4222 pcre_exec2() (for example) with extra arguments, but this would not have been
4223 extensible, and would also have required all calls to the original function to
4224 be mapped to the new one. Instead, I have chosen to extend the mechanism that
4225 is used for passing in "extra" data from pcre_study().
4226
4227 The pcre_extra structure is now exposed and defined in pcre.h. It currently
4228 contains the following fields:
4229
4230 flags a bitmap indicating which of the following fields are set
4231 study_data opaque data from pcre_study()
4232 match_limit a way of specifying a limit on match() calls for a specific
4233 call to pcre_exec()
4234 callout_data data for callouts (see 49 below)
4235
4236 The flag bits are also defined in pcre.h, and are
4237
4238 PCRE_EXTRA_STUDY_DATA
4239 PCRE_EXTRA_MATCH_LIMIT
4240 PCRE_EXTRA_CALLOUT_DATA
4241
4242 The pcre_study() function now returns one of these new pcre_extra blocks, with
4243 the actual study data pointed to by the study_data field, and the
4244 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
4245 before. That is, this change is entirely upwards-compatible and requires no
4246 change to existing code.
4247
4248 If you want to pass in additional data to pcre_exec(), you can either place it
4249 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
4250 block.
4251
4252 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
4253 data string contains the escape sequence \M, pcretest calls pcre_exec() several
4254 times with different match limits, until it finds the minimum value needed for
4255 pcre_exec() to complete. The value is then output. This can be instructive; for
4256 most simple matches the number is quite small, but for pathological cases it
4257 gets very large very quickly.
4258
4259 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
4260 returns the size of the data block pointed to by the study_data field in a
4261 pcre_extra block, that is, the value that was passed as the argument to
4262 pcre_malloc() when PCRE was getting memory in which to place the information
4263 created by pcre_study(). The fourth argument should point to a size_t variable.
4264 pcretest has been extended so that this information is shown after a successful
4265 pcre_study() call when information about the compiled regex is being displayed.
4266
4267 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
4268 because what follows is always an absolute path. (Later: it turns out that this
4269 is more than cosmetic for MinGW, because it doesn't like empty path
4270 components.)
4271
4272 49. Some changes have been made to the callout feature (see 28 above):
4273
4274 (i) A callout function now has three choices for what it returns:
4275
4276 0 => success, carry on matching
4277 > 0 => failure at this point, but backtrack if possible
4278 < 0 => serious error, return this value from pcre_exec()
4279
4280 Negative values should normally be chosen from the set of PCRE_ERROR_xxx
4281 values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
4282 "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
4283 use by callout functions. It will never be used by PCRE itself.
4284
4285 (ii) The pcre_extra structure (see 45 above) has a void * field called
4286 callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
4287 pcre_callout_block structure has a field of the same name. The contents of
4288 the field passed in the pcre_extra structure are passed to the callout
4289 function in the corresponding field in the callout block. This makes it
4290 easier to use the same callout-containing regex from multiple threads. For
4291 testing, the pcretest program has a new data escape
4292
4293 \C*n pass the number n (may be negative) as callout_data
4294
4295 If the callout function in pcretest receives a non-zero value as
4296 callout_data, it returns that value.
4297
4298 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
4299 there were some redundant $(CFLAGS) in commands that are now specified as
4300 $(LINK), which already includes $(CFLAGS).
4301
4302 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
4303 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
4304 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
4305 one-byte characters throughout. Note that case-insensitive matching applies
4306 only to characters whose values are less than 256. PCRE doesn't support the
4307 notion of cases for higher-valued characters.
4308
4309 (i) A character class whose characters are all within 0-255 is handled as
4310 a bit map, and the map is inverted for negative classes. Previously, a
4311 character > 255 always failed to match such a class; however it should
4312 match if the class was a negative one (e.g. [^ab]). This has been fixed.
4313
4314 (ii) A negated character class with a single character < 255 is coded as
4315 "not this character" (OP_NOT). This wasn't working properly when the test
4316 character was multibyte, either singly or repeated.
4317
4318 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
4319 mode, for example: \x{100}{2,3}.
4320
4321 (iv) The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
4322 singly or repeated) now correctly test multibyte characters. However,
4323 PCRE doesn't recognize any characters with values greater than 255 as
4324 digits, spaces, or word characters. Such characters always match \D, \S,
4325 and \W, and never match \d, \s, or \w.
4326
4327 (v) Classes may now contain characters and character ranges with values
4328 greater than 255. For example: [ab\x{100}-\x{400}].
4329
4330 (vi) pcregrep now has a --utf-8 option (synonym -u) which makes it call
4331 PCRE in UTF-8 mode.
4332
4333 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
4334 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
4335 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
4336 value.)
4337
4338 53. The single man page has become too large. I have therefore split it up into
4339 a number of separate man pages. These also give rise to individual HTML pages;
4340 these are now put in a separate directory, and there is an index.html page that
4341 lists them all. Some hyperlinking between the pages has been installed.
4342
4343 54. Added convenience functions for handling named capturing parentheses.
4344
4345 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
4346 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
4347 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
4348 are faulted.
4349
4350 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
4351 calling configure. These values are used when compiling the dftables.c program
4352 which is run to generate the source of the default character tables. They
4353 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
4354 you will need to set these values.
4355
4356 57. Updated the building process for Windows DLL, as provided by Fred Cox.
4357
4358
4359 Version 3.9 02-Jan-02
4360 ---------------------
4361
4362 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
4363
4364 2. If --disable-static was given, the building process failed when trying to
4365 build pcretest and pcregrep. (For some reason it was using libtool to compile
4366 them, which is not right, as they aren't part of the library.)
4367
4368
4369 Version 3.8 18-Dec-01
4370 ---------------------
4371
4372 1. The experimental UTF-8 code was completely screwed up. It was packing the
4373 bytes in the wrong order. How dumb can you get?
4374
4375
4376 Version 3.7 29-Oct-01
4377 ---------------------
4378
4379 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
4380 This caused pcretest, when used on the test data, to segfault. Unfortunately,
4381 this didn't happen under Solaris 8, where I normally test things.
4382
4383 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
4384 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
4385 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
4386 this fix an hour or so after the initial 3.7 release.)
4387
4388
4389 Version 3.6 23-Oct-01
4390 ---------------------
4391
4392 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
4393 offsets passed as NULL with zero offset count.
4394
4395 2. The config.guess and config.sub files had not been updated when I moved to
4396 the latest autoconf.
4397
4398
4399 Version 3.5 15-Aug-01
4400 ---------------------
4401
4402 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
4403 had been forgotten.
4404
4405 2. By using declared but undefined structures, we can avoid using "void"
4406 definitions in pcre.h while keeping the internal definitions of the structures
4407 private.
4408
4409 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
4410 user point of view, this means that both static and shared libraries are built
4411 by default, but this can be individually controlled. More of the work of
4412 handling this static/shared cases is now inside libtool instead of PCRE's make
4413 file.
4414
4415 4. The pcretest utility is now installed along with pcregrep because it is
4416 useful for users (to test regexs) and by doing this, it automatically gets
4417 relinked by libtool. The documentation has been turned into a man page, so
4418 there are now .1, .txt, and .html versions in /doc.
4419
4420 5. Upgrades to pcregrep:
4421 (i) Added long-form option names like gnu grep.
4422 (ii) Added --help to list all options with an explanatory phrase.
4423 (iii) Added -r, --recursive to recurse into sub-directories.
4424 (iv) Added -f, --file to read patterns from a file.
4425
4426 6. pcre_exec() was referring to its "code" argument before testing that
4427 argument for NULL (and giving an error if it was NULL).
4428
4429 7. Upgraded Makefile.in to allow for compiling in a different directory from
4430 the source directory.
4431
4432 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
4433 options bits, the pointer it was passed was to an int instead of to an unsigned
4434 long int. This mattered only on 64-bit systems.
4435
4436 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
4437 generated) instead of pcre.in, which it its source. Also made the same change
4438 in several of the .c files.
4439
4440 10. A new release of gcc defines printf() as a macro, which broke pcretest
4441 because it had an ifdef in the middle of a string argument for printf(). Fixed
4442 by using separate calls to printf().
4443
4444 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
4445 script, to force use of CR or LF instead of \n in the source. On non-Unix
4446 systems, the value can be set in config.h.
4447
4448 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
4449 absolute limit. Changed the text of the error message to make this clear, and
4450 likewise updated the man page.
4451
4452 13. The limit of 99 on the number of capturing subpatterns has been removed.
4453 The new limit is 65535, which I hope will not be a "real" limit.
4454
4455
4456 Version 3.4 22-Aug-00
4457 ---------------------
4458
4459 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
4460
4461 2. Diagnose condition (?(0) as an error instead of crashing on matching.
4462
4463
4464 Version 3.3 01-Aug-00
4465 ---------------------
4466
4467 1. If an octal character was given, but the value was greater than \377, it
4468 was not getting masked to the least significant bits, as documented. This could
4469 lead to crashes in some systems.
4470
4471 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
4472 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
4473
4474 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
4475 These just pass their arguments on to (pcre_free)(), but they are provided
4476 because some uses of PCRE bind it to non-C systems that can call its functions,
4477 but cannot call free() or pcre_free() directly.
4478
4479 4. Add "make test" as a synonym for "make check". Corrected some comments in
4480 the Makefile.
4481
4482 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
4483 Makefile.
4484
4485 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
4486 command called pgrep for grepping around the active processes.
4487
4488 7. Added the beginnings of support for UTF-8 character strings.
4489
4490 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
4491 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
4492 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
4493 out for the ar command.)
4494
4495
4496 Version 3.2 12-May-00
4497 ---------------------
4498
4499 This is purely a bug fixing release.
4500
4501 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
4502 of ZA. This was just one example of several cases that could provoke this bug,
4503 which was introduced by change 9 of version 2.00. The code for breaking
4504 infinite loops after an iteration that matches an empty string was't working
4505 correctly.
4506
4507 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
4508 when matched against abbab (for example). After matching an empty string, it
4509 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
4510 caused it to match further down the string than it should.
4511
4512 3. The code contained an inclusion of sys/types.h. It isn't clear why this
4513 was there because it doesn't seem to be needed, and it causes trouble on some
4514 systems, as it is not a Standard C header. It has been removed.
4515
4516 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
4517 were reported on the Macintosh. The changes were from
4518
4519 while ((c = *(++ptr)) != 0 && c != '\n');
4520 to
4521 while ((c = *(++ptr)) != 0 && c != '\n') ;
4522
4523 Totally extraordinary, but if that's what it takes...
4524
4525 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
4526 available. Added HAVE_BCOPY and an autoconf test for it; if neither
4527 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
4528 assumes the way PCRE uses memmove() (always moving upwards).
4529
4530 6. PCRE is being used in one environment where strchr() is not available. There
4531 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
4532 faster code anyway.
4533
4534
4535 Version 3.1 09-Feb-00
4536 ---------------------
4537
4538 The only change in this release is the fixing of some bugs in Makefile.in for
4539 the "install" target:
4540
4541 (1) It was failing to install pcreposix.h.
4542
4543 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
4544
4545
4546 Version 3.0 01-Feb-00
4547 ---------------------
4548
4549 1. Add support for the /+ modifier to perltest (to output $` like it does in
4550 pcretest).
4551
4552 2. Add support for the /g modifier to perltest.
4553
4554 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
4555 matches null strings.
4556
4557 4. Fix perltest so that it doesn't do unwanted things when fed an empty
4558 pattern. Perl treats empty patterns specially - it reuses the most recent
4559 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
4560 effect.
4561
4562 5. The POSIX interface was broken in that it was just handing over the POSIX
4563 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
4564 required a bigger vector, with some working space on the end. This means that
4565 the POSIX wrapper now has to get and free some memory, and copy the results.
4566
4567 6. Added some simple autoconf support, placing the test data and the
4568 documentation in separate directories, re-organizing some of the
4569 information files, and making it build pcre-config (a GNU standard). Also added
4570 libtool support for building PCRE as a shared library, which is now the
4571 default.
4572
4573 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
4574 09 are not valid octal constants. Single digits will be used for minor values
4575 less than 10.
4576
4577 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
4578 existing programs that set these in the POSIX interface can use PCRE without
4579 modification.
4580
4581 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
4582 return all that pcre_info() returns, plus additional data. The pcre_info()
4583 function is retained for compatibility, but is considered to be obsolete.
4584
4585 10. Added experimental recursion feature (?R) to handle one common case that
4586 Perl 5.6 will be able to do with (?p{...}).
4587
4588 11. Added support for POSIX character classes like [:alpha:], which Perl is
4589 adopting.
4590
4591
4592 Version 2.08 31-Aug-99
4593 ----------------------
4594
4595 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
4596 trying to match at the startoffset position, but instead was moving forward to
4597 the next newline as if a previous match had failed.
4598
4599 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
4600 and could get into a loop if a null string was matched other than at the start
4601 of the subject.
4602
4603 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
4604 be distinguished at compile time, and for completeness also added PCRE_DATE.
4605
4606 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
4607 in GnuWin32 environments.
4608
4609
4610 Version 2.07 29-Jul-99
4611 ----------------------
4612
4613 1. The documentation is now supplied in plain text form and HTML as well as in
4614 the form of man page sources.
4615
4616 2. C++ compilers don't like assigning (void *) values to other pointer types.
4617 In particular this affects malloc(). Although there is no problem in Standard
4618 C, I've put in casts to keep C++ compilers happy.
4619
4620 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
4621 should be (const char *).
4622
4623 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
4624 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
4625 However, I haven't made this a standard facility. The documentation doesn't
4626 mention it, and the Makefile doesn't support it.
4627
4628 5. The Makefile now contains an "install" target, with editable destinations at
4629 the top of the file. The pcretest program is not installed.
4630
4631 6. pgrep -V now gives the PCRE version number and date.
4632
4633 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
4634 causing the entire string to be ignored, instead of just the last character.
4635
4636 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
4637 non-matching string, it can take a very, very long time, even for strings of
4638 quite modest length, because of the nested recursion. PCRE now does better in
4639 some of these cases. It does this by remembering the last required literal
4640 character in the pattern, and pre-searching the subject to ensure it is present
4641 before running the real match. In other words, it applies a heuristic to detect
4642 some types of certain failure quickly, and in the above example, if presented
4643 with a string that has no trailing " it gives "no match" very quickly.
4644
4645 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
4646 other alternatives are tried instead.
4647
4648
4649 Version 2.06 09-Jun-99
4650 ----------------------
4651
4652 1. Change pcretest's output for amount of store used to show just the code
4653 space, because the remainder (the data block) varies in size between 32-bit and
4654 64-bit systems.
4655
4656 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
4657 start matching at. This allows lookbehinds to work when searching for multiple
4658 occurrences in a string.
4659
4660 3. Added additional options to pcretest for testing multiple occurrences:
4661
4662 /+ outputs the rest of the string that follows a match
4663 /g loops for multiple occurrences, using the new startoffset argument
4664 /G loops for multiple occurrences by passing an incremented pointer
4665
4666 4. PCRE wasn't doing the "first character" optimization for patterns starting
4667 with \b or \B, though it was doing it for other lookbehind assertions. That is,
4668 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
4669 the letter 'x'. On long subject strings, this gives a significant speed-up.
4670
4671
4672 Version 2.05 21-Apr-99
4673 ----------------------
4674
4675 1. Changed the type of magic_number from int to long int so that it works
4676 properly on 16-bit systems.
4677
4678 2. Fixed a bug which caused patterns starting with .* not to work correctly
4679 when the subject string contained newline characters. PCRE was assuming
4680 anchoring for such patterns in all cases, which is not correct because .* will
4681 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
4682 DOTALL is set at top level; otherwise it knows that patterns starting with .*
4683 must be retried after every newline in the subject.
4684
4685
4686 Version 2.04 18-Feb-99
4687 ----------------------
4688
4689 1. For parenthesized subpatterns with repeats whose minimum was zero, the
4690 computation of the store needed to hold the pattern was incorrect (too large).
4691 If such patterns were nested a few deep, this could multiply and become a real
4692 problem.
4693
4694 2. Added /M option to pcretest to show the memory requirement of a specific
4695 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
4696
4697 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
4698 compiled in such a way that the backtracking after subsequent failure was
4699 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
4700 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
4701
4702
4703 Version 2.03 02-Feb-99
4704 ----------------------
4705
4706 1. Fixed typo and small mistake in man page.
4707
4708 2. Added 4th condition (GPL supersedes if conflict) and created separate
4709 LICENCE file containing the conditions.
4710
4711 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
4712 Perl, that is the internal \ allows the delimiter to be included in the
4713 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
4714 the final delimiter, add \ to the end of the pattern (to test the error).
4715
4716 4. Added the convenience functions for extracting substrings after a successful
4717 match. Updated pcretest to make it able to test these functions.
4718
4719
4720 Version 2.02 14-Jan-99
4721 ----------------------
4722
4723 1. Initialized the working variables associated with each extraction so that
4724 their saving and restoring doesn't refer to uninitialized store.
4725
4726 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
4727 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
4728 fix the problem.
4729
4730 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
4731 calls, and wasn't printing the correct value for compiling calls. Increased the
4732 default value of LOOPREPEAT, and the number of significant figures in the
4733 times.
4734
4735 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
4736
4737 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
4738 a building problem on Windows NT with a FAT file system.
4739
4740
4741 Version 2.01 21-Oct-98
4742 ----------------------
4743
4744 1. Changed the API for pcre_compile() to allow for the provision of a pointer
4745 to character tables built by pcre_maketables() in the current locale. If NULL
4746 is passed, the default tables are used.
4747
4748
4749 Version 2.00 24-Sep-98
4750 ----------------------
4751
4752 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
4753 it any more.
4754
4755 2. Allow quantification of (?>) groups, and make it work correctly.
4756
4757 3. The first character computation wasn't working for (?>) groups.
4758
4759 4. Correct the implementation of \Z (it is permitted to match on the \n at the
4760 end of the subject) and add 5.005's \z, which really does match only at the
4761 very end of the subject.
4762
4763 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
4764
4765 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
4766 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
4767 localized options. All options to pcre_study() were also removed.
4768
4769 7. Add other new features from 5.005:
4770
4771 $(?<= positive lookbehind
4772 $(?<! negative lookbehind
4773 (?imsx-imsx) added the unsetting capability
4774 such a setting is global if at outer level; local otherwise
4775 (?imsx-imsx:) non-capturing groups with option setting
4776 (?(cond)re|re) conditional pattern matching
4777
4778 A backreference to itself in a repeated group matches the previous
4779 captured string.
4780
4781 8. General tidying up of studying (both automatic and via "study")
4782 consequential on the addition of new assertions.
4783
4784 9. As in 5.005, unlimited repeated groups that could match an empty substring
4785 are no longer faulted at compile time. Instead, the loop is forcibly broken at
4786 runtime if any iteration does actually match an empty substring.
4787
4788 10. Include the RunTest script in the distribution.
4789
4790 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
4791 discrepancies, some of which were old and were also with respect to 5.004. They
4792 have now been fixed.
4793
4794
4795 Version 1.09 28-Apr-98
4796 ----------------------
4797
4798 1. A negated single character class followed by a quantifier with a minimum
4799 value of one (e.g. [^x]{1,6} ) was not compiled correctly. This could lead to
4800 program crashes, or just wrong answers. This did not apply to negated classes
4801 containing more than one character, or to minima other than one.
4802
4803
4804 Version 1.08 27-Mar-98
4805 ----------------------
4806
4807 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
4808
4809 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
4810 latter must appear before anything that relies on it in the pattern.
4811
4812
4813 Version 1.07 16-Feb-98
4814 ----------------------
4815
4816 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
4817 repeat of a potentially empty string).
4818
4819
4820 Version 1.06 23-Jan-98
4821 ----------------------
4822
4823 1. Added Markus Oberhumer's little patches for C++.
4824
4825 2. Literal strings longer than 255 characters were broken.
4826
4827
4828 Version 1.05 23-Dec-97
4829 ----------------------
4830
4831 1. Negated character classes containing more than one character were failing if
4832 PCRE_CASELESS was set at run time.
4833
4834
4835 Version 1.04 19-Dec-97
4836 ----------------------
4837
4838 1. Corrected the man page, where some "const" qualifiers had been omitted.
4839
4840 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
4841 input syntax.
4842
4843 3. Fixed memory leak which occurred when a regex with back references was
4844 matched with an offsets vector that wasn't big enough. The temporary memory
4845 that is used in this case wasn't being freed if the match failed.
4846
4847 4. Tidied pcretest to ensure it frees memory that it gets.
4848
4849 5. Temporary memory was being obtained in the case where the passed offsets
4850 vector was exactly big enough.
4851
4852 6. Corrected definition of offsetof() from change 5 below.
4853
4854 7. I had screwed up change 6 below and broken the rules for the use of
4855 setjmp(). Now fixed.
4856
4857
4858 Version 1.03 18-Dec-97
4859 ----------------------
4860
4861 1. A erroneous regex with a missing opening parenthesis was correctly
4862 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
4863 on some systems.
4864
4865 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
4866 it was reported that one broken compiler failed on the former because "code" is
4867 also an independent variable.
4868
4869 3. The erroneous regex a[]b caused an array overrun reference.
4870
4871 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
4872 fail on data ending with that character. (It was going on too far, and checking
4873 the next character, typically a binary zero.) This was specific to the
4874 optimized code for single-character negative classes.
4875
4876 5. Added a contributed patch from the TIN world which does the following:
4877
4878 + Add an undef for memmove, in case the the system defines a macro for it.
4879
4880 + Add a definition of offsetof(), in case there isn't one. (I don't know
4881 the reason behind this - offsetof() is part of the ANSI standard - but
4882 it does no harm).
4883
4884 + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
4885 most of the places where whitespace preceded '#'. I have given up and
4886 allowed the remaining 2 cases to be at the margin.
4887
4888 + Rename some variables in pcre to eliminate shadowing. This seems very
4889 pedantic, but does no harm, of course.
4890
4891 6. Moved the call to setjmp() into its own function, to get rid of warnings
4892 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
4893
4894 7. Constructs such as \d{8,} were compiling into the equivalent of
4895 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
4896 outcome, but in this particular case used more store than had been allocated,
4897 which caused the bug to be discovered because it threw up an internal error.
4898
4899 8. The debugging code in both pcre and pcretest for outputting the compiled
4900 form of a regex was going wrong in the case of back references followed by
4901 curly-bracketed repeats.
4902
4903
4904 Version 1.02 12-Dec-97
4905 ----------------------
4906
4907 1. Typos in pcre.3 and comments in the source fixed.
4908
4909 2. Applied a contributed patch to get rid of places where it used to remove
4910 'const' from variables, and fixed some signed/unsigned and uninitialized
4911 variable warnings.
4912
4913 3. Added the "runtest" target to Makefile.
4914
4915 4. Set default compiler flag to -O2 rather than just -O.
4916
4917
4918 Version 1.01 19-Nov-97
4919 ----------------------
4920
4921 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
4922 like /([ab]*)*/, that is, for classes with more than one character in them.
4923
4924 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
4925 as /((?>a*))*/ (a PCRE_EXTRA facility).
4926
4927
4928 Version 1.00 18-Nov-97
4929 ----------------------
4930
4931 1. Added compile-time macros to support systems such as SunOS4 which don't have
4932 memmove() or strerror() but have other things that can be used instead.
4933
4934 2. Arranged that "make clean" removes the executables.
4935
4936
4937 Version 0.99 27-Oct-97
4938 ----------------------
4939
4940 1. Fixed bug in code for optimizing classes with only one character. It was
4941 initializing a 32-byte map regardless, which could cause it to run off the end
4942 of the memory it had got.
4943
4944 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
4945
4946
4947 Version 0.98 22-Oct-97
4948 ----------------------
4949
4950 1. Fixed bug in code for handling temporary memory usage when there are more
4951 back references than supplied space in the ovector. This could cause segfaults.
4952
4953
4954 Version 0.97 21-Oct-97
4955 ----------------------
4956
4957 1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
4958
4959 2. Optimized negated single characters not to use a bit map.
4960
4961 3. Brought error texts together as macro definitions; clarified some of them;
4962 fixed one that was wrong - it said "range out of order" when it meant "invalid
4963 escape sequence".
4964
4965 4. Changed some char * arguments to const char *.
4966
4967 5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
4968
4969 6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
4970 pcretest.
4971
4972
4973 Version 0.96 16-Oct-97
4974 ----------------------
4975
4976 1. Added a simple "pgrep" utility to the distribution.
4977
4978 2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
4979 unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
4980 where "ddd" means "one or more decimal digits".
4981
4982 3. Fixed serious bug. If a pattern had a back reference, but the call to
4983 pcre_exec() didn't supply a large enough ovector to record the related
4984 identifying subpattern, the match always failed. PCRE now remembers the number
4985 of the largest back reference, and gets some temporary memory in which to save
4986 the offsets during matching if necessary, in order to ensure that
4987 backreferences always work.
4988
4989 4. Increased the compatibility with Perl in a number of ways:
4990
4991 (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
4992 to request this handling. The option can be set at compile or exec time.
4993
4994 (b) $ matches before a terminating newline by default; an option
4995 PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
4996 mode). The option can be set at compile or exec time.
4997
4998 (c) The handling of \ followed by a digit other than 0 is now supposed to be
4999 the same as Perl's. If the decimal number it represents is less than 10
5000 or there aren't that many previous left capturing parentheses, an octal
5001 escape is read. Inside a character class, it's always an octal escape,
5002 even if it is a single digit.
5003
5004 (d) An escaped but undefined alphabetic character is taken as a literal,
5005 unless PCRE_EXTRA is set. Currently this just reserves the remaining
5006 escapes.
5007
5008 (e) {0} is now permitted. (The previous item is removed from the compiled
5009 pattern).
5010
5011 5. Changed all the names of code files so that the basic parts are no longer
5012 than 10 characters, and abolished the teeny "globals.c" file.
5013
5014 6. Changed the handling of character classes; they are now done with a 32-byte
5015 bit map always.
5016
5017 7. Added the -d and /D options to pcretest to make it possible to look at the
5018 internals of compilation without having to recompile pcre.
5019
5020
5021 Version 0.95 23-Sep-97
5022 ----------------------
5023
5024 1. Fixed bug in pre-pass concerning escaped "normal" characters such as \x5c or
5025 \x20 at the start of a run of normal characters. These were being treated as
5026 real characters, instead of the source characters being re-checked.
5027
5028
5029 Version 0.94 18-Sep-97
5030 ----------------------
5031
5032 1. The functions are now thread-safe, with the caveat that the global variables
5033 containing pointers to malloc() and free() or alternative functions are the
5034 same for all threads.
5035
5036 2. Get pcre_study() to generate a bitmap of initial characters for non-
5037 anchored patterns when this is possible, and use it if passed to pcre_exec().
5038
5039
5040 Version 0.93 15-Sep-97
5041 ----------------------
5042
5043 1. /(b)|(:+)/ was computing an incorrect first character.
5044
5045 2. Add pcre_study() to the API and the passing of pcre_extra to pcre_exec(),
5046 but not actually doing anything yet.
5047
5048 3. Treat "-" characters in classes that cannot be part of ranges as literals,
5049 as Perl does (e.g. [-az] or [az-]).
5050
5051 4. Set the anchored flag if a branch starts with .* or .*? because that tests
5052 all possible positions.
5053
5054 5. Split up into different modules to avoid including unneeded functions in a
5055 compiled binary. However, compile and exec are still in one module. The "study"
5056 function is split off.
5057
5058 6. The character tables are now in a separate module whose source is generated
5059 by an auxiliary program - but can then be edited by hand if required. There are
5060 now no calls to isalnum(), isspace(), isdigit(), isxdigit(), tolower() or
5061 toupper() in the code.
5062
5063 7. Turn the malloc/free funtions variables into pcre_malloc and pcre_free and
5064 make them global. Abolish the function for setting them, as the caller can now
5065 set them directly.
5066
5067
5068 Version 0.92 11-Sep-97
5069 ----------------------
5070
5071 1. A repeat with a fixed maximum and a minimum of 1 for an ordinary character
5072 (e.g. /a{1,3}/) was broken (I mis-optimized it).
5073
5074 2. Caseless matching was not working in character classes if the characters in
5075 the pattern were in upper case.
5076
5077 3. Make ranges like [W-c] work in the same way as Perl for caseless matching.
5078
5079 4. Make PCRE_ANCHORED public and accept as a compile option.
5080
5081 5. Add an options word to pcre_exec() and accept PCRE_ANCHORED and
5082 PCRE_CASELESS at run time. Add escapes \A and \I to pcretest to cause it to
5083 pass them.
5084
5085 6. Give an error if bad option bits passed at compile or run time.
5086
5087 7. Add PCRE_MULTILINE at compile and exec time, and (?m) as well. Add \M to
5088 pcretest to cause it to pass that flag.
5089
5090 8. Add pcre_info(), to get the number of identifying subpatterns, the stored
5091 options, and the first character, if set.
5092
5093 9. Recognize C+ or C{n,m} where n >= 1 as providing a fixed starting character.
5094
5095
5096 Version 0.91 10-Sep-97
5097 ----------------------
5098
5099 1. PCRE was failing to diagnose unlimited repeats of subpatterns that could
5100 match the empty string as in /(a*)*/. It was looping and ultimately crashing.
5101
5102 2. PCRE was looping on encountering an indefinitely repeated back reference to
5103 a subpattern that had matched an empty string, e.g. /(a|)\1*/. It now does what
5104 Perl does - treats the match as successful.
5105
5106 ****

Properties

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

  ViewVC Help
Powered by ViewVC 1.1.5