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

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1035 - (show annotations)
Mon Sep 10 16:23:12 2012 UTC (7 years, 2 months ago) by ph10
File size: 235659 byte(s)
Error occurred while calculating annotation data.
Make pcregrep use PCRE_STUDY_EXTRA_NEEDED.
1 ChangeLog for PCRE
2 ------------------
3
4 Version 8.32
5 ------------
6
7 1. Improved JIT compiler optimizations for first character search and single
8 character iterators.
9
10 2. Supporting IBM XL C compilers for PPC architectures in the JIT compiler.
11 Patch by Daniel Richard G.
12
13 3. Single character iterator optimizations in the JIT compiler.
14
15 4. Improved JIT compiler optimizations for character ranges.
16
17 5. Rename the "leave" variable names to "quit" to improve WinCE compatibility.
18 Reported by Giuseppe D'Angelo.
19
20 6. The PCRE_STARTLINE bit, indicating that a match can occur only at the start
21 of a line, was being set incorrectly in cases where .* appeared inside
22 atomic brackets at the start of a pattern, or where there was a subsequent
23 *PRUNE or *SKIP.
24
25 7. Improved instruction cache flush for POWER/PowerPC.
26 Patch by Daniel Richard G.
27
28 8. Fixed a number of issues in pcregrep, making it more compatible with GNU
29 grep:
30
31 (a) There is now no limit to the number of patterns to be matched.
32
33 (b) An error is given if a pattern is too long.
34
35 (c) Multiple uses of --exclude, --exclude-dir, --include, and --include-dir
36 are now supported.
37
38 (d) --exclude-from and --include-from (multiple use) have been added.
39
40 (e) Exclusions and inclusions now apply to all files and directories, not
41 just to those obtained from scanning a directory recursively.
42
43 (f) Multiple uses of -f and --file-list are now supported.
44
45 (g) In a Windows environment, the default for -d has been changed from
46 "read" (the GNU grep default) to "skip", because otherwise the presence
47 of a directory in the file list provokes an error.
48
49 (h) The documentation has been revised and clarified in places.
50
51 9. Improve the matching speed of capturing brackets.
52
53 10. Changed the meaning of \X so that it now matches a Unicode extended
54 grapheme cluster.
55
56 11. Patch by Daniel Richard G to the autoconf files to add a macro for sorting
57 out POSIX threads when JIT support is configured.
58
59 12. Added support for PCRE_STUDY_EXTRA_NEEDED.
60
61 13. In the POSIX wrapper regcomp() function, setting re_nsub field in the preg
62 structure could go wrong in environments where size_t is not the same size
63 as int.
64
65 14. Applied user-supplied patch to pcrecpp.cc to allow PCRE_NO_UTF8_CHECK to be
66 set.
67
68 15. The EBCDIC support had decayed; later updates to the code had included
69 explicit references to (e.g.) \x0a instead of CHAR_LF. There has been a
70 general tidy up of EBCDIC-related issues, and the documentation was also
71 not quite right. There is now a test that can be run on ASCII systems to
72 check some of the EBCDIC-related things (but is it not a full test).
73
74 16. The new PCRE_STUDY_EXTRA_NEEDED option is now used by pcregrep, resulting
75 in a small tidy to the code.
76
77
78
79 Version 8.31 06-July-2012
80 -------------------------
81
82 1. Fixing a wrong JIT test case and some compiler warnings.
83
84 2. Removed a bashism from the RunTest script.
85
86 3. Add a cast to pcre_exec.c to fix the warning "unary minus operator applied
87 to unsigned type, result still unsigned" that was given by an MS compiler
88 on encountering the code "-sizeof(xxx)".
89
90 4. Partial matching support is added to the JIT compiler.
91
92 5. Fixed several bugs concerned with partial matching of items that consist
93 of more than one character:
94
95 (a) /^(..)\1/ did not partially match "aba" because checking references was
96 done on an "all or nothing" basis. This also applied to repeated
97 references.
98
99 (b) \R did not give a hard partial match if \r was found at the end of the
100 subject.
101
102 (c) \X did not give a hard partial match after matching one or more
103 characters at the end of the subject.
104
105 (d) When newline was set to CRLF, a pattern such as /a$/ did not recognize
106 a partial match for the string "\r".
107
108 (e) When newline was set to CRLF, the metacharacter "." did not recognize
109 a partial match for a CR character at the end of the subject string.
110
111 6. If JIT is requested using /S++ or -s++ (instead of just /S+ or -s+) when
112 running pcretest, the text "(JIT)" added to the output whenever JIT is
113 actually used to run the match.
114
115 7. Individual JIT compile options can be set in pcretest by following -s+[+]
116 or /S+[+] with a digit between 1 and 7.
117
118 8. OP_NOT now supports any UTF character not just single-byte ones.
119
120 9. (*MARK) control verb is now supported by the JIT compiler.
121
122 10. The command "./RunTest list" lists the available tests without actually
123 running any of them. (Because I keep forgetting what they all are.)
124
125 11. Add PCRE_INFO_MAXLOOKBEHIND.
126
127 12. Applied a (slightly modified) user-supplied patch that improves performance
128 when the heap is used for recursion (compiled with --disable-stack-for-
129 recursion). Instead of malloc and free for each heap frame each time a
130 logical recursion happens, frames are retained on a chain and re-used where
131 possible. This sometimes gives as much as 30% improvement.
132
133 13. As documented, (*COMMIT) is now confined to within a recursive subpattern
134 call.
135
136 14. As documented, (*COMMIT) is now confined to within a positive assertion.
137
138 15. It is now possible to link pcretest with libedit as an alternative to
139 libreadline.
140
141 16. (*COMMIT) control verb is now supported by the JIT compiler.
142
143 17. The Unicode data tables have been updated to Unicode 6.1.0.
144
145 18. Added --file-list option to pcregrep.
146
147 19. Added binary file support to pcregrep, including the -a, --binary-files,
148 -I, and --text options.
149
150 20. The madvise function is renamed for posix_madvise for QNX compatibility
151 reasons. Fixed by Giuseppe D'Angelo.
152
153 21. Fixed a bug for backward assertions with REVERSE 0 in the JIT compiler.
154
155 22. Changed the option for creating symbolic links for 16-bit man pages from
156 -s to -sf so that re-installing does not cause issues.
157
158 23. Support PCRE_NO_START_OPTIMIZE in JIT as (*MARK) support requires it.
159
160 24. Fixed a very old bug in pcretest that caused errors with restarted DFA
161 matches in certain environments (the workspace was not being correctly
162 retained). Also added to pcre_dfa_exec() a simple plausibility check on
163 some of the workspace data at the beginning of a restart.
164
165 25. \s*\R was auto-possessifying the \s* when it should not, whereas \S*\R
166 was not doing so when it should - probably a typo introduced by SVN 528
167 (change 8.10/14).
168
169 26. When PCRE_UCP was not set, \w+\x{c4} was incorrectly auto-possessifying the
170 \w+ when the character tables indicated that \x{c4} was a word character.
171 There were several related cases, all because the tests for doing a table
172 lookup were testing for characters less than 127 instead of 255.
173
174 27. If a pattern contains capturing parentheses that are not used in a match,
175 their slots in the ovector are set to -1. For those that are higher than
176 any matched groups, this happens at the end of processing. In the case when
177 there were back references that the ovector was too small to contain
178 (causing temporary malloc'd memory to be used during matching), and the
179 highest capturing number was not used, memory off the end of the ovector
180 was incorrectly being set to -1. (It was using the size of the temporary
181 memory instead of the true size.)
182
183 28. To catch bugs like 27 using valgrind, when pcretest is asked to specify an
184 ovector size, it uses memory at the end of the block that it has got.
185
186 29. Check for an overlong MARK name and give an error at compile time. The
187 limit is 255 for the 8-bit library and 65535 for the 16-bit library.
188
189 30. JIT compiler update.
190
191 31. JIT is now supported on jailbroken iOS devices. Thanks for Ruiger
192 Rill for the patch.
193
194 32. Put spaces around SLJIT_PRINT_D in the JIT compiler. Required by CXX11.
195
196 33. Variable renamings in the PCRE-JIT compiler. No functionality change.
197
198 34. Fixed typos in pcregrep: in two places there was SUPPORT_LIBZ2 instead of
199 SUPPORT_LIBBZ2. This caused a build problem when bzip2 but not gzip (zlib)
200 was enabled.
201
202 35. Improve JIT code generation for greedy plus quantifier.
203
204 36. When /((?:a?)*)*c/ or /((?>a?)*)*c/ was matched against "aac", it set group
205 1 to "aa" instead of to an empty string. The bug affected repeated groups
206 that could potentially match an empty string.
207
208 37. Optimizing single character iterators in JIT.
209
210 38. Wide characters specified with \uxxxx in JavaScript mode are now subject to
211 the same checks as \x{...} characters in non-JavaScript mode. Specifically,
212 codepoints that are too big for the mode are faulted, and in a UTF mode,
213 disallowed codepoints are also faulted.
214
215 39. If PCRE was compiled with UTF support, in three places in the DFA
216 matcher there was code that should only have been obeyed in UTF mode, but
217 was being obeyed unconditionally. In 8-bit mode this could cause incorrect
218 processing when bytes with values greater than 127 were present. In 16-bit
219 mode the bug would be provoked by values in the range 0xfc00 to 0xdc00. In
220 both cases the values are those that cannot be the first data item in a UTF
221 character. The three items that might have provoked this were recursions,
222 possessively repeated groups, and atomic groups.
223
224 40. Ensure that libpcre is explicitly listed in the link commands for pcretest
225 and pcregrep, because some OS require shared objects to be explicitly
226 passed to ld, causing the link step to fail if they are not.
227
228 41. There were two incorrect #ifdefs in pcre_study.c, meaning that, in 16-bit
229 mode, patterns that started with \h* or \R* might be incorrectly matched.
230
231
232 Version 8.30 04-February-2012
233 -----------------------------
234
235 1. Renamed "isnumber" as "is_a_number" because in some Mac environments this
236 name is defined in ctype.h.
237
238 2. Fixed a bug in fixed-length calculation for lookbehinds that would show up
239 only in quite long subpatterns.
240
241 3. Removed the function pcre_info(), which has been obsolete and deprecated
242 since it was replaced by pcre_fullinfo() in February 2000.
243
244 4. For a non-anchored pattern, if (*SKIP) was given with a name that did not
245 match a (*MARK), and the match failed at the start of the subject, a
246 reference to memory before the start of the subject could occur. This bug
247 was introduced by fix 17 of release 8.21.
248
249 5. A reference to an unset group with zero minimum repetition was giving
250 totally wrong answers (in non-JavaScript-compatibility mode). For example,
251 /(another)?(\1?)test/ matched against "hello world test". This bug was
252 introduced in release 8.13.
253
254 6. Add support for 16-bit character strings (a large amount of work involving
255 many changes and refactorings).
256
257 7. RunGrepTest failed on msys because \r\n was replaced by whitespace when the
258 command "pattern=`printf 'xxx\r\njkl'`" was run. The pattern is now taken
259 from a file.
260
261 8. Ovector size of 2 is also supported by JIT based pcre_exec (the ovector size
262 rounding is not applied in this particular case).
263
264 9. The invalid Unicode surrogate codepoints U+D800 to U+DFFF are now rejected
265 if they appear, or are escaped, in patterns.
266
267 10. Get rid of a number of -Wunused-but-set-variable warnings.
268
269 11. The pattern /(?=(*:x))(q|)/ matches an empty string, and returns the mark
270 "x". The similar pattern /(?=(*:x))((*:y)q|)/ did not return a mark at all.
271 Oddly, Perl behaves the same way. PCRE has been fixed so that this pattern
272 also returns the mark "x". This bug applied to capturing parentheses,
273 non-capturing parentheses, and atomic parentheses. It also applied to some
274 assertions.
275
276 12. Stephen Kelly's patch to CMakeLists.txt allows it to parse the version
277 information out of configure.ac instead of relying on pcre.h.generic, which
278 is not stored in the repository.
279
280 13. Applied Dmitry V. Levin's patch for a more portable method for linking with
281 -lreadline.
282
283 14. ZH added PCRE_CONFIG_JITTARGET; added its output to pcretest -C.
284
285 15. Applied Graycode's patch to put the top-level frame on the stack rather
286 than the heap when not using the stack for recursion. This gives a
287 performance improvement in many cases when recursion is not deep.
288
289 16. Experimental code added to "pcretest -C" to output the stack frame size.
290
291
292 Version 8.21 12-Dec-2011
293 ------------------------
294
295 1. Updating the JIT compiler.
296
297 2. JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
298 are added as well.
299
300 3. Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
301 PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
302 calling _pcre_jit_exec. Some extra comments are added.
303
304 4. (*MARK) settings inside atomic groups that do not contain any capturing
305 parentheses, for example, (?>a(*:m)), were not being passed out. This bug
306 was introduced by change 18 for 8.20.
307
308 5. Supporting of \x, \U and \u in JavaScript compatibility mode based on the
309 ECMA-262 standard.
310
311 6. Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
312 erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
313 This bug was probably introduced by change 9 of 8.13.
314
315 7. While fixing 6 above, I noticed that a number of other items were being
316 incorrectly rejected as "not fixed length". This arose partly because newer
317 opcodes had not been added to the fixed-length checking code. I have (a)
318 corrected the bug and added tests for these items, and (b) arranged for an
319 error to occur if an unknown opcode is encountered while checking for fixed
320 length instead of just assuming "not fixed length". The items that were
321 rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
322 (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
323 repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
324
325 8. A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
326 being incorrectly compiled and would have given unpredicatble results.
327
328 9. A possessively repeated subpattern with minimum repeat count greater than
329 one behaved incorrectly. For example, (A){2,}+ behaved as if it was
330 (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
331 the first (A) could occur when it should not.
332
333 10. Add a cast and remove a redundant test from the code.
334
335 11. JIT should use pcre_malloc/pcre_free for allocation.
336
337 12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
338 best practice nowadays, and helps with cross-compiling. (If the exec_prefix
339 is anything other than /usr, -L is still shown).
340
341 13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
342
343 14. Perl does not support \N without a following name in a [] class; PCRE now
344 also gives an error.
345
346 15. If a forward reference was repeated with an upper limit of around 2000,
347 it caused the error "internal error: overran compiling workspace". The
348 maximum number of forward references (including repeats) was limited by the
349 internal workspace, and dependent on the LINK_SIZE. The code has been
350 rewritten so that the workspace expands (via pcre_malloc) if necessary, and
351 the default depends on LINK_SIZE. There is a new upper limit (for safety)
352 of around 200,000 forward references. While doing this, I also speeded up
353 the filling in of repeated forward references.
354
355 16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
356 incorrectly expecting the subject to contain another "a" after the start.
357
358 17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
359 in the match, the SKIP should be ignored. This was not happening; instead
360 the SKIP was being treated as NOMATCH. For patterns such as
361 /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
362 tested.
363
364 18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
365 now much more compatible with Perl, in particular in cases where the result
366 is a non-match for a non-anchored pattern. For example, if
367 /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
368 "m", where previously it did not return a name. A side effect of this
369 change is that for partial matches, the last encountered mark name is
370 returned, as for non matches. A number of tests that were previously not
371 Perl-compatible have been moved into the Perl-compatible test files. The
372 refactoring has had the pleasing side effect of removing one argument from
373 the match() function, thus reducing its stack requirements.
374
375 19. If the /S+ option was used in pcretest to study a pattern using JIT,
376 subsequent uses of /S (without +) incorrectly behaved like /S+.
377
378 21. Retrieve executable code size support for the JIT compiler and fixing
379 some warnings.
380
381 22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
382 not work when the shorter character appeared right at the end of the
383 subject string.
384
385 23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
386 systems.
387
388 24. Added PCRE_INFO_JITSIZE to pass on the value from (21) above, and also
389 output it when the /M option is used in pcretest.
390
391 25. The CheckMan script was not being included in the distribution. Also, added
392 an explicit "perl" to run Perl scripts from the PrepareRelease script
393 because this is reportedly needed in Windows.
394
395 26. If study data was being save in a file and studying had not found a set of
396 "starts with" bytes for the pattern, the data written to the file (though
397 never used) was taken from uninitialized memory and so caused valgrind to
398 complain.
399
400 27. Updated RunTest.bat as provided by Sheri Pierce.
401
402 28. Fixed a possible uninitialized memory bug in pcre_jit_compile.c.
403
404 29. Computation of memory usage for the table of capturing group names was
405 giving an unnecessarily large value.
406
407
408 Version 8.20 21-Oct-2011
409 ------------------------
410
411 1. Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
412 a POSIX class. After further experiments with Perl, which convinced me that
413 Perl has bugs and confusions, a closing square bracket is no longer allowed
414 in a POSIX name. This bug also affected patterns with classes that started
415 with full stops.
416
417 2. If a pattern such as /(a)b|ac/ is matched against "ac", there is no
418 captured substring, but while checking the failing first alternative,
419 substring 1 is temporarily captured. If the output vector supplied to
420 pcre_exec() was not big enough for this capture, the yield of the function
421 was still zero ("insufficient space for captured substrings"). This cannot
422 be totally fixed without adding another stack variable, which seems a lot
423 of expense for a edge case. However, I have improved the situation in cases
424 such as /(a)(b)x|abc/ matched against "abc", where the return code
425 indicates that fewer than the maximum number of slots in the ovector have
426 been set.
427
428 3. Related to (2) above: when there are more back references in a pattern than
429 slots in the output vector, pcre_exec() uses temporary memory during
430 matching, and copies in the captures as far as possible afterwards. It was
431 using the entire output vector, but this conflicts with the specification
432 that only 2/3 is used for passing back captured substrings. Now it uses
433 only the first 2/3, for compatibility. This is, of course, another edge
434 case.
435
436 4. Zoltan Herczeg's just-in-time compiler support has been integrated into the
437 main code base, and can be used by building with --enable-jit. When this is
438 done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
439 runtime --no-jit option is given.
440
441 5. When the number of matches in a pcre_dfa_exec() run exactly filled the
442 ovector, the return from the function was zero, implying that there were
443 other matches that did not fit. The correct "exactly full" value is now
444 returned.
445
446 6. If a subpattern that was called recursively or as a subroutine contained
447 (*PRUNE) or any other control that caused it to give a non-standard return,
448 invalid errors such as "Error -26 (nested recursion at the same subject
449 position)" or even infinite loops could occur.
450
451 7. If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
452 computing the minimum length on reaching *ACCEPT, and so ended up with the
453 wrong value of 1 rather than 0. Further investigation indicates that
454 computing a minimum subject length in the presence of *ACCEPT is difficult
455 (think back references, subroutine calls), and so I have changed the code
456 so that no minimum is registered for a pattern that contains *ACCEPT.
457
458 8. If (*THEN) was present in the first (true) branch of a conditional group,
459 it was not handled as intended. [But see 16 below.]
460
461 9. Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
462 Sheri Pierce.
463
464 10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
465 the first byte in a match must be "a".
466
467 11. Change 17 for 8.13 increased the recursion depth for patterns like
468 /a(?:.)*?a/ drastically. I've improved things by remembering whether a
469 pattern contains any instances of (*THEN). If it does not, the old
470 optimizations are restored. It would be nice to do this on a per-group
471 basis, but at the moment that is not feasible.
472
473 12. In some environments, the output of pcretest -C is CRLF terminated. This
474 broke RunTest's code that checks for the link size. A single white space
475 character after the value is now allowed for.
476
477 13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
478 For "fr", it uses the Windows-specific input and output files.
479
480 14. If (*THEN) appeared in a group that was called recursively or as a
481 subroutine, it did not work as intended. [But see next item.]
482
483 15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
484 pattern fragments (but not containing any | characters). If A and B are
485 matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
486 was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
487 D. In other words, Perl considers parentheses that do not contain any |
488 characters to be part of a surrounding alternative, whereas PCRE was
489 treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
490 differently. PCRE now behaves in the same way as Perl, except in the case
491 of subroutine/recursion calls such as (?1) which have in any case always
492 been different (but PCRE had them first :-).
493
494 16. Related to 15 above: Perl does not treat the | in a conditional group as
495 creating alternatives. Such a group is treated in the same way as an
496 ordinary group without any | characters when processing (*THEN). PCRE has
497 been changed to match Perl's behaviour.
498
499 17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
500 RunGrepTest script failed.
501
502 18. Change 22 for version 13 caused atomic groups to use more stack. This is
503 inevitable for groups that contain captures, but it can lead to a lot of
504 stack use in large patterns. The old behaviour has been restored for atomic
505 groups that do not contain any capturing parentheses.
506
507 19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
508 suppress the check for a minimum subject length at run time. (If it was
509 given to pcre_exec() or pcre_dfa_exec() it did work.)
510
511 20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
512 fail to work when decoding hex characters in data strings in EBCDIC
513 environments.
514
515 21. It appears that in at least one Mac OS environment, the isxdigit() function
516 is implemented as a macro that evaluates to its argument more than once,
517 contravening the C 90 Standard (I haven't checked a later standard). There
518 was an instance in pcretest which caused it to go wrong when processing
519 \x{...} escapes in subject strings. The has been rewritten to avoid using
520 things like p++ in the argument of isxdigit().
521
522
523 Version 8.13 16-Aug-2011
524 ------------------------
525
526 1. The Unicode data tables have been updated to Unicode 6.0.0.
527
528 2. Two minor typos in pcre_internal.h have been fixed.
529
530 3. Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
531 pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
532 in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
533
534 4. There were a number of related bugs in the code for matching backrefences
535 caselessly in UTF-8 mode when codes for the characters concerned were
536 different numbers of bytes. For example, U+023A and U+2C65 are an upper
537 and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
538 (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
539 code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
540 2-byte code at the end of the subject (it thought there wasn't enough data
541 left).
542
543 5. Comprehensive information about what went wrong is now returned by
544 pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
545 as the output vector has at least 2 elements. The offset of the start of
546 the failing character and a reason code are placed in the vector.
547
548 6. When the UTF-8 string check fails for pcre_compile(), the offset that is
549 now returned is for the first byte of the failing character, instead of the
550 last byte inspected. This is an incompatible change, but I hope it is small
551 enough not to be a problem. It makes the returned offset consistent with
552 pcre_exec() and pcre_dfa_exec().
553
554 7. pcretest now gives a text phrase as well as the error number when
555 pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
556 failure, the offset and reason code are output.
557
558 8. When \R was used with a maximizing quantifier it failed to skip backwards
559 over a \r\n pair if the subsequent match failed. Instead, it just skipped
560 back over a single character (\n). This seems wrong (because it treated the
561 two characters as a single entity when going forwards), conflicts with the
562 documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
563 behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
564 has been changed.
565
566 9. Some internal refactoring has changed the processing so that the handling
567 of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
568 time (the PCRE_DOTALL option was changed this way some time ago: version
569 7.7 change 16). This has made it possible to abolish the OP_OPT op code,
570 which was always a bit of a fudge. It also means that there is one less
571 argument for the match() function, which reduces its stack requirements
572 slightly. This change also fixes an incompatibility with Perl: the pattern
573 (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
574
575 10. More internal refactoring has drastically reduced the number of recursive
576 calls to match() for possessively repeated groups such as (abc)++ when
577 using pcre_exec().
578
579 11. While implementing 10, a number of bugs in the handling of groups were
580 discovered and fixed:
581
582 (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
583 (a|)*(?1) gave a compile-time internal error.
584 ((a|)+)+ did not notice that the outer group could match an empty string.
585 (^a|^)+ was not marked as anchored.
586 (.*a|.*)+ was not marked as matching at start or after a newline.
587
588 12. Yet more internal refactoring has removed another argument from the match()
589 function. Special calls to this function are now indicated by setting a
590 value in a variable in the "match data" data block.
591
592 13. Be more explicit in pcre_study() instead of relying on "default" for
593 opcodes that mean there is no starting character; this means that when new
594 ones are added and accidentally left out of pcre_study(), testing should
595 pick them up.
596
597 14. The -s option of pcretest has been documented for ages as being an old
598 synonym of -m (show memory usage). I have changed it to mean "force study
599 for every regex", that is, assume /S for every regex. This is similar to -i
600 and -d etc. It's slightly incompatible, but I'm hoping nobody is still
601 using it. It makes it easier to run collections of tests with and without
602 study enabled, and thereby test pcre_study() more easily. All the standard
603 tests are now run with and without -s (but some patterns can be marked as
604 "never study" - see 20 below).
605
606 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
607 restoration of the capturing data to the outer values was not happening
608 correctly.
609
610 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
611 empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
612 pattern had matched an empty string, and so incorrectly returned a no
613 match.
614
615 17. There was optimizing code for the last branch of non-capturing parentheses,
616 and also for the obeyed branch of a conditional subexpression, which used
617 tail recursion to cut down on stack usage. Unfortunately, now that there is
618 the possibility of (*THEN) occurring in these branches, tail recursion is
619 no longer possible because the return has to be checked for (*THEN). These
620 two optimizations have therefore been removed. [But see 8.20/11 above.]
621
622 18. If a pattern containing \R was studied, it was assumed that \R always
623 matched two bytes, thus causing the minimum subject length to be
624 incorrectly computed because \R can also match just one byte.
625
626 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
627 was incorrectly computed.
628
629 20. If /S is present twice on a test pattern in pcretest input, it now
630 *disables* studying, thereby overriding the use of -s on the command line
631 (see 14 above). This is necessary for one or two tests to keep the output
632 identical in both cases.
633
634 21. When (*ACCEPT) was used in an assertion that matched an empty string and
635 PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
636
637 22. When an atomic group that contained a capturing parenthesis was
638 successfully matched, but the branch in which it appeared failed, the
639 capturing was not being forgotten if a higher numbered group was later
640 captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
641 group 1 to "a", when in fact it should be unset. This applied to multi-
642 branched capturing and non-capturing groups, repeated or not, and also to
643 positive assertions (capturing in negative assertions does not happen
644 in PCRE) and also to nested atomic groups.
645
646 23. Add the ++ qualifier feature to pcretest, to show the remainder of the
647 subject after a captured substring, to make it easier to tell which of a
648 number of identical substrings has been captured.
649
650 24. The way atomic groups are processed by pcre_exec() has been changed so that
651 if they are repeated, backtracking one repetition now resets captured
652 values correctly. For example, if ((?>(a+)b)+aabab) is matched against
653 "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
654 "aaa". Previously, it would have been "a". As part of this code
655 refactoring, the way recursive calls are handled has also been changed.
656
657 25. If an assertion condition captured any substrings, they were not passed
658 back unless some other capturing happened later. For example, if
659 (?(?=(a))a) was matched against "a", no capturing was returned.
660
661 26. When studying a pattern that contained subroutine calls or assertions,
662 the code for finding the minimum length of a possible match was handling
663 direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
664 group 1 called group 2 while simultaneously a separate group 2 called group
665 1). A stack overflow occurred in this case. I have fixed this by limiting
666 the recursion depth to 10.
667
668 27. Updated RunTest.bat in the distribution to the version supplied by Tom
669 Fortmann. This supports explicit test numbers on the command line, and has
670 argument validation and error reporting.
671
672 28. An instance of \X with an unlimited repeat could fail if at any point the
673 first character it looked at was a mark character.
674
675 29. Some minor code refactoring concerning Unicode properties and scripts
676 should reduce the stack requirement of match() slightly.
677
678 30. Added the '=' option to pcretest to check the setting of unused capturing
679 slots at the end of the pattern, which are documented as being -1, but are
680 not included in the return count.
681
682 31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
683 compiled something random. Now it gives a compile-time error (as does
684 Perl).
685
686 32. A *MARK encountered during the processing of a positive assertion is now
687 recorded and passed back (compatible with Perl).
688
689 33. If --only-matching or --colour was set on a pcregrep call whose pattern
690 had alternative anchored branches, the search for a second match in a line
691 was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
692 matched the line "0102" twice. The same bug affected patterns that started
693 with a backwards assertion. For example /\b01|\b02/ also matched "0102"
694 twice.
695
696 34. Previously, PCRE did not allow quantification of assertions. However, Perl
697 does, and because of capturing effects, quantifying parenthesized
698 assertions may at times be useful. Quantifiers are now allowed for
699 parenthesized assertions.
700
701 35. A minor code tidy in pcre_compile() when checking options for \R usage.
702
703 36. \g was being checked for fancy things in a character class, when it should
704 just be a literal "g".
705
706 37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
707 appearance of a nested POSIX class supersedes an apparent external class.
708 For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
709 unescaped square brackets may also appear as part of class names. For
710 example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
711 more like Perl. (But see 8.20/1 above.)
712
713 38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
714 was because it thought it was \N{name}, which is not supported).
715
716 39. Add minix to OS list not supporting the -S option in pcretest.
717
718 40. PCRE tries to detect cases of infinite recursion at compile time, but it
719 cannot analyze patterns in sufficient detail to catch mutual recursions
720 such as ((?1))((?2)). There is now a runtime test that gives an error if a
721 subgroup is called recursively as a subpattern for a second time at the
722 same position in the subject string. In previous releases this might have
723 been caught by the recursion limit, or it might have run out of stack.
724
725 41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
726 happen only once. PCRE was, however incorrectly giving a compile time error
727 "recursive call could loop indefinitely" because it cannot analyze the
728 pattern in sufficient detail. The compile time test no longer happens when
729 PCRE is compiling a conditional subpattern, but actual runaway loops are
730 now caught at runtime (see 40 above).
731
732 42. It seems that Perl allows any characters other than a closing parenthesis
733 to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
734 has been changed to be the same.
735
736 43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
737 as not to get warnings when autogen.sh is called. Also changed
738 AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
739
740 44. To help people who use pcregrep to scan files containing exceedingly long
741 lines, the following changes have been made:
742
743 (a) The default value of the buffer size parameter has been increased from
744 8K to 20K. (The actual buffer used is three times this size.)
745
746 (b) The default can be changed by ./configure --with-pcregrep-bufsize when
747 PCRE is built.
748
749 (c) A --buffer-size=n option has been added to pcregrep, to allow the size
750 to be set at run time.
751
752 (d) Numerical values in pcregrep options can be followed by K or M, for
753 example --buffer-size=50K.
754
755 (e) If a line being scanned overflows pcregrep's buffer, an error is now
756 given and the return code is set to 2.
757
758 45. Add a pointer to the latest mark to the callout data block.
759
760 46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
761 partial match of an empty string instead of no match. This was specific to
762 the use of ".".
763
764 47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
765 complete match instead of a partial match. This bug was dependent on both
766 the PCRE_UTF8 and PCRE_DOTALL options being set.
767
768 48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
769 starting byte set, because \b was not being ignored.
770
771
772 Version 8.12 15-Jan-2011
773 ------------------------
774
775 1. Fixed some typos in the markup of the man pages, and wrote a script that
776 checks for such things as part of the documentation building process.
777
778 2. On a big-endian 64-bit system, pcregrep did not correctly process the
779 --match-limit and --recursion-limit options (added for 8.11). In
780 particular, this made one of the standard tests fail. (The integer value
781 went into the wrong half of a long int.)
782
783 3. If the --colour option was given to pcregrep with -v (invert match), it
784 did strange things, either producing crazy output, or crashing. It should,
785 of course, ignore a request for colour when reporting lines that do not
786 match.
787
788 4. Another pcregrep bug caused similar problems if --colour was specified with
789 -M (multiline) and the pattern match finished with a line ending.
790
791 5. In pcregrep, when a pattern that ended with a literal newline sequence was
792 matched in multiline mode, the following line was shown as part of the
793 match. This seems wrong, so I have changed it.
794
795 6. Another pcregrep bug in multiline mode, when --colour was specified, caused
796 the check for further matches in the same line (so they could be coloured)
797 to overrun the end of the current line. If another match was found, it was
798 incorrectly shown (and then shown again when found in the next line).
799
800 7. If pcregrep was compiled under Windows, there was a reference to the
801 function pcregrep_exit() before it was defined. I am assuming this was
802 the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
803 reported by a user. I've moved the definition above the reference.
804
805
806 Version 8.11 10-Dec-2010
807 ------------------------
808
809 1. (*THEN) was not working properly if there were untried alternatives prior
810 to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
811 backtracked to try for "b" instead of moving to the next alternative branch
812 at the same level (in this case, to look for "c"). The Perl documentation
813 is clear that when (*THEN) is backtracked onto, it goes to the "next
814 alternative in the innermost enclosing group".
815
816 2. (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
817 such as (A(*COMMIT)B(*THEN)C|D) any failure after matching A should
818 result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
819 (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
820 (*THEN).
821
822 3. If \s appeared in a character class, it removed the VT character from
823 the class, even if it had been included by some previous item, for example
824 in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
825 of \s, but is part of the POSIX "space" class.)
826
827 4. A partial match never returns an empty string (because you can always
828 match an empty string at the end of the subject); however the checking for
829 an empty string was starting at the "start of match" point. This has been
830 changed to the "earliest inspected character" point, because the returned
831 data for a partial match starts at this character. This means that, for
832 example, /(?<=abc)def/ gives a partial match for the subject "abc"
833 (previously it gave "no match").
834
835 5. Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
836 of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
837 previously a full match would be given. However, setting PCRE_PARTIAL_HARD
838 has an implication that the given string is incomplete (because a partial
839 match is preferred over a full match). For this reason, these items now
840 give a partial match in this situation. [Aside: previously, the one case
841 /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
842 match rather than a full match, which was wrong by the old rules, but is
843 now correct.]
844
845 6. There was a bug in the handling of #-introduced comments, recognized when
846 PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
847 If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
848 UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
849 scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
850 but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
851 places in pcre_compile().
852
853 7. Related to (6) above, when pcre_compile() was skipping #-introduced
854 comments when looking ahead for named forward references to subpatterns,
855 the only newline sequence it recognized was NL. It now handles newlines
856 according to the set newline convention.
857
858 8. SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
859 former, but used strtoul(), whereas pcretest avoided strtoul() but did not
860 cater for a lack of strerror(). These oversights have been fixed.
861
862 9. Added --match-limit and --recursion-limit to pcregrep.
863
864 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
865
866 11. When the -o option was used, pcregrep was setting a return code of 1, even
867 when matches were found, and --line-buffered was not being honoured.
868
869 12. Added an optional parentheses number to the -o and --only-matching options
870 of pcregrep.
871
872 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
873 can match an empty string. The code to do it in pcretest and pcredemo
874 needed fixing:
875
876 (a) When the newline convention was "crlf", pcretest got it wrong, skipping
877 only one byte after an empty string match just before CRLF (this case
878 just got forgotten; "any" and "anycrlf" were OK).
879
880 (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
881 mode when an empty string match preceded an ASCII character followed by
882 a non-ASCII character. (The code for advancing by one character rather
883 than one byte was nonsense.)
884
885 (c) The pcredemo.c sample program did not have any code at all to handle
886 the cases when CRLF is a valid newline sequence.
887
888 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
889 as a starting offset was within the subject string. There is now a new
890 error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
891 negative or greater than the length of the string. In order to test this,
892 pcretest is extended to allow the setting of negative starting offsets.
893
894 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
895 starting offset points to the beginning of a UTF-8 character was
896 unnecessarily clumsy. I tidied it up.
897
898 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
899 bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
900
901 17. Nobody had reported that the --include_dir option, which was added in
902 release 7.7 should have been called --include-dir (hyphen, not underscore)
903 for compatibility with GNU grep. I have changed it to --include-dir, but
904 left --include_dir as an undocumented synonym, and the same for
905 --exclude-dir, though that is not available in GNU grep, at least as of
906 release 2.5.4.
907
908 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
909 characters from a string of bytes) have been redefined so as not to use
910 loops, in order to improve performance in some environments. At the same
911 time, I abstracted some of the common code into auxiliary macros to save
912 repetition (this should not affect the compiled code).
913
914 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
915 compile-time error is now given if \c is not followed by an ASCII
916 character, that is, a byte less than 128. (In EBCDIC mode, the code is
917 different, and any byte value is allowed.)
918
919 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
920 START_OPTIMIZE option, which is now allowed at compile time - but just
921 passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
922 to pcregrep and other applications that have no direct access to PCRE
923 options. The new /Y option in pcretest sets this option when calling
924 pcre_compile().
925
926 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
927 back references. Groups containing recursive back references were forced to
928 be atomic by that change, but in the case of named groups, the amount of
929 memory required was incorrectly computed, leading to "Failed: internal
930 error: code overflow". This has been fixed.
931
932 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
933 pcretest.c, to avoid build problems in some Borland environments.
934
935
936 Version 8.10 25-Jun-2010
937 ------------------------
938
939 1. Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
940 THEN.
941
942 2. (*ACCEPT) was not working when inside an atomic group.
943
944 3. Inside a character class, \B is treated as a literal by default, but
945 faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
946 causes the error). The code is unchanged, but I tidied the documentation.
947
948 4. Inside a character class, PCRE always treated \R and \X as literals,
949 whereas Perl faults them if its -w option is set. I have changed PCRE so
950 that it faults them when PCRE_EXTRA is set.
951
952 5. Added support for \N, which always matches any character other than
953 newline. (It is the same as "." when PCRE_DOTALL is not set.)
954
955 6. When compiling pcregrep with newer versions of gcc which may have
956 FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
957 declared with attribute warn_unused_result" were given. Just casting the
958 result to (void) does not stop the warnings; a more elaborate fudge is
959 needed. I've used a macro to implement this.
960
961 7. Minor change to pcretest.c to avoid a compiler warning.
962
963 8. Added four artifical Unicode properties to help with an option to make
964 \s etc use properties (see next item). The new properties are: Xan
965 (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
966
967 9. Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
968 use Unicode properties. (*UCP) at the start of a pattern can be used to set
969 this option. Modified pcretest to add /W to test this facility. Added
970 REG_UCP to make it available via the POSIX interface.
971
972 10. Added --line-buffered to pcregrep.
973
974 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
975 studied, and the match started with a letter with a code point greater than
976 127 whose first byte was different to the first byte of the other case of
977 the letter, the other case of this starting letter was not recognized
978 (#976).
979
980 12. If a pattern that was studied started with a repeated Unicode property
981 test, for example, \p{Nd}+, there was the theoretical possibility of
982 setting up an incorrect bitmap of starting bytes, but fortunately it could
983 not have actually happened in practice until change 8 above was made (it
984 added property types that matched character-matching opcodes).
985
986 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
987 possible starting bytes for non-anchored patterns.
988
989 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
990 \R, and also a number of cases that involve Unicode properties, both
991 explicit and implicit when PCRE_UCP is set.
992
993 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
994 input, it could crash or give wrong results if characters with values
995 greater than 0xc0 were present in the subject string. (Detail: it assumed
996 UTF-8 input when processing these items.)
997
998 16. Added a lot of (int) casts to avoid compiler warnings in systems where
999 size_t is 64-bit (#991).
1000
1001 17. Added a check for running out of memory when PCRE is compiled with
1002 --disable-stack-for-recursion (#990).
1003
1004 18. If the last data line in a file for pcretest does not have a newline on
1005 the end, a newline was missing in the output.
1006
1007 19. The default pcre_chartables.c file recognizes only ASCII characters (values
1008 less than 128) in its various bitmaps. However, there is a facility for
1009 generating tables according to the current locale when PCRE is compiled. It
1010 turns out that in some environments, 0x85 and 0xa0, which are Unicode space
1011 characters, are recognized by isspace() and therefore were getting set in
1012 these tables, and indeed these tables seem to approximate to ISO 8859. This
1013 caused a problem in UTF-8 mode when pcre_study() was used to create a list
1014 of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
1015 which of course cannot start UTF-8 characters. I have changed the code so
1016 that only real ASCII characters (less than 128) and the correct starting
1017 bytes for UTF-8 encodings are set for characters greater than 127 when in
1018 UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
1019 altogether.)
1020
1021 20. Added the /T option to pcretest so as to be able to run tests with non-
1022 standard character tables, thus making it possible to include the tests
1023 used for 19 above in the standard set of tests.
1024
1025 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
1026 reference to a subpattern the other side of a comment that contains an
1027 opening parenthesis caused either an internal compiling error, or a
1028 reference to the wrong subpattern.
1029
1030
1031 Version 8.02 19-Mar-2010
1032 ------------------------
1033
1034 1. The Unicode data tables have been updated to Unicode 5.2.0.
1035
1036 2. Added the option --libs-cpp to pcre-config, but only when C++ support is
1037 configured.
1038
1039 3. Updated the licensing terms in the pcregexp.pas file, as agreed with the
1040 original author of that file, following a query about its status.
1041
1042 4. On systems that do not have stdint.h (e.g. Solaris), check for and include
1043 inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
1044
1045 5. A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
1046 quantifier applied to a forward-referencing subroutine call, could compile
1047 incorrect code or give the error "internal error: previously-checked
1048 referenced subpattern not found".
1049
1050 6. Both MS Visual Studio and Symbian OS have problems with initializing
1051 variables to point to external functions. For these systems, therefore,
1052 pcre_malloc etc. are now initialized to local functions that call the
1053 relevant global functions.
1054
1055 7. There were two entries missing in the vectors called coptable and poptable
1056 in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
1057 I've fixed the data, and added a kludgy way of testing at compile time that
1058 the lengths are correct (equal to the number of opcodes).
1059
1060 8. Following on from 7, I added a similar kludge to check the length of the
1061 eint vector in pcreposix.c.
1062
1063 9. Error texts for pcre_compile() are held as one long string to avoid too
1064 much relocation at load time. To find a text, the string is searched,
1065 counting zeros. There was no check for running off the end of the string,
1066 which could happen if a new error number was added without updating the
1067 string.
1068
1069 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
1070
1071 11. \K was not working if it appeared in an atomic group or in a group that
1072 was called as a "subroutine", or in an assertion. Perl 5.11 documents that
1073 \K is "not well defined" if used in an assertion. PCRE now accepts it if
1074 the assertion is positive, but not if it is negative.
1075
1076 12. Change 11 fortuitously reduced the size of the stack frame used in the
1077 "match()" function of pcre_exec.c by one pointer. Forthcoming
1078 implementation of support for (*MARK) will need an extra pointer on the
1079 stack; I have reserved it now, so that the stack frame size does not
1080 decrease.
1081
1082 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
1083 item in branch that calls a recursion is a subroutine call - as in the
1084 second branch in the above example - was incorrectly given the compile-
1085 time error "recursive call could loop indefinitely" because pcre_compile()
1086 was not correctly checking the subroutine for matching a non-empty string.
1087
1088 14. The checks for overrunning compiling workspace could trigger after an
1089 overrun had occurred. This is a "should never occur" error, but it can be
1090 triggered by pathological patterns such as hundreds of nested parentheses.
1091 The checks now trigger 100 bytes before the end of the workspace.
1092
1093 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
1094
1095
1096 Version 8.01 19-Jan-2010
1097 ------------------------
1098
1099 1. If a pattern contained a conditional subpattern with only one branch (in
1100 particular, this includes all (*DEFINE) patterns), a call to pcre_study()
1101 computed the wrong minimum data length (which is of course zero for such
1102 subpatterns). This could cause incorrect "no match" results.
1103
1104 2. For patterns such as (?i)a(?-i)b|c where an option setting at the start of
1105 the pattern is reset in the first branch, pcre_compile() failed with
1106 "internal error: code overflow at offset...". This happened only when
1107 the reset was to the original external option setting. (An optimization
1108 abstracts leading options settings into an external setting, which was the
1109 cause of this.)
1110
1111 3. A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
1112 of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
1113 assertion pattern did not match (meaning that the assertion was true), it
1114 was incorrectly treated as false if the SKIP had been reached during the
1115 matching. This also applied to assertions used as conditions.
1116
1117 4. If an item that is not supported by pcre_dfa_exec() was encountered in an
1118 assertion subpattern, including such a pattern used as a condition,
1119 unpredictable results occurred, instead of the error return
1120 PCRE_ERROR_DFA_UITEM.
1121
1122 5. The C++ GlobalReplace function was not working like Perl for the special
1123 situation when an empty string is matched. It now does the fancy magic
1124 stuff that is necessary.
1125
1126 6. In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
1127 removed. (These were left over from very, very early versions of PCRE.)
1128
1129 7. Some cosmetic changes to the code to make life easier when compiling it
1130 as part of something else:
1131
1132 (a) Change DEBUG to PCRE_DEBUG.
1133
1134 (b) In pcre_compile(), rename the member of the "branch_chain" structure
1135 called "current" as "current_branch", to prevent a collision with the
1136 Linux macro when compiled as a kernel module.
1137
1138 (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
1139 prevent a collision with the Linux macro when compiled as a kernel
1140 module.
1141
1142 8. In pcre_compile() there are some checks for integer overflows that used to
1143 cast potentially large values to (double). This has been changed to that
1144 when building, a check for int64_t is made, and if it is found, it is used
1145 instead, thus avoiding the use of floating point arithmetic. (There is no
1146 other use of FP in PCRE.) If int64_t is not found, the fallback is to
1147 double.
1148
1149 9. Added two casts to avoid signed/unsigned warnings from VS Studio Express
1150 2005 (difference between two addresses compared to an unsigned value).
1151
1152 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
1153 custom one, because of the following reported problem in Windows:
1154
1155 - libbz2 uses the Pascal calling convention (WINAPI) for the functions
1156 under Win32.
1157 - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
1158 therefore missing the function definition.
1159 - The compiler thus generates a "C" signature for the test function.
1160 - The linker fails to find the "C" function.
1161 - PCRE fails to configure if asked to do so against libbz2.
1162
1163 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
1164 messages were output:
1165
1166 Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
1167 rerunning libtoolize, to keep the correct libtool macros in-tree.
1168 Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
1169
1170 I have done both of these things.
1171
1172 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
1173 most of the time, it *can* run out if it is given a pattern that contains a
1174 runaway infinite recursion. I updated the discussion in the pcrestack man
1175 page.
1176
1177 13. Now that we have gone to the x.xx style of version numbers, the minor
1178 version may start with zero. Using 08 or 09 is a bad idea because users
1179 might check the value of PCRE_MINOR in their code, and 08 or 09 may be
1180 interpreted as invalid octal numbers. I've updated the previous comment in
1181 configure.ac, and also added a check that gives an error if 08 or 09 are
1182 used.
1183
1184 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
1185 causing partial matching to fail when the end of the subject matched \W
1186 in a UTF-8 pattern where \W was quantified with a minimum of 3.
1187
1188 15. There were some discrepancies between the declarations in pcre_internal.h
1189 of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
1190 their definitions. The declarations used "const uschar *" and the
1191 definitions used USPTR. Even though USPTR is normally defined as "const
1192 unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
1193 reported that: "This difference in casting confuses some C++ compilers, for
1194 example, SunCC recognizes above declarations as different functions and
1195 generates broken code for hbpcre." I have changed the declarations to use
1196 USPTR.
1197
1198 16. GNU libtool is named differently on some systems. The autogen.sh script now
1199 tries several variants such as glibtoolize (MacOSX) and libtoolize1x
1200 (FreeBSD).
1201
1202 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
1203 (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
1204 comment: "Figure out how to create a longlong from a string: strtoll and
1205 equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
1206 instance, but it only takes 2 args instead of 3!"
1207
1208 18. A subtle bug concerned with back references has been fixed by a change of
1209 specification, with a corresponding code fix. A pattern such as
1210 ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
1211 refers, was giving matches when it shouldn't. For example, xa=xaaa would
1212 match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
1213 same bug. Such groups have to be quantified to be useful, or contained
1214 inside another quantified group. (If there's no repetition, the reference
1215 can never match.) The problem arises because, having left the group and
1216 moved on to the rest of the pattern, a later failure that backtracks into
1217 the group uses the captured value from the final iteration of the group
1218 rather than the correct earlier one. I have fixed this in PCRE by forcing
1219 any group that contains a reference to itself to be an atomic group; that
1220 is, there cannot be any backtracking into it once it has completed. This is
1221 similar to recursive and subroutine calls.
1222
1223
1224 Version 8.00 19-Oct-09
1225 ----------------------
1226
1227 1. The table for translating pcre_compile() error codes into POSIX error codes
1228 was out-of-date, and there was no check on the pcre_compile() error code
1229 being within the table. This could lead to an OK return being given in
1230 error.
1231
1232 2. Changed the call to open a subject file in pcregrep from fopen(pathname,
1233 "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
1234 in a Windows environment.
1235
1236 3. The pcregrep --count option prints the count for each file even when it is
1237 zero, as does GNU grep. However, pcregrep was also printing all files when
1238 --files-with-matches was added. Now, when both options are given, it prints
1239 counts only for those files that have at least one match. (GNU grep just
1240 prints the file name in this circumstance, but including the count seems
1241 more useful - otherwise, why use --count?) Also ensured that the
1242 combination -clh just lists non-zero counts, with no names.
1243
1244 4. The long form of the pcregrep -F option was incorrectly implemented as
1245 --fixed_strings instead of --fixed-strings. This is an incompatible change,
1246 but it seems right to fix it, and I didn't think it was worth preserving
1247 the old behaviour.
1248
1249 5. The command line items --regex=pattern and --regexp=pattern were not
1250 recognized by pcregrep, which required --regex pattern or --regexp pattern
1251 (with a space rather than an '='). The man page documented the '=' forms,
1252 which are compatible with GNU grep; these now work.
1253
1254 6. No libpcreposix.pc file was created for pkg-config; there was just
1255 libpcre.pc and libpcrecpp.pc. The omission has been rectified.
1256
1257 7. Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
1258 when UCP support is not needed, by modifying the Python script that
1259 generates it from Unicode data files. This should not matter if the module
1260 is correctly used as a library, but I received one complaint about 50K of
1261 unwanted data. My guess is that the person linked everything into his
1262 program rather than using a library. Anyway, it does no harm.
1263
1264 8. A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
1265 was a minimum greater than 1 for a wide character in a possessive
1266 repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
1267 which had an unlimited repeat of a nested, fixed maximum repeat of a wide
1268 character. Chaos in the form of incorrect output or a compiling loop could
1269 result.
1270
1271 9. The restrictions on what a pattern can contain when partial matching is
1272 requested for pcre_exec() have been removed. All patterns can now be
1273 partially matched by this function. In addition, if there are at least two
1274 slots in the offset vector, the offset of the earliest inspected character
1275 for the match and the offset of the end of the subject are set in them when
1276 PCRE_ERROR_PARTIAL is returned.
1277
1278 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
1279 synonymous with PCRE_PARTIAL, for backwards compatibility, and
1280 PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
1281 and may be more useful for multi-segment matching.
1282
1283 11. Partial matching with pcre_exec() is now more intuitive. A partial match
1284 used to be given if ever the end of the subject was reached; now it is
1285 given only if matching could not proceed because another character was
1286 needed. This makes a difference in some odd cases such as Z(*FAIL) with the
1287 string "Z", which now yields "no match" instead of "partial match". In the
1288 case of pcre_dfa_exec(), "no match" is given if every matching path for the
1289 final character ended with (*FAIL).
1290
1291 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
1292 if the pattern had a "must contain" character that was already found in the
1293 earlier partial match, unless partial matching was again requested. For
1294 example, with the pattern /dog.(body)?/, the "must contain" character is
1295 "g". If the first part-match was for the string "dog", restarting with
1296 "sbody" failed. This bug has been fixed.
1297
1298 13. The string returned by pcre_dfa_exec() after a partial match has been
1299 changed so that it starts at the first inspected character rather than the
1300 first character of the match. This makes a difference only if the pattern
1301 starts with a lookbehind assertion or \b or \B (\K is not supported by
1302 pcre_dfa_exec()). It's an incompatible change, but it makes the two
1303 matching functions compatible, and I think it's the right thing to do.
1304
1305 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
1306 so that the demonstration program is easily available in environments where
1307 PCRE has not been installed from source.
1308
1309 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
1310 libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
1311 library.
1312
1313 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
1314 It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
1315 is not the first non-POSIX option to be added. Clearly some people find
1316 these options useful.
1317
1318 17. If a caller to the POSIX matching function regexec() passes a non-zero
1319 value for nmatch with a NULL value for pmatch, the value of
1320 nmatch is forced to zero.
1321
1322 18. RunGrepTest did not have a test for the availability of the -u option of
1323 the diff command, as RunTest does. It now checks in the same way as
1324 RunTest, and also checks for the -b option.
1325
1326 19. If an odd number of negated classes containing just a single character
1327 interposed, within parentheses, between a forward reference to a named
1328 subpattern and the definition of the subpattern, compilation crashed with
1329 an internal error, complaining that it could not find the referenced
1330 subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
1331 [The bug was that it was starting one character too far in when skipping
1332 over the character class, thus treating the ] as data rather than
1333 terminating the class. This meant it could skip too much.]
1334
1335 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
1336 /g option in pcretest when the pattern contains \K, which makes it possible
1337 to have an empty string match not at the start, even when the pattern is
1338 anchored. Updated pcretest and pcredemo to use this option.
1339
1340 21. If the maximum number of capturing subpatterns in a recursion was greater
1341 than the maximum at the outer level, the higher number was returned, but
1342 with unset values at the outer level. The correct (outer level) value is
1343 now given.
1344
1345 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
1346 PCRE did not set those parentheses (unlike Perl). I have now found a way to
1347 make it do so. The string so far is captured, making this feature
1348 compatible with Perl.
1349
1350 23. The tests have been re-organized, adding tests 11 and 12, to make it
1351 possible to check the Perl 5.10 features against Perl 5.10.
1352
1353 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
1354 pattern matches a fixed length string. PCRE did not allow this; now it
1355 does. Neither allows recursion.
1356
1357 25. I finally figured out how to implement a request to provide the minimum
1358 length of subject string that was needed in order to match a given pattern.
1359 (It was back references and recursion that I had previously got hung up
1360 on.) This code has now been added to pcre_study(); it finds a lower bound
1361 to the length of subject needed. It is not necessarily the greatest lower
1362 bound, but using it to avoid searching strings that are too short does give
1363 some useful speed-ups. The value is available to calling programs via
1364 pcre_fullinfo().
1365
1366 26. While implementing 25, I discovered to my embarrassment that pcretest had
1367 not been passing the result of pcre_study() to pcre_dfa_exec(), so the
1368 study optimizations had never been tested with that matching function.
1369 Oops. What is worse, even when it was passed study data, there was a bug in
1370 pcre_dfa_exec() that meant it never actually used it. Double oops. There
1371 were also very few tests of studied patterns with pcre_dfa_exec().
1372
1373 27. If (?| is used to create subpatterns with duplicate numbers, they are now
1374 allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
1375 on the other side of the coin, they are no longer allowed to have different
1376 names, because these cannot be distinguished in PCRE, and this has caused
1377 confusion. (This is a difference from Perl.)
1378
1379 28. When duplicate subpattern names are present (necessarily with different
1380 numbers, as required by 27 above), and a test is made by name in a
1381 conditional pattern, either for a subpattern having been matched, or for
1382 recursion in such a pattern, all the associated numbered subpatterns are
1383 tested, and the overall condition is true if the condition is true for any
1384 one of them. This is the way Perl works, and is also more like the way
1385 testing by number works.
1386
1387
1388 Version 7.9 11-Apr-09
1389 ---------------------
1390
1391 1. When building with support for bzlib/zlib (pcregrep) and/or readline
1392 (pcretest), all targets were linked against these libraries. This included
1393 libpcre, libpcreposix, and libpcrecpp, even though they do not use these
1394 libraries. This caused unwanted dependencies to be created. This problem
1395 has been fixed, and now only pcregrep is linked with bzlib/zlib and only
1396 pcretest is linked with readline.
1397
1398 2. The "typedef int BOOL" in pcre_internal.h that was included inside the
1399 "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
1400 moved outside it again, because FALSE and TRUE are already defined in AIX,
1401 but BOOL is not.
1402
1403 3. The pcre_config() function was treating the PCRE_MATCH_LIMIT and
1404 PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
1405
1406 4. The pcregrep documentation said spaces were inserted as well as colons (or
1407 hyphens) following file names and line numbers when outputting matching
1408 lines. This is not true; no spaces are inserted. I have also clarified the
1409 wording for the --colour (or --color) option.
1410
1411 5. In pcregrep, when --colour was used with -o, the list of matching strings
1412 was not coloured; this is different to GNU grep, so I have changed it to be
1413 the same.
1414
1415 6. When --colo(u)r was used in pcregrep, only the first matching substring in
1416 each matching line was coloured. Now it goes on to look for further matches
1417 of any of the test patterns, which is the same behaviour as GNU grep.
1418
1419 7. A pattern that could match an empty string could cause pcregrep to loop; it
1420 doesn't make sense to accept an empty string match in pcregrep, so I have
1421 locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
1422 seems to be how GNU grep behaves.
1423
1424 8. The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
1425 start or after a newline", because the conditional assertion was not being
1426 correctly handled. The rule now is that both the assertion and what follows
1427 in the first alternative must satisfy the test.
1428
1429 9. If auto-callout was enabled in a pattern with a conditional group whose
1430 condition was an assertion, PCRE could crash during matching, both with
1431 pcre_exec() and pcre_dfa_exec().
1432
1433 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
1434 used for matching.
1435
1436 11. Unicode property support in character classes was not working for
1437 characters (bytes) greater than 127 when not in UTF-8 mode.
1438
1439 12. Added the -M command line option to pcretest.
1440
1441 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
1442
1443 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
1444
1445 16. Added comments and documentation about mis-use of no_arg in the C++
1446 wrapper.
1447
1448 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
1449 from Martin Jerabek that uses macro names for all relevant character and
1450 string constants.
1451
1452 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
1453 SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
1454 SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
1455 these, but not everybody uses configure.
1456
1457 19. A conditional group that had only one branch was not being correctly
1458 recognized as an item that could match an empty string. This meant that an
1459 enclosing group might also not be so recognized, causing infinite looping
1460 (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
1461 with the subject "ab", where knowledge that the repeated group can match
1462 nothing is needed in order to break the loop.
1463
1464 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
1465 exec(), but without supplying a callout function, matching went wrong.
1466
1467 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
1468 leak if the size of the offset vector was greater than 30. When the vector
1469 is smaller, the saved offsets during recursion go onto a local stack
1470 vector, but for larger vectors malloc() is used. It was failing to free
1471 when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
1472 error, in fact).
1473
1474 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
1475 heapframe that is used only when UTF-8 support is enabled. This caused no
1476 problem, but was untidy.
1477
1478 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
1479 CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
1480 included within another project.
1481
1482 24. Steven Van Ingelgem's patches to add more options to the CMake support,
1483 slightly modified by me:
1484
1485 (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
1486 not building pcregrep.
1487
1488 (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
1489 if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
1490
1491 25. Forward references, both numeric and by name, in patterns that made use of
1492 duplicate group numbers, could behave incorrectly or give incorrect errors,
1493 because when scanning forward to find the reference group, PCRE was not
1494 taking into account the duplicate group numbers. A pattern such as
1495 ^X(?3)(a)(?|(b)|(q))(Y) is an example.
1496
1497 26. Changed a few more instances of "const unsigned char *" to USPTR, making
1498 the feature of a custom pointer more persuasive (as requested by a user).
1499
1500 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
1501 pcretest.c, inside #ifndefs, because it seems they are sometimes already
1502 pre-defined.
1503
1504 28. Added support for (*UTF8) at the start of a pattern.
1505
1506 29. Arrange for flags added by the "release type" setting in CMake to be shown
1507 in the configuration summary.
1508
1509
1510 Version 7.8 05-Sep-08
1511 ---------------------
1512
1513 1. Replaced UCP searching code with optimized version as implemented for Ad
1514 Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
1515 stage table and inline lookup instead of a function, giving speed ups of 2
1516 to 5 times on some simple patterns that I tested. Permission was given to
1517 distribute the MultiStage2.py script that generates the tables (it's not in
1518 the tarball, but is in the Subversion repository).
1519
1520 2. Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
1521 scripts.
1522
1523 3. Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
1524 a group with a zero qualifier. The result of the study could be incorrect,
1525 or the function might crash, depending on the pattern.
1526
1527 4. Caseless matching was not working for non-ASCII characters in back
1528 references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
1529 It now works when Unicode Property Support is available.
1530
1531 5. In pcretest, an escape such as \x{de} in the data was always generating
1532 a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
1533 non-UTF-8 mode. If the value is greater than 255, it gives a warning about
1534 truncation.
1535
1536 6. Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
1537
1538 7. Added two (int) casts to pcregrep when printing the difference of two
1539 pointers, in case they are 64-bit values.
1540
1541 8. Added comments about Mac OS X stack usage to the pcrestack man page and to
1542 test 2 if it fails.
1543
1544 9. Added PCRE_CALL_CONVENTION just before the names of all exported functions,
1545 and a #define of that name to empty if it is not externally set. This is to
1546 allow users of MSVC to set it if necessary.
1547
1548 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
1549 the convenience functions in the pcre_get.c source file.
1550
1551 11. An option change at the start of a pattern that had top-level alternatives
1552 could cause overwriting and/or a crash. This command provoked a crash in
1553 some environments:
1554
1555 printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
1556
1557 This potential security problem was recorded as CVE-2008-2371.
1558
1559 12. For a pattern where the match had to start at the beginning or immediately
1560 after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
1561 pcre_dfa_exec() could read past the end of the passed subject if there was
1562 no match. To help with detecting such bugs (e.g. with valgrind), I modified
1563 pcretest so that it places the subject at the end of its malloc-ed buffer.
1564
1565 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
1566 exec() might read past the end of the data buffer in UTF-8 mode.
1567
1568 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
1569 the data contained the byte 0x85 as part of a UTF-8 character within its
1570 first line. This applied both to normal and DFA matching.
1571
1572 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
1573 /^[^d]*?$/8 failed to match "abc".
1574
1575 16. Added a missing copyright notice to pcrecpp_internal.h.
1576
1577 17. Make it more clear in the documentation that values returned from
1578 pcre_exec() in ovector are byte offsets, not character counts.
1579
1580 18. Tidied a few places to stop certain compilers from issuing warnings.
1581
1582 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
1583 supplied by Stefan Weber. I made a further small update for 7.8 because
1584 there is a change of source arrangements: the pcre_searchfuncs.c module is
1585 replaced by pcre_ucd.c.
1586
1587
1588 Version 7.7 07-May-08
1589 ---------------------
1590
1591 1. Applied Craig's patch to sort out a long long problem: "If we can't convert
1592 a string to a long long, pretend we don't even have a long long." This is
1593 done by checking for the strtoq, strtoll, and _strtoi64 functions.
1594
1595 2. Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
1596 pre-7.6 versions, which defined a global no_arg variable instead of putting
1597 it in the RE class. (See also #8 below.)
1598
1599 3. Remove a line of dead code, identified by coverity and reported by Nuno
1600 Lopes.
1601
1602 4. Fixed two related pcregrep bugs involving -r with --include or --exclude:
1603
1604 (1) The include/exclude patterns were being applied to the whole pathnames
1605 of files, instead of just to the final components.
1606
1607 (2) If there was more than one level of directory, the subdirectories were
1608 skipped unless they satisfied the include/exclude conditions. This is
1609 inconsistent with GNU grep (and could even be seen as contrary to the
1610 pcregrep specification - which I improved to make it absolutely clear).
1611 The action now is always to scan all levels of directory, and just
1612 apply the include/exclude patterns to regular files.
1613
1614 5. Added the --include_dir and --exclude_dir patterns to pcregrep, and used
1615 --exclude_dir in the tests to avoid scanning .svn directories.
1616
1617 6. Applied Craig's patch to the QuoteMeta function so that it escapes the
1618 NUL character as backslash + 0 rather than backslash + NUL, because PCRE
1619 doesn't support NULs in patterns.
1620
1621 7. Added some missing "const"s to declarations of static tables in
1622 pcre_compile.c and pcre_dfa_exec.c.
1623
1624 8. Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
1625 caused by fix #2 above. (Subsequently also a second patch to fix the
1626 first patch. And a third patch - this was a messy problem.)
1627
1628 9. Applied Craig's patch to remove the use of push_back().
1629
1630 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
1631 matching function regexec().
1632
1633 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
1634 which, however, unlike Perl's \g{...}, are subroutine calls, not back
1635 references. PCRE supports relative numbers with this syntax (I don't think
1636 Oniguruma does).
1637
1638 12. Previously, a group with a zero repeat such as (...){0} was completely
1639 omitted from the compiled regex. However, this means that if the group
1640 was called as a subroutine from elsewhere in the pattern, things went wrong
1641 (an internal error was given). Such groups are now left in the compiled
1642 pattern, with a new opcode that causes them to be skipped at execution
1643 time.
1644
1645 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
1646 to the way PCRE behaves:
1647
1648 (a) A lone ] character is dis-allowed (Perl treats it as data).
1649
1650 (b) A back reference to an unmatched subpattern matches an empty string
1651 (Perl fails the current match path).
1652
1653 (c) A data ] in a character class must be notated as \] because if the
1654 first data character in a class is ], it defines an empty class. (In
1655 Perl it is not possible to have an empty class.) The empty class []
1656 never matches; it forces failure and is equivalent to (*FAIL) or (?!).
1657 The negative empty class [^] matches any one character, independently
1658 of the DOTALL setting.
1659
1660 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
1661 non-existent subpattern following a character class starting with ']' and
1662 containing () gave an internal compiling error instead of "reference to
1663 non-existent subpattern". Fortunately, when the pattern did exist, the
1664 compiled code was correct. (When scanning forwards to check for the
1665 existencd of the subpattern, it was treating the data ']' as terminating
1666 the class, so got the count wrong. When actually compiling, the reference
1667 was subsequently set up correctly.)
1668
1669 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
1670 it was being rejected as not supported by pcre_dfa_exec(), even though
1671 other assertions are supported. I have made pcre_dfa_exec() support
1672 (*FAIL).
1673
1674 16. The implementation of 13c above involved the invention of a new opcode,
1675 OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
1676 cannot be changed at match time, I realized I could make a small
1677 improvement to matching performance by compiling OP_ALLANY instead of
1678 OP_ANY for "." when DOTALL was set, and then removing the runtime tests
1679 on the OP_ANY path.
1680
1681 17. Compiling pcretest on Windows with readline support failed without the
1682 following two fixes: (1) Make the unistd.h include conditional on
1683 HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
1684
1685 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
1686 ncurses library to be included for pcretest when ReadLine support is
1687 requested, but also to allow for it to be overridden. This patch came from
1688 Daniel Bergström.
1689
1690 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
1691 as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
1692 any errors with the current Unicode tables. Thanks to Peter Kankowski for
1693 spotting this.
1694
1695
1696 Version 7.6 28-Jan-08
1697 ---------------------
1698
1699 1. A character class containing a very large number of characters with
1700 codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
1701 overflow.
1702
1703 2. Patch to cut out the "long long" test in pcrecpp_unittest when
1704 HAVE_LONG_LONG is not defined.
1705
1706 3. Applied Christian Ehrlicher's patch to update the CMake build files to
1707 bring them up to date and include new features. This patch includes:
1708
1709 - Fixed PH's badly added libz and libbz2 support.
1710 - Fixed a problem with static linking.
1711 - Added pcredemo. [But later removed - see 7 below.]
1712 - Fixed dftables problem and added an option.
1713 - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
1714 HAVE_LONG_LONG.
1715 - Added readline support for pcretest.
1716 - Added an listing of the option settings after cmake has run.
1717
1718 4. A user submitted a patch to Makefile that makes it easy to create
1719 "pcre.dll" under mingw when using Configure/Make. I added stuff to
1720 Makefile.am that cause it to include this special target, without
1721 affecting anything else. Note that the same mingw target plus all
1722 the other distribution libraries and programs are now supported
1723 when configuring with CMake (see 6 below) instead of with
1724 Configure/Make.
1725
1726 5. Applied Craig's patch that moves no_arg into the RE class in the C++ code.
1727 This is an attempt to solve the reported problem "pcrecpp::no_arg is not
1728 exported in the Windows port". It has not yet been confirmed that the patch
1729 solves the problem, but it does no harm.
1730
1731 6. Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
1732 NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
1733 with CMake, and also correct the comment about stack recursion.
1734
1735 7. Remove the automatic building of pcredemo from the ./configure system and
1736 from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
1737 of a program that users should build themselves after PCRE is installed, so
1738 building it automatically is not really right. What is more, it gave
1739 trouble in some build environments.
1740
1741 8. Further tidies to CMakeLists.txt from Sheri and Christian.
1742
1743
1744 Version 7.5 10-Jan-08
1745 ---------------------
1746
1747 1. Applied a patch from Craig: "This patch makes it possible to 'ignore'
1748 values in parens when parsing an RE using the C++ wrapper."
1749
1750 2. Negative specials like \S did not work in character classes in UTF-8 mode.
1751 Characters greater than 255 were excluded from the class instead of being
1752 included.
1753
1754 3. The same bug as (2) above applied to negated POSIX classes such as
1755 [:^space:].
1756
1757 4. PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
1758 defined or documented. It seems to have been a typo for PCRE_STATIC, so
1759 I have changed it.
1760
1761 5. The construct (?&) was not diagnosed as a syntax error (it referenced the
1762 first named subpattern) and a construct such as (?&a) would reference the
1763 first named subpattern whose name started with "a" (in other words, the
1764 length check was missing). Both these problems are fixed. "Subpattern name
1765 expected" is now given for (?&) (a zero-length name), and this patch also
1766 makes it give the same error for \k'' (previously it complained that that
1767 was a reference to a non-existent subpattern).
1768
1769 6. The erroneous patterns (?+-a) and (?-+a) give different error messages;
1770 this is right because (?- can be followed by option settings as well as by
1771 digits. I have, however, made the messages clearer.
1772
1773 7. Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
1774 than the number used in the conditional) now cause a compile-time error.
1775 This is actually not compatible with Perl, which accepts such patterns, but
1776 treats the conditional as always being FALSE (as PCRE used to), but it
1777 seems to me that giving a diagnostic is better.
1778
1779 8. Change "alphameric" to the more common word "alphanumeric" in comments
1780 and messages.
1781
1782 9. Fix two occurrences of "backslash" in comments that should have been
1783 "backspace".
1784
1785 10. Remove two redundant lines of code that can never be obeyed (their function
1786 was moved elsewhere).
1787
1788 11. The program that makes PCRE's Unicode character property table had a bug
1789 which caused it to generate incorrect table entries for sequences of
1790 characters that have the same character type, but are in different scripts.
1791 It amalgamated them into a single range, with the script of the first of
1792 them. In other words, some characters were in the wrong script. There were
1793 thirteen such cases, affecting characters in the following ranges:
1794
1795 U+002b0 - U+002c1
1796 U+0060c - U+0060d
1797 U+0061e - U+00612
1798 U+0064b - U+0065e
1799 U+0074d - U+0076d
1800 U+01800 - U+01805
1801 U+01d00 - U+01d77
1802 U+01d9b - U+01dbf
1803 U+0200b - U+0200f
1804 U+030fc - U+030fe
1805 U+03260 - U+0327f
1806 U+0fb46 - U+0fbb1
1807 U+10450 - U+1049d
1808
1809 12. The -o option (show only the matching part of a line) for pcregrep was not
1810 compatible with GNU grep in that, if there was more than one match in a
1811 line, it showed only the first of them. It now behaves in the same way as
1812 GNU grep.
1813
1814 13. If the -o and -v options were combined for pcregrep, it printed a blank
1815 line for every non-matching line. GNU grep prints nothing, and pcregrep now
1816 does the same. The return code can be used to tell if there were any
1817 non-matching lines.
1818
1819 14. Added --file-offsets and --line-offsets to pcregrep.
1820
1821 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
1822 infinitely looping recursion. The bug was that positive lookaheads were not
1823 being skipped when checking for a possible empty match (negative lookaheads
1824 and both kinds of lookbehind were skipped).
1825
1826 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
1827 inclusion of <windows.h> to before rather than after the definition of
1828 INVALID_FILE_ATTRIBUTES (patch from David Byron).
1829
1830 17. Specifying a possessive quantifier with a specific limit for a Unicode
1831 character property caused pcre_compile() to compile bad code, which led at
1832 runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
1833 are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
1834 caused the error; without that there was no problem.
1835
1836 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
1837
1838 19. Added --enable-pcretest-libreadline.
1839
1840 20. In pcrecpp.cc, the variable 'count' was incremented twice in
1841 RE::GlobalReplace(). As a result, the number of replacements returned was
1842 double what it should be. I removed one of the increments, but Craig sent a
1843 later patch that removed the other one (the right fix) and added unit tests
1844 that check the return values (which was not done before).
1845
1846 21. Several CMake things:
1847
1848 (1) Arranged that, when cmake is used on Unix, the libraries end up with
1849 the names libpcre and libpcreposix, not just pcre and pcreposix.
1850
1851 (2) The above change means that pcretest and pcregrep are now correctly
1852 linked with the newly-built libraries, not previously installed ones.
1853
1854 (3) Added PCRE_SUPPORT_LIBREADLINE, PCRE_SUPPORT_LIBZ, PCRE_SUPPORT_LIBBZ2.
1855
1856 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
1857 crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
1858 UTF-8 newline character). The key issue is that the pattern starts .*;
1859 this means that the match must be either at the beginning, or after a
1860 newline. The bug was in the code for advancing after a failed match and
1861 checking that the new position followed a newline. It was not taking
1862 account of UTF-8 characters correctly.
1863
1864 23. PCRE was behaving differently from Perl in the way it recognized POSIX
1865 character classes. PCRE was not treating the sequence [:...:] as a
1866 character class unless the ... were all letters. Perl, however, seems to
1867 allow any characters between [: and :], though of course it rejects as
1868 unknown any "names" that contain non-letters, because all the known class
1869 names consist only of letters. Thus, Perl gives an error for [[:1234:]],
1870 for example, whereas PCRE did not - it did not recognize a POSIX character
1871 class. This seemed a bit dangerous, so the code has been changed to be
1872 closer to Perl. The behaviour is not identical to Perl, because PCRE will
1873 diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
1874 treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
1875 Perl does, and where it didn't before.
1876
1877 24. Rewrite so as to remove the single use of %n from pcregrep because in some
1878 Windows environments %n is disabled by default.
1879
1880
1881 Version 7.4 21-Sep-07
1882 ---------------------
1883
1884 1. Change 7.3/28 was implemented for classes by looking at the bitmap. This
1885 means that a class such as [\s] counted as "explicit reference to CR or
1886 LF". That isn't really right - the whole point of the change was to try to
1887 help when there was an actual mention of one of the two characters. So now
1888 the change happens only if \r or \n (or a literal CR or LF) character is
1889 encountered.
1890
1891 2. The 32-bit options word was also used for 6 internal flags, but the numbers
1892 of both had grown to the point where there were only 3 bits left.
1893 Fortunately, there was spare space in the data structure, and so I have
1894 moved the internal flags into a new 16-bit field to free up more option
1895 bits.
1896
1897 3. The appearance of (?J) at the start of a pattern set the DUPNAMES option,
1898 but did not set the internal JCHANGED flag - either of these is enough to
1899 control the way the "get" function works - but the PCRE_INFO_JCHANGED
1900 facility is supposed to tell if (?J) was ever used, so now (?J) at the
1901 start sets both bits.
1902
1903 4. Added options (at build time, compile time, exec time) to change \R from
1904 matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
1905
1906 5. doc/pcresyntax.html was missing from the distribution.
1907
1908 6. Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
1909 compatibility, even though it is no longer used.
1910
1911 7. Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
1912 strtoull to pcrecpp.cc to select the available functions in WIN32 when the
1913 windows.h file is present (where different names are used). [This was
1914 reversed later after testing - see 16 below.]
1915
1916 8. Changed all #include <config.h> to #include "config.h". There were also
1917 some further <pcre.h> cases that I changed to "pcre.h".
1918
1919 9. When pcregrep was used with the --colour option, it missed the line ending
1920 sequence off the lines that it output.
1921
1922 10. It was pointed out to me that arrays of string pointers cause lots of
1923 relocations when a shared library is dynamically loaded. A technique of
1924 using a single long string with a table of offsets can drastically reduce
1925 these. I have refactored PCRE in four places to do this. The result is
1926 dramatic:
1927
1928 Originally: 290
1929 After changing UCP table: 187
1930 After changing error message table: 43
1931 After changing table of "verbs" 36
1932 After changing table of Posix names 22
1933
1934 Thanks to the folks working on Gregex for glib for this insight.
1935
1936 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
1937 unicode-properties was also set.
1938
1939 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
1940
1941 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
1942 checked only for CRLF.
1943
1944 14. Added casts to pcretest.c to avoid compiler warnings.
1945
1946 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
1947
1948 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
1949 and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
1950 entirely. This removes changes made in 7 above.
1951
1952 17. The CMake files have been updated, and there is now more information about
1953 building with CMake in the NON-UNIX-USE document.
1954
1955
1956 Version 7.3 28-Aug-07
1957 ---------------------
1958
1959 1. In the rejigging of the build system that eventually resulted in 7.1, the
1960 line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
1961 brackets there is not right, since it causes compilers to look for an
1962 installed pcre.h, not the version that is in the source that is being
1963 compiled (which of course may be different). I have changed it back to:
1964
1965 #include "pcre.h"
1966
1967 I have a vague recollection that the change was concerned with compiling in
1968 different directories, but in the new build system, that is taken care of
1969 by the VPATH setting the Makefile.
1970
1971 2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
1972 when the subject happened to end in the byte 0x85 (e.g. if the last
1973 character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
1974 characters but of course it shouldn't be taken as a newline when it is part
1975 of another character. The bug was that, for an unlimited repeat of . in
1976 not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
1977 characters when looking for a newline.
1978
1979 3. A small performance improvement in the DOTALL UTF-8 mode .* case.
1980
1981 4. Debugging: adjusted the names of opcodes for different kinds of parentheses
1982 in debug output.
1983
1984 5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
1985 long printing in the pcrecpp unittest when running under MinGW.
1986
1987 6. ESC_K was left out of the EBCDIC table.
1988
1989 7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
1990 parentheses; I made it 1000, which seemed large enough. Unfortunately, the
1991 limit also applies to "virtual nesting" when a pattern is recursive, and in
1992 this case 1000 isn't so big. I have been able to remove this limit at the
1993 expense of backing off one optimization in certain circumstances. Normally,
1994 when pcre_exec() would call its internal match() function recursively and
1995 immediately return the result unconditionally, it uses a "tail recursion"
1996 feature to save stack. However, when a subpattern that can match an empty
1997 string has an unlimited repetition quantifier, it no longer makes this
1998 optimization. That gives it a stack frame in which to save the data for
1999 checking that an empty string has been matched. Previously this was taken
2000 from the 1000-entry workspace that had been reserved. So now there is no
2001 explicit limit, but more stack is used.
2002
2003 8. Applied Daniel's patches to solve problems with the import/export magic
2004 syntax that is required for Windows, and which was going wrong for the
2005 pcreposix and pcrecpp parts of the library. These were overlooked when this
2006 problem was solved for the main library.
2007
2008 9. There were some crude static tests to avoid integer overflow when computing
2009 the size of patterns that contain repeated groups with explicit upper
2010 limits. As the maximum quantifier is 65535, the maximum group length was
2011 set at 30,000 so that the product of these two numbers did not overflow a
2012 32-bit integer. However, it turns out that people want to use groups that
2013 are longer than 30,000 bytes (though not repeat them that many times).
2014 Change 7.0/17 (the refactoring of the way the pattern size is computed) has
2015 made it possible to implement the integer overflow checks in a much more
2016 dynamic way, which I have now done. The artificial limitation on group
2017 length has been removed - we now have only the limit on the total length of
2018 the compiled pattern, which depends on the LINK_SIZE setting.
2019
2020 10. Fixed a bug in the documentation for get/copy named substring when
2021 duplicate names are permitted. If none of the named substrings are set, the
2022 functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
2023 empty string.
2024
2025 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
2026 instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
2027 because the ] is interpreted as the first data character and the
2028 terminating ] is not found. PCRE has been made compatible with Perl in this
2029 regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
2030 cause memory overwriting.
2031
2032 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
2033 string has been matched (to stop an infinite loop). It was not recognizing
2034 a conditional subpattern that could match an empty string if that
2035 subpattern was within another subpattern. For example, it looped when
2036 trying to match (((?(1)X|))*) but it was OK with ((?(1)X|)*) where the
2037 condition was not nested. This bug has been fixed.
2038
2039 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
2040 past the start of the subject in the presence of bytes with the top bit
2041 set, for example "\x8aBCD".
2042
2043 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
2044 (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
2045
2046 14. Optimized (?!) to (*FAIL).
2047
2048 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
2049 This restricts code points to be within the range 0 to 0x10FFFF, excluding
2050 the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
2051 full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
2052 does: it's just the validity check that is more restrictive.
2053
2054 16. Inserted checks for integer overflows during escape sequence (backslash)
2055 processing, and also fixed erroneous offset values for syntax errors during
2056 backslash processing.
2057
2058 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
2059 for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
2060
2061 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
2062 caused an overrun.
2063
2064 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
2065 something other than just ASCII characters) inside a group that had an
2066 unlimited repeat caused a loop at compile time (while checking to see
2067 whether the group could match an empty string).
2068
2069 20. Debugging a pattern containing \p or \P could cause a crash. For example,
2070 [\P{Any}] did so. (Error in the code for printing property names.)
2071
2072 21. An orphan \E inside a character class could cause a crash.
2073
2074 22. A repeated capturing bracket such as (A)? could cause a wild memory
2075 reference during compilation.
2076
2077 23. There are several functions in pcre_compile() that scan along a compiled
2078 expression for various reasons (e.g. to see if it's fixed length for look
2079 behind). There were bugs in these functions when a repeated \p or \P was
2080 present in the pattern. These operators have additional parameters compared
2081 with \d, etc, and these were not being taken into account when moving along
2082 the compiled data. Specifically:
2083
2084 (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
2085 length.
2086
2087 (b) An item such as \pL+ within a repeated group could cause crashes or
2088 loops.
2089
2090 (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
2091 "reference to non-existent subpattern" error.
2092
2093 (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
2094
2095 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
2096 characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
2097
2098 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
2099
2100 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
2101 character were causing crashes (broken optimization).
2102
2103 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
2104 \p or \P) caused a compile-time loop.
2105
2106 28. More problems have arisen in unanchored patterns when CRLF is a valid line
2107 break. For example, the unstudied pattern [\r\n]A does not match the string
2108 "\r\nA" because change 7.0/46 below moves the current point on by two
2109 characters after failing to match at the start. However, the pattern \nA
2110 *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
2111 the same is true. There doesn't seem any very clean way out of this, but
2112 what I have chosen to do makes the common cases work: PCRE now takes note
2113 of whether there can be an explicit match for \r or \n anywhere in the
2114 pattern, and if so, 7.0/46 no longer applies. As part of this change,
2115 there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
2116 pattern has explicit CR or LF references.
2117
2118 29. Added (*CR) etc for changing newline setting at start of pattern.
2119
2120
2121 Version 7.2 19-Jun-07
2122 ---------------------
2123
2124 1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
2125 which is apparently normally available under Windows.
2126
2127 2. Re-jig the pcregrep tests with different newline settings in an attempt
2128 to make them independent of the local environment's newline setting.
2129
2130 3. Add code to configure.ac to remove -g from the CFLAGS default settings.
2131
2132 4. Some of the "internals" tests were previously cut out when the link size
2133 was not 2, because the output contained actual offsets. The recent new
2134 "Z" feature of pcretest means that these can be cut out, making the tests
2135 usable with all link sizes.
2136
2137 5. Implemented Stan Switzer's goto replacement for longjmp() when not using
2138 stack recursion. This gives a massive performance boost under BSD, but just
2139 a small improvement under Linux. However, it saves one field in the frame
2140 in all cases.
2141
2142 6. Added more features from the forthcoming Perl 5.10:
2143
2144 (a) (?-n) (where n is a string of digits) is a relative subroutine or
2145 recursion call. It refers to the nth most recently opened parentheses.
2146
2147 (b) (?+n) is also a relative subroutine call; it refers to the nth next
2148 to be opened parentheses.
2149
2150 (c) Conditions that refer to capturing parentheses can be specified
2151 relatively, for example, (?(-2)... or (?(+3)...
2152
2153 (d) \K resets the start of the current match so that everything before
2154 is not part of it.
2155
2156 (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
2157
2158 (f) \g{name} is another synonym - part of Perl 5.10's unification of
2159 reference syntax.
2160
2161 (g) (?| introduces a group in which the numbering of parentheses in each
2162 alternative starts with the same number.
2163
2164 (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
2165
2166 7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
2167 PCRE_INFO_JCHANGED.
2168
2169 8. A pattern such as (.*(.)?)* caused pcre_exec() to fail by either not
2170 terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
2171 for detecting groups that can match an empty string.
2172
2173 9. A pattern with a very large number of alternatives (more than several
2174 hundred) was running out of internal workspace during the pre-compile
2175 phase, where pcre_compile() figures out how much memory will be needed. A
2176 bit of new cunning has reduced the workspace needed for groups with
2177 alternatives. The 1000-alternative test pattern now uses 12 bytes of
2178 workspace instead of running out of the 4096 that are available.
2179
2180 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
2181
2182 11. Applied patch from Google to remove an optimization that didn't quite work.
2183 The report of the bug said:
2184
2185 pcrecpp::RE("a*").FullMatch("aaa") matches, while
2186 pcrecpp::RE("a*?").FullMatch("aaa") does not, and
2187 pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
2188
2189 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
2190 it matched the wrong number of bytes.
2191
2192
2193 Version 7.1 24-Apr-07
2194 ---------------------
2195
2196 1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
2197 that is more "standard", making use of automake and other Autotools. There
2198 is some re-arrangement of the files and adjustment of comments consequent
2199 on this.
2200
2201 2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
2202 for recursive directory scanning broke on some systems because the files
2203 are not scanned in any specific order and on different systems the order
2204 was different. A call to "sort" has been inserted into RunGrepTest for the
2205 approprate test as a short-term fix. In the longer term there may be an
2206 alternative.
2207
2208 3. I had an email from Eric Raymond about problems translating some of PCRE's
2209 man pages to HTML (despite the fact that I distribute HTML pages, some
2210 people do their own conversions for various reasons). The problems
2211 concerned the use of low-level troff macros .br and .in. I have therefore
2212 removed all such uses from the man pages (some were redundant, some could
2213 be replaced by .nf/.fi pairs). The 132html script that I use to generate
2214 HTML has been updated to handle .nf/.fi and to complain if it encounters
2215 .br or .in.
2216
2217 4. Updated comments in configure.ac that get placed in config.h.in and also
2218 arranged for config.h to be included in the distribution, with the name
2219 config.h.generic, for the benefit of those who have to compile without
2220 Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
2221
2222 5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
2223 Weber: (1) pcre_internal.h was missing some function renames; (2) updated
2224 makevp.bat for the current PCRE, using the additional files
2225 makevp_c.txt, makevp_l.txt, and pcregexp.pas.
2226
2227 6. A Windows user reported a minor discrepancy with test 2, which turned out
2228 to be caused by a trailing space on an input line that had got lost in his
2229 copy. The trailing space was an accident, so I've just removed it.
2230
2231 7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
2232 that is needed.
2233
2234 8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
2235 as "const" (a) because they are and (b) because it helps the PHP
2236 maintainers who have recently made a script to detect big data structures
2237 in the php code that should be moved to the .rodata section. I remembered
2238 to update Builducptable as well, so it won't revert if ucptable.h is ever
2239 re-created.
2240
2241 9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
2242 pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
2243 order to be able to cut out the UTF-8 tables in the latter when UTF-8
2244 support is not required. This saves 1.5-2K of code, which is important in
2245 some applications.
2246
2247 Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
2248 so as not to refer to the tables, even though these functions will never be
2249 called when UTF-8 support is disabled. Otherwise there are problems with a
2250 shared library.
2251
2252 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
2253
2254 (a) It was defining its arguments as char * instead of void *.
2255
2256 (b) It was assuming that all moves were upwards in memory; this was true
2257 a long time ago when I wrote it, but is no longer the case.
2258
2259 The emulated memove() is provided for those environments that have neither
2260 memmove() nor bcopy(). I didn't think anyone used it these days, but that
2261 is clearly not the case, as these two bugs were recently reported.
2262
2263 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
2264 and Detrail to create the HTML documentation, the .txt form of the man
2265 pages, and it removes trailing spaces from listed files. It also creates
2266 pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
2267 case, it wraps all the #defines with #ifndefs. This script should be run
2268 before "make dist".
2269
2270 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
2271 with Unicode property support.
2272
2273 (a) For a maximizing quantifier, if the two different cases of the
2274 character were of different lengths in their UTF-8 codings (there are
2275 some cases like this - I found 11), and the matching function had to
2276 back up over a mixture of the two cases, it incorrectly assumed they
2277 were both the same length.
2278
2279 (b) When PCRE was configured to use the heap rather than the stack for
2280 recursion during matching, it was not correctly preserving the data for
2281 the other case of a UTF-8 character when checking ahead for a match
2282 while processing a minimizing repeat. If the check also involved
2283 matching a wide character, but failed, corruption could cause an
2284 erroneous result when trying to check for a repeat of the original
2285 character.
2286
2287 13. Some tidying changes to the testing mechanism:
2288
2289 (a) The RunTest script now detects the internal link size and whether there
2290 is UTF-8 and UCP support by running ./pcretest -C instead of relying on
2291 values substituted by "configure". (The RunGrepTest script already did
2292 this for UTF-8.) The configure.ac script no longer substitutes the
2293 relevant variables.
2294
2295 (b) The debugging options /B and /D in pcretest show the compiled bytecode
2296 with length and offset values. This means that the output is different
2297 for different internal link sizes. Test 2 is skipped for link sizes
2298 other than 2 because of this, bypassing the problem. Unfortunately,
2299 there was also a test in test 3 (the locale tests) that used /B and
2300 failed for link sizes other than 2. Rather than cut the whole test out,
2301 I have added a new /Z option to pcretest that replaces the length and
2302 offset values with spaces. This is now used to make test 3 independent
2303 of link size. (Test 2 will be tidied up later.)
2304
2305 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
2306 segmentation fault instead of returning the appropriate error message.
2307
2308 15. In multiline mode when the newline sequence was set to "any", the pattern
2309 ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
2310 This doesn't seem right; it now treats the CRLF combination as the line
2311 ending, and so does not match in that case. It's only a pattern such as ^$
2312 that would hit this one: something like ^ABC$ would have failed after \r
2313 and then tried again after \r\n.
2314
2315 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
2316 in an attempt to make files that differ only in their line terminators
2317 compare equal. This works on Linux.
2318
2319 17. Under certain error circumstances pcregrep might try to free random memory
2320 as it exited. This is now fixed, thanks to valgrind.
2321
2322 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
2323 "abc\r\n\r\n", it found an unwanted second match after the second \r. This
2324 was because its rules for how to advance for /g after matching an empty
2325 string at the end of a line did not allow for this case. They now check for
2326 it specially.
2327
2328 20. pcretest is supposed to handle patterns and data of any length, by
2329 extending its buffers when necessary. It was getting this wrong when the
2330 buffer for a data line had to be extended.
2331
2332 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
2333 CRLF as a newline sequence.
2334
2335 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
2336 out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
2337 I have nevertheless tidied it up.
2338
2339 23. Added some casts to kill warnings from HP-UX ia64 compiler.
2340
2341 24. Added a man page for pcre-config.
2342
2343
2344 Version 7.0 19-Dec-06
2345 ---------------------
2346
2347 1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
2348 moving to gcc 4.1.1.
2349
2350 2. The -S option for pcretest uses setrlimit(); I had omitted to #include
2351 sys/time.h, which is documented as needed for this function. It doesn't
2352 seem to matter on Linux, but it showed up on some releases of OS X.
2353
2354 3. It seems that there are systems where bytes whose values are greater than
2355 127 match isprint() in the "C" locale. The "C" locale should be the
2356 default when a C program starts up. In most systems, only ASCII printing
2357 characters match isprint(). This difference caused the output from pcretest
2358 to vary, making some of the tests fail. I have changed pcretest so that:
2359
2360 (a) When it is outputting text in the compiled version of a pattern, bytes
2361 other than 32-126 are always shown as hex escapes.
2362
2363 (b) When it is outputting text that is a matched part of a subject string,
2364 it does the same, unless a different locale has been set for the match
2365 (using the /L modifier). In this case, it uses isprint() to decide.
2366
2367 4. Fixed a major bug that caused incorrect computation of the amount of memory
2368 required for a compiled pattern when options that changed within the
2369 pattern affected the logic of the preliminary scan that determines the
2370 length. The relevant options are -x, and -i in UTF-8 mode. The result was
2371 that the computed length was too small. The symptoms of this bug were
2372 either the PCRE error "internal error: code overflow" from pcre_compile(),
2373 or a glibc crash with a message such as "pcretest: free(): invalid next
2374 size (fast)". Examples of patterns that provoked this bug (shown in
2375 pcretest format) are:
2376
2377 /(?-x: )/x
2378 /(?x)(?-x: \s*#\s*)/
2379 /((?i)[\x{c0}])/8
2380 /(?i:[\x{c0}])/8
2381
2382 HOWEVER: Change 17 below makes this fix obsolete as the memory computation
2383 is now done differently.
2384
2385 5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
2386 wrapper classes; (b) implement a new function in the C++ scanner that is
2387 more efficient than the old way of doing things because it avoids levels of
2388 recursion in the regex matching; (c) add a paragraph to the documentation
2389 for the FullMatch() function.
2390
2391 6. The escape sequence \n was being treated as whatever was defined as
2392 "newline". Not only was this contrary to the documentation, which states
2393 that \n is character 10 (hex 0A), but it also went horribly wrong when
2394 "newline" was defined as CRLF. This has been fixed.
2395
2396 7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
2397 was being set to -1 for the "end of line" case (supposedly a value that no
2398 character can have). Though this value is never used (the check for end of
2399 line is "zero bytes in current character"), it caused compiler complaints.
2400 I've changed it to 0xffffffff.
2401
2402 8. In pcre_version.c, the version string was being built by a sequence of
2403 C macros that, in the event of PCRE_PRERELEASE being defined as an empty
2404 string (as it is for production releases) called a macro with an empty
2405 argument. The C standard says the result of this is undefined. The gcc
2406 compiler treats it as an empty string (which was what was wanted) but it is
2407 reported that Visual C gives an error. The source has been hacked around to
2408 avoid this problem.
2409
2410 9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
2411 builds of pcretest, and changed the call to _setmode() to use _O_BINARY
2412 instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
2413 of them did).
2414
2415 10. Originally, pcretest opened its input and output without "b"; then I was
2416 told that "b" was needed in some environments, so it was added for release
2417 5.0 to both the input and output. (It makes no difference on Unix-like
2418 systems.) Later I was told that it is wrong for the input on Windows. I've
2419 now abstracted the modes into two macros, to make it easier to fiddle with
2420 them, and removed "b" from the input mode under Windows.
2421
2422 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
2423
2424 12. Added -help and --help to pcretest as an official way of being reminded
2425 of the options.
2426
2427 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
2428 and pcrecpp.cc because they annoy compilers at high warning levels.
2429
2430 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
2431
2432 15. Fixed an occurrence of == in configure.ac that should have been = (shell
2433 scripts are not C programs :-) and which was not noticed because it works
2434 on Linux.
2435
2436 16. pcretest is supposed to handle any length of pattern and data line (as one
2437 line or as a continued sequence of lines) by extending its input buffer if
2438 necessary. This feature was broken for very long pattern lines, leading to
2439 a string of junk being passed to pcre_compile() if the pattern was longer
2440 than about 50K.
2441
2442 17. I have done a major re-factoring of the way pcre_compile() computes the
2443 amount of memory needed for a compiled pattern. Previously, there was code
2444 that made a preliminary scan of the pattern in order to do this. That was
2445 OK when PCRE was new, but as the facilities have expanded, it has become
2446 harder and harder to keep it in step with the real compile phase, and there
2447 have been a number of bugs (see for example, 4 above). I have now found a
2448 cunning way of running the real compile function in a "fake" mode that
2449 enables it to compute how much memory it would need, while actually only
2450 ever using a few hundred bytes of working memory and without too many
2451 tests of the mode. This should make future maintenance and development
2452 easier. A side effect of this work is that the limit of 200 on the nesting
2453 depth of parentheses has been removed (though this was never a serious
2454 limitation, I suspect). However, there is a downside: pcre_compile() now
2455 runs more slowly than before (30% or more, depending on the pattern). I
2456 hope this isn't a big issue. There is no effect on runtime performance.
2457
2458 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
2459 newline (only possible for the last line of a file) and it was a
2460 pattern that set a locale (followed by /Lsomething), pcretest crashed.
2461
2462 19. Added additional timing features to pcretest. (1) The -tm option now times
2463 matching only, not compiling. (2) Both -t and -tm can be followed, as a
2464 separate command line item, by a number that specifies the number of
2465 repeats to use when timing. The default is 50000; this gives better
2466 precision, but takes uncomfortably long for very large patterns.
2467
2468 20. Extended pcre_study() to be more clever in cases where a branch of a
2469 subpattern has no definite first character. For example, (a*|b*)[cd] would
2470 previously give no result from pcre_study(). Now it recognizes that the
2471 first character must be a, b, c, or d.
2472
2473 21. There was an incorrect error "recursive call could loop indefinitely" if
2474 a subpattern (or the entire pattern) that was being tested for matching an
2475 empty string contained only one non-empty item after a nested subpattern.
2476 For example, the pattern (?>\x{100}*)\d(?R) provoked this error
2477 incorrectly, because the \d was being skipped in the check.
2478
2479 22. The pcretest program now has a new pattern option /B and a command line
2480 option -b, which is equivalent to adding /B to every pattern. This causes
2481 it to show the compiled bytecode, without the additional information that
2482 -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
2483 is the same as /B/I).
2484
2485 23. A new optimization is now able automatically to treat some sequences such
2486 as a*b as a*+b. More specifically, if something simple (such as a character
2487 or a simple class like \d) has an unlimited quantifier, and is followed by
2488 something that cannot possibly match the quantified thing, the quantifier
2489 is automatically "possessified".
2490
2491 24. A recursive reference to a subpattern whose number was greater than 39
2492 went wrong under certain circumstances in UTF-8 mode. This bug could also
2493 have affected the operation of pcre_study().
2494
2495 25. Realized that a little bit of performance could be had by replacing
2496 (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
2497
2498 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
2499
2500 27. Possessive quantifiers such as a++ were previously implemented by turning
2501 them into atomic groups such as ($>a+). Now they have their own opcodes,
2502 which improves performance. This includes the automatically created ones
2503 from 23 above.
2504
2505 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
2506 lookahead was broken if it was not anchored. PCRE was mistakenly expecting
2507 the first matched character to be a colon. This applied both to named and
2508 numbered groups.
2509
2510 29. The ucpinternal.h header file was missing its idempotency #ifdef.
2511
2512 30. I was sent a "project" file called libpcre.a.dev which I understand makes
2513 building PCRE on Windows easier, so I have included it in the distribution.
2514
2515 31. There is now a check in pcretest against a ridiculously large number being
2516 returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
2517 loop, the loop is abandoned.
2518
2519 32. Forward references to subpatterns in conditions such as (?(2)...) where
2520 subpattern 2 is defined later cause pcre_compile() to search forwards in
2521 the pattern for the relevant set of parentheses. This search went wrong
2522 when there were unescaped parentheses in a character class, parentheses
2523 escaped with \Q...\E, or parentheses in a #-comment in /x mode.
2524
2525 33. "Subroutine" calls and backreferences were previously restricted to
2526 referencing subpatterns earlier in the regex. This restriction has now
2527 been removed.
2528
2529 34. Added a number of extra features that are going to be in Perl 5.10. On the
2530 whole, these are just syntactic alternatives for features that PCRE had
2531 previously implemented using the Python syntax or my own invention. The
2532 other formats are all retained for compatibility.
2533
2534 (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
2535 as (?P<name>...). The new forms, as well as being in Perl 5.10, are
2536 also .NET compatible.
2537
2538 (b) A recursion or subroutine call to a named group can now be defined as
2539 (?&name) as well as (?P>name).
2540
2541 (c) A backreference to a named group can now be defined as \k<name> or
2542 \k'name' as well as (?P=name). The new forms, as well as being in Perl
2543 5.10, are also .NET compatible.
2544
2545 (d) A conditional reference to a named group can now use the syntax
2546 (?(<name>) or (?('name') as well as (?(name).
2547
2548 (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
2549 groups (named and numbered) that are never evaluated inline, but can be
2550 called as "subroutines" from elsewhere. In effect, the DEFINE condition
2551 is always false. There may be only one alternative in such a group.
2552
2553 (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
2554 as the simple (?(R). The condition is true only if the most recent
2555 recursion is that of the given number or name. It does not search out
2556 through the entire recursion stack.
2557
2558 (g) The escape \gN or \g{N} has been added, where N is a positive or
2559 negative number, specifying an absolute or relative reference.
2560
2561 35. Tidied to get rid of some further signed/unsigned compiler warnings and
2562 some "unreachable code" warnings.
2563
2564 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
2565 things, this adds five new scripts.
2566
2567 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
2568 There were also incompatibilities regarding the handling of \Q..\E inside
2569 character classes, for example with patterns like [\Qa\E-\Qz\E] where the
2570 hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
2571
2572 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
2573 matches an empty string, and forcibly breaks the loop. There were bugs in
2574 this code in non-simple cases. For a pattern such as ^(a()*)* matched
2575 against aaaa the result was just "a" rather than "aaaa", for example. Two
2576 separate and independent bugs (that affected different cases) have been
2577 fixed.
2578
2579 39. Refactored the code to abolish the use of different opcodes for small
2580 capturing bracket numbers. This is a tidy that I avoided doing when I
2581 removed the limit on the number of capturing brackets for 3.5 back in 2001.
2582 The new approach is not only tidier, it makes it possible to reduce the
2583 memory needed to fix the previous bug (38).
2584
2585 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
2586 sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
2587 processing dot, circumflex, or dollar metacharacters, or #-comments in /x
2588 mode.
2589
2590 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
2591 report.
2592
2593 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
2594 copy construction and assignment in the C++ wrapper.
2595
2596 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
2597 couple of bugs that could have given wrong results in the "--newline=crlf"
2598 case.
2599
2600 44. Added a number of casts and did some reorganization of signed/unsigned int
2601 variables following suggestions from Dair Grant. Also renamed the variable
2602 "this" as "item" because it is a C++ keyword.
2603
2604 45. Arranged for dftables to add
2605
2606 #include "pcre_internal.h"
2607
2608 to pcre_chartables.c because without it, gcc 4.x may remove the array
2609 definition from the final binary if PCRE is built into a static library and
2610 dead code stripping is activated.
2611
2612 46. For an unanchored pattern, if a match attempt fails at the start of a
2613 newline sequence, and the newline setting is CRLF or ANY, and the next two
2614 characters are CRLF, advance by two characters instead of one.
2615
2616
2617 Version 6.7 04-Jul-06
2618 ---------------------
2619
2620 1. In order to handle tests when input lines are enormously long, pcretest has
2621 been re-factored so that it automatically extends its buffers when
2622 necessary. The code is crude, but this _is_ just a test program. The
2623 default size has been increased from 32K to 50K.
2624
2625 2. The code in pcre_study() was using the value of the re argument before
2626 testing it for NULL. (Of course, in any sensible call of the function, it
2627 won't be NULL.)
2628
2629 3. The memmove() emulation function in pcre_internal.h, which is used on
2630 systems that lack both memmove() and bcopy() - that is, hardly ever -
2631 was missing a "static" storage class specifier.
2632
2633 4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
2634 containing an extended class (one that cannot be represented by a bitmap
2635 because it contains high-valued characters or Unicode property items, e.g.
2636 [\pZ]). Almost always one would set UTF-8 mode when processing such a
2637 pattern, but PCRE should not loop if you do not (it no longer does).
2638 [Detail: two cases were found: (a) a repeated subpattern containing an
2639 extended class; (b) a recursive reference to a subpattern that followed a
2640 previous extended class. It wasn't skipping over the extended class
2641 correctly when UTF-8 mode was not set.]
2642
2643 5. A negated single-character class was not being recognized as fixed-length
2644 in lookbehind assertions such as (?<=[^f]), leading to an incorrect
2645 compile error "lookbehind assertion is not fixed length".
2646
2647 6. The RunPerlTest auxiliary script was showing an unexpected difference
2648 between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
2649 write a Perl script that can interpret lines of an input file either as
2650 byte characters or as UTF-8, which is what "perltest" was being required to
2651 do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
2652 can't do is switch easily at run time between having the "use utf8;" pragma
2653 or not. In the end, I fudged it by using the RunPerlTest script to insert
2654 "use utf8;" explicitly for the UTF-8 tests.
2655
2656 7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
2657 the end of the subject string, contrary to the documentation and to what
2658 Perl does. This was true of both matching functions. Now it matches only at
2659 the start of the subject and immediately after *internal* newlines.
2660
2661 8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
2662 a pointer to an int instead of a pointer to an unsigned long int. This
2663 caused problems on 64-bit systems.
2664
2665 9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
2666 instance of the 'standard' template library not being so standard".
2667
2668 10. There was no check on the number of named subpatterns nor the maximum
2669 length of a subpattern name. The product of these values is used to compute
2670 the size of the memory block for a compiled pattern. By supplying a very
2671 long subpattern name and a large number of named subpatterns, the size
2672 computation could be caused to overflow. This is now prevented by limiting
2673 the length of names to 32 characters, and the number of named subpatterns
2674 to 10,000.
2675
2676 11. Subpatterns that are repeated with specific counts have to be replicated in
2677 the compiled pattern. The size of memory for this was computed from the
2678 length of the subpattern and the repeat count. The latter is limited to
2679 65535, but there was no limit on the former, meaning that integer overflow
2680 could in principle occur. The compiled length of a repeated subpattern is
2681 now limited to 30,000 bytes in order to prevent this.
2682
2683 12. Added the optional facility to have named substrings with the same name.
2684
2685 13. Added the ability to use a named substring as a condition, using the
2686 Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
2687 are numbers (not recommended). Forward references are permitted.
2688
2689 14. Added forward references in named backreferences (if you see what I mean).
2690
2691 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
2692 pattern could run off the end of the subject. For example, the pattern
2693 "(?s)(.{1,5})"8 did this with the subject "ab".
2694
2695 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
2696 PCRE_CASELESS was set when matching characters that were quantified with ?
2697 or *.
2698
2699 17. A character class other than a single negated character that had a minimum
2700 but no maximum quantifier - for example [ab]{6,} - was not handled
2701 correctly by pce_dfa_exec(). It would match only one character.
2702
2703 18. A valid (though odd) pattern that looked like a POSIX character
2704 class but used an invalid character after [ (for example [[,abc,]]) caused
2705 pcre_compile() to give the error "Failed: internal error: code overflow" or
2706 in some cases to crash with a glibc free() error. This could even happen if
2707 the pattern terminated after [[ but there just happened to be a sequence of
2708 letters, a binary zero, and a closing ] in the memory that followed.
2709
2710 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
2711 over the years. Originally (before any Unicode support), just the bottom 8
2712 bits were taken. Thus, for example, \500 really meant \100. Nowadays the
2713 output from "man perlunicode" includes this:
2714
2715 The regular expression compiler produces polymorphic opcodes. That
2716 is, the pattern adapts to the data and automatically switches to
2717 the Unicode character scheme when presented with Unicode data--or
2718 instead uses a traditional byte scheme when presented with byte
2719 data.
2720
2721 Sadly, a wide octal escape does not cause a switch, and in a string with
2722 no other multibyte characters, these octal escapes are treated as before.
2723 Thus, in Perl, the pattern /\500/ actually matches \100 but the pattern
2724 /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
2725 Unicode string.
2726
2727 I have not perpetrated such confusion in PCRE. Up till now, it took just
2728 the bottom 8 bits, as in old Perl. I have now made octal escapes with
2729 values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
2730 translate to the appropriate multibyte character.
2731
2732 29. Applied some refactoring to reduce the number of warnings from Microsoft
2733 and Borland compilers. This has included removing the fudge introduced
2734 seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
2735 a warning about an unused variable.
2736
2737 21. PCRE has not included VT (character 0x0b) in the set of whitespace
2738 characters since release 4.0, because Perl (from release 5.004) does not.
2739 [Or at least, is documented not to: some releases seem to be in conflict
2740 with the documentation.] However, when a pattern was studied with
2741 pcre_study() and all its branches started with \s, PCRE still included VT
2742 as a possible starting character. Of course, this did no harm; it just
2743 caused an unnecessary match attempt.
2744
2745 22. Removed a now-redundant internal flag bit that recorded the fact that case
2746 dependency changed within the pattern. This was once needed for "required
2747 byte" processing, but is no longer used. This recovers a now-scarce options
2748 bit. Also moved the least significant internal flag bit to the most-
2749 significant bit of the word, which was not previously used (hangover from
2750 the days when it was an int rather than a uint) to free up another bit for
2751 the future.
2752
2753 23. Added support for CRLF line endings as well as CR and LF. As well as the
2754 default being selectable at build time, it can now be changed at runtime
2755 via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
2756 specify that it is scanning data with non-default line endings.
2757
2758 24. Changed the definition of CXXLINK to make it agree with the definition of
2759 LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
2760
2761 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
2762 recursions. This makes a big different to stack usage for some patterns.
2763
2764 26. If a subpattern containing a named recursion or subroutine reference such
2765 as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
2766 the space required for the compiled pattern went wrong and gave too small a
2767 value. Depending on the environment, this could lead to "Failed: internal
2768 error: code overflow at offset 49" or "glibc detected double free or
2769 corruption" errors.
2770
2771 27. Applied patches from Google (a) to support the new newline modes and (b) to
2772 advance over multibyte UTF-8 characters in GlobalReplace.
2773
2774 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
2775 difference for some implementation of PCRE in some Windows version.
2776
2777 29. Added some extra testing facilities to pcretest:
2778
2779 \q<number> in a data line sets the "match limit" value
2780 \Q<number> in a data line sets the "match recursion limt" value
2781 -S <number> sets the stack size, where <number> is in megabytes
2782
2783 The -S option isn't available for Windows.
2784
2785
2786 Version 6.6 06-Feb-06
2787 ---------------------
2788
2789 1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
2790 in pcreposix.h. I have copied the definition from pcre.h.
2791
2792 2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
2793 because pcre.h is no longer a built file.
2794
2795 3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
2796 not normally included in the compiled code.
2797
2798
2799 Version 6.5 01-Feb-06
2800 ---------------------
2801
2802 1. When using the partial match feature with pcre_dfa_exec(), it was not
2803 anchoring the second and subsequent partial matches at the new starting
2804 point. This could lead to incorrect results. For example, with the pattern
2805 /1234/, partially matching against "123" and then "a4" gave a match.
2806
2807 2. Changes to pcregrep:
2808
2809 (a) All non-match returns from pcre_exec() were being treated as failures
2810 to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
2811 error message is output. Some extra information is given for the
2812 PCRE_ERROR_MATCHLIMIT and PCRE_ERROR_RECURSIONLIMIT errors, which are
2813 probably the only errors that are likely to be caused by users (by
2814 specifying a regex that has nested indefinite repeats, for instance).
2815 If there are more than 20 of these errors, pcregrep is abandoned.
2816
2817 (b) A binary zero was treated as data while matching, but terminated the
2818 output line if it was written out. This has been fixed: binary zeroes
2819 are now no different to any other data bytes.
2820
2821 (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
2822 used to set a locale for matching. The --locale=xxxx long option has
2823 been added (no short equivalent) to specify a locale explicitly on the
2824 pcregrep command, overriding the environment variables.
2825
2826 (d) When -B was used with -n, some line numbers in the output were one less
2827 than they should have been.
2828
2829 (e) Added the -o (--only-matching) option.
2830
2831 (f) If -A or -C was used with -c (count only), some lines of context were
2832 accidentally printed for the final match.
2833
2834 (g) Added the -H (--with-filename) option.
2835
2836 (h) The combination of options -rh failed to suppress file names for files
2837 that were found from directory arguments.
2838
2839 (i) Added the -D (--devices) and -d (--directories) options.
2840
2841 (j) Added the -F (--fixed-strings) option.
2842
2843 (k) Allow "-" to be used as a file name for -f as well as for a data file.
2844
2845 (l) Added the --colo(u)r option.
2846
2847 (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
2848 is not present by default.
2849
2850 3. A nasty bug was discovered in the handling of recursive patterns, that is,
2851 items such as (?R) or (?1), when the recursion could match a number of
2852 alternatives. If it matched one of the alternatives, but subsequently,
2853 outside the recursion, there was a failure, the code tried to back up into
2854 the recursion. However, because of the way PCRE is implemented, this is not
2855 possible, and the result was an incorrect result from the match.
2856
2857 In order to prevent this happening, the specification of recursion has
2858 been changed so that all such subpatterns are automatically treated as
2859 atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
2860
2861 4. I had overlooked the fact that, in some locales, there are characters for
2862 which isalpha() is true but neither isupper() nor islower() are true. In
2863 the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
2864 and ordfeminine) are like this. This affected the treatment of \w and \W
2865 when they appeared in character classes, but not when they appeared outside
2866 a character class. The bit map for "word" characters is now created
2867 separately from the results of isalnum() instead of just taking it from the
2868 upper, lower, and digit maps. (Plus the underscore character, of course.)
2869
2870 5. The above bug also affected the handling of POSIX character classes such as
2871 [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
2872 permanent tables. Instead, the bit maps for such a class were previously
2873 created as the appropriate unions of the upper, lower, and digit bitmaps.
2874 Now they are created by subtraction from the [[:word:]] class, which has
2875 its own bitmap.
2876
2877 6. The [[:blank:]] character class matches horizontal, but not vertical space.
2878 It is created by subtracting the vertical space characters (\x09, \x0a,
2879 \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
2880 subtraction was done in the overall bitmap for a character class, meaning
2881 that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
2882 be recognized. This bug has been fixed.
2883
2884 7. Patches from the folks at Google:
2885
2886 (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
2887 real life, but is still worth protecting against".
2888
2889 (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
2890 regular expressions".
2891
2892 (c) pcre_scanner.cc: avoid use of std::count() because not all systems
2893 have it.
2894
2895 (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
2896 "configure" and the latter not, in order to fix a problem somebody had
2897 with compiling the Arg class on HP-UX.
2898
2899 (e) Improve the error-handling of the C++ wrapper a little bit.
2900
2901 (f) New tests for checking recursion limiting.
2902
2903 8. The pcre_memmove() function, which is used only if the environment does not
2904 have a standard memmove() function (and is therefore rarely compiled),
2905 contained two bugs: (a) use of int instead of size_t, and (b) it was not
2906 returning a result (though PCRE never actually uses the result).
2907
2908 9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
2909 large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
2910 returned instead of calling malloc() with an overflowing number that would
2911 most likely cause subsequent chaos.
2912
2913 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
2914
2915 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
2916 with this option is matched, the nmatch and pmatch options of regexec() are
2917 ignored.
2918
2919 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
2920 provided in case anyone wants to the the POSIX interface with UTF-8
2921 strings.
2922
2923 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
2924 C++ linking (needed for some HP-UX environments).
2925
2926 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
2927 (unused parameter) and in the pcre_printint() function (omitted "default"
2928 switch label when the default is to do nothing).
2929
2930 15. Added some code to make it possible, when PCRE is compiled as a C++
2931 library, to replace subject pointers for pcre_exec() with a smart pointer
2932 class, thus making it possible to process discontinuous strings.
2933
2934 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
2935 much the same function. They were added by different people who were trying
2936 to make PCRE easy to compile on non-Unix systems. It has been suggested
2937 that PCRE_EXPORT be abolished now that there is more automatic apparatus
2938 for compiling on Windows systems. I have therefore replaced it with
2939 PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
2940 defaults to "extern" for C or "extern C" for C++, which works fine on
2941 Unix-like systems. It is now possible to override the value of PCRE_DATA_
2942 SCOPE with something explicit in config.h. In addition:
2943
2944 (a) pcreposix.h still had just "extern" instead of either of these macros;
2945 I have replaced it with PCRE_DATA_SCOPE.
2946
2947 (b) Functions such as _pcre_xclass(), which are internal to the library,
2948 but external in the C sense, all had PCRE_EXPORT in their definitions.
2949 This is apparently wrong for the Windows case, so I have removed it.
2950 (It makes no difference on Unix-like systems.)
2951
2952 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
2953 of recursive calls to match(). This is different to MATCH_LIMIT because
2954 that limits the total number of calls to match(), not all of which increase
2955 the depth of recursion. Limiting the recursion depth limits the amount of
2956 stack (or heap if NO_RECURSE is set) that is used. The default can be set
2957 when PCRE is compiled, and changed at run time. A patch from Google adds
2958 this functionality to the C++ interface.
2959
2960 18. Changes to the handling of Unicode character properties:
2961
2962 (a) Updated the table to Unicode 4.1.0.
2963
2964 (b) Recognize characters that are not in the table as "Cn" (undefined).
2965
2966 (c) I revised the way the table is implemented to a much improved format
2967 which includes recognition of ranges. It now supports the ranges that
2968 are defined in UnicodeData.txt, and it also amalgamates other
2969 characters into ranges. This has reduced the number of entries in the
2970 table from around 16,000 to around 3,000, thus reducing its size
2971 considerably. I realized I did not need to use a tree structure after
2972 all - a binary chop search is just as efficient. Having reduced the
2973 number of entries, I extended their size from 6 bytes to 8 bytes to
2974 allow for more data.
2975
2976 (d) Added support for Unicode script names via properties such as \p{Han}.
2977
2978 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
2979 matching that character.
2980
2981 20. When matching a repeated Unicode property with a minimum greater than zero,
2982 (for example \pL{2,}), PCRE could look past the end of the subject if it
2983 reached it while seeking the minimum number of characters. This could
2984 happen only if some of the characters were more than one byte long, because
2985 there is a check for at least the minimum number of bytes.
2986
2987 21. Refactored the implementation of \p and \P so as to be more general, to
2988 allow for more different types of property in future. This has changed the
2989 compiled form incompatibly. Anybody with saved compiled patterns that use
2990 \p or \P will have to recompile them.
2991
2992 22. Added "Any" and "L&" to the supported property types.
2993
2994 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
2995 but give a compile time error if the value is greater than 0xff.
2996
2997 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
2998 accidentally not being installed or uninstalled.
2999
3000 25. The pcre.h file was built from pcre.h.in, but the only changes that were
3001 made were to insert the current release number. This seemed silly, because
3002 it made things harder for people building PCRE on systems that don't run
3003 "configure". I have turned pcre.h into a distributed file, no longer built
3004 by "configure", with the version identification directly included. There is
3005 no longer a pcre.h.in file.
3006
3007 However, this change necessitated a change to the pcre-config script as
3008 well. It is built from pcre-config.in, and one of the substitutions was the
3009 release number. I have updated configure.ac so that ./configure now finds
3010 the release number by grepping pcre.h.
3011
3012 26. Added the ability to run the tests under valgrind.
3013
3014
3015 Version 6.4 05-Sep-05
3016 ---------------------
3017
3018 1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
3019 "--" to be printed when multiple files were scanned, even when none of the
3020 -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
3021 consider it to be a bug, and have restored the previous behaviour.
3022
3023 2. A couple of code tidies to get rid of compiler warnings.
3024
3025 3. The pcretest program used to cheat by referring to symbols in the library
3026 whose names begin with _pcre_. These are internal symbols that are not
3027 really supposed to be visible externally, and in some environments it is
3028 possible to suppress them. The cheating is now confined to including
3029 certain files from the library's source, which is a bit cleaner.
3030
3031 4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
3032 file's purpose clearer.
3033
3034 5. Reorganized pcre_ucp_findchar().
3035
3036
3037 Version 6.3 15-Aug-05
3038 ---------------------
3039
3040 1. The file libpcre.pc.in did not have general read permission in the tarball.
3041
3042 2. There were some problems when building without C++ support:
3043
3044 (a) If C++ support was not built, "make install" and "make test" still
3045 tried to test it.
3046
3047 (b) There were problems when the value of CXX was explicitly set. Some
3048 changes have been made to try to fix these, and ...
3049
3050 (c) --disable-cpp can now be used to explicitly disable C++ support.
3051
3052 (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
3053 backslash in a target when C++ was disabled. This confuses some
3054 versions of "make", apparently. Using an intermediate variable solves
3055 this. (Same for CPP_LOBJ.)
3056
3057 3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
3058 (non-Windows) now includes $(CFLAGS) because these flags are sometimes
3059 necessary on certain architectures.
3060
3061 4. Added a setting of -export-symbols-regex to the link command to remove
3062 those symbols that are exported in the C sense, but actually are local
3063 within the library, and not documented. Their names all begin with
3064 "_pcre_". This is not a perfect job, because (a) we have to except some
3065 symbols that pcretest ("illegally") uses, and (b) the facility isn't always
3066 available (and never for static libraries). I have made a note to try to
3067 find a way round (a) in the future.
3068
3069
3070 Version 6.2 01-Aug-05
3071 ---------------------
3072
3073 1. There was no test for integer overflow of quantifier values. A construction
3074 such as {1111111111111111} would give undefined results. What is worse, if
3075 a minimum quantifier for a parenthesized subpattern overflowed and became
3076 negative, the calculation of the memory size went wrong. This could have
3077 led to memory overwriting.
3078
3079 2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
3080
3081 3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
3082 operating environments where this matters.
3083
3084 4. Applied Giuseppe Maxia's patch to add additional features for controlling
3085 PCRE options from within the C++ wrapper.
3086
3087 5. Named capturing subpatterns were not being correctly counted when a pattern
3088 was compiled. This caused two problems: (a) If there were more than 100
3089 such subpatterns, the calculation of the memory needed for the whole
3090 compiled pattern went wrong, leading to an overflow error. (b) Numerical
3091 back references of the form \12, where the number was greater than 9, were
3092 not recognized as back references, even though there were sufficient
3093 previous subpatterns.
3094
3095 6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
3096 versions of gcc, e.g. 2.95.4.
3097
3098
3099 Version 6.1 21-Jun-05
3100 ---------------------
3101
3102 1. There was one reference to the variable "posix" in pcretest.c that was not
3103 surrounded by "#if !defined NOPOSIX".
3104
3105 2. Make it possible to compile pcretest without DFA support, UTF8 support, or
3106 the cross-check on the old pcre_info() function, for the benefit of the
3107 cut-down version of PCRE that is currently imported into Exim.
3108
3109 3. A (silly) pattern starting with (?i)(?-i) caused an internal space
3110 allocation error. I've done the easy fix, which wastes 2 bytes for sensible
3111 patterns that start (?i) but I don't think that matters. The use of (?i) is
3112 just an example; this all applies to the other options as well.
3113
3114 4. Since libtool seems to echo the compile commands it is issuing, the output
3115 from "make" can be reduced a bit by putting "@" in front of each libtool
3116 compile command.
3117
3118 5. Patch from the folks at Google for configure.in to be a bit more thorough
3119 in checking for a suitable C++ installation before trying to compile the
3120 C++ stuff. This should fix a reported problem when a compiler was present,
3121 but no suitable headers.
3122
3123 6. The man pages all had just "PCRE" as their title. I have changed them to
3124 be the relevant file name. I have also arranged that these names are
3125 retained in the file doc/pcre.txt, which is a concatenation in text format
3126 of all the man pages except the little individual ones for each function.
3127
3128 7. The NON-UNIX-USE file had not been updated for the different set of source
3129 files that come with release 6. I also added a few comments about the C++
3130 wrapper.
3131
3132
3133 Version 6.0 07-Jun-05
3134 ---------------------
3135
3136 1. Some minor internal re-organization to help with my DFA experiments.
3137
3138 2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
3139 didn't matter for the library itself when fully configured, but did matter
3140 when compiling without UCP support, or within Exim, where the ucp files are
3141 not imported.
3142
3143 3. Refactoring of the library code to split up the various functions into
3144 different source modules. The addition of the new DFA matching code (see
3145 below) to a single monolithic source would have made it really too
3146 unwieldy, quite apart from causing all the code to be include in a
3147 statically linked application, when only some functions are used. This is
3148 relevant even without the DFA addition now that patterns can be compiled in
3149 one application and matched in another.
3150
3151 The downside of splitting up is that there have to be some external
3152 functions and data tables that are used internally in different modules of
3153 the library but which are not part of the API. These have all had their
3154 names changed to start with "_pcre_" so that they are unlikely to clash
3155 with other external names.
3156
3157 4. Added an alternate matching function, pcre_dfa_exec(), which matches using
3158 a different (DFA) algorithm. Although it is slower than the original
3159 function, it does have some advantages for certain types of matching
3160 problem.
3161
3162 5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
3163 including restarting after a partial match.
3164
3165 6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
3166 defined when compiling for Windows was sent to me. I have put it into the
3167 code, though I have no means of testing or verifying it.
3168
3169 7. Added the pcre_refcount() auxiliary function.
3170
3171 8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
3172 match before or at the first newline in the subject string. In pcretest,
3173 the /f option on a pattern can be used to set this.
3174
3175 9. A repeated \w when used in UTF-8 mode with characters greater than 256
3176 would behave wrongly. This has been present in PCRE since release 4.0.
3177
3178 10. A number of changes to the pcregrep command:
3179
3180 (a) Refactored how -x works; insert ^(...)$ instead of setting
3181 PCRE_ANCHORED and checking the length, in preparation for adding
3182 something similar for -w.
3183
3184 (b) Added the -w (match as a word) option.
3185
3186 (c) Refactored the way lines are read and buffered so as to have more
3187 than one at a time available.
3188
3189 (d) Implemented a pcregrep test script.
3190
3191 (e) Added the -M (multiline match) option. This allows patterns to match
3192 over several lines of the subject. The buffering ensures that at least
3193 8K, or the rest of the document (whichever is the shorter) is available
3194 for matching (and similarly the previous 8K for lookbehind assertions).
3195
3196 (f) Changed the --help output so that it now says
3197
3198 -w, --word-regex(p)
3199
3200 instead of two lines, one with "regex" and the other with "regexp"
3201 because that confused at least one person since the short forms are the
3202 same. (This required a bit of code, as the output is generated
3203 automatically from a table. It wasn't just a text change.)
3204
3205 (g) -- can be used to terminate pcregrep options if the next thing isn't an
3206 option but starts with a hyphen. Could be a pattern or a path name
3207 starting with a hyphen, for instance.
3208
3209 (h) "-" can be given as a file name to represent stdin.
3210
3211 (i) When file names are being printed, "(standard input)" is used for
3212 the standard input, for compatibility with GNU grep. Previously
3213 "<stdin>" was used.
3214
3215 (j) The option --label=xxx can be used to supply a name to be used for
3216 stdin when file names are being printed. There is no short form.
3217
3218 (k) Re-factored the options decoding logic because we are going to add
3219 two more options that take data. Such options can now be given in four
3220 different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
3221
3222 (l) Added the -A, -B, and -C options for requesting that lines of context
3223 around matches be printed.
3224
3225 (m) Added the -L option to print the names of files that do not contain
3226 any matching lines, that is, the complement of -l.
3227
3228 (n) The return code is 2 if any file cannot be opened, but pcregrep does
3229 continue to scan other files.
3230
3231 (o) The -s option was incorrectly implemented. For compatibility with other
3232 greps, it now suppresses the error message for a non-existent or non-
3233 accessible file (but not the return code). There is a new option called
3234 -q that suppresses the output of matching lines, which was what -s was
3235 previously doing.
3236
3237 (p) Added --include and --exclude options to specify files for inclusion
3238 and exclusion when recursing.
3239
3240 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
3241 Hopefully, it now does.
3242
3243 12. Missing cast in pcre_study().
3244
3245 13. Added an "uninstall" target to the makefile.
3246
3247 14. Replaced "extern" in the function prototypes in Makefile.in with
3248 "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
3249 world, but is set differently for Windows.
3250
3251 15. Added a second compiling function called pcre_compile2(). The only
3252 difference is that it has an extra argument, which is a pointer to an
3253 integer error code. When there is a compile-time failure, this is set
3254 non-zero, in addition to the error test pointer being set to point to an
3255 error message. The new argument may be NULL if no error number is required
3256 (but then you may as well call pcre_compile(), which is now just a
3257 wrapper). This facility is provided because some applications need a
3258 numeric error indication, but it has also enabled me to tidy up the way
3259 compile-time errors are handled in the POSIX wrapper.
3260
3261 16. Added VPATH=.libs to the makefile; this should help when building with one
3262 prefix path and installing with another. (Or so I'm told by someone who
3263 knows more about this stuff than I do.)
3264
3265 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
3266 passes PCRE_DOTALL to the pcre_compile() function, making the "." character
3267 match everything, including newlines. This is not POSIX-compatible, but
3268 somebody wanted the feature. From pcretest it can be activated by using
3269 both the P and the s flags.
3270
3271 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
3272
3273 19. libpcre.pc was being incorrectly installed as executable.
3274
3275 20. A couple of places in pcretest check for end-of-line by looking for '\n';
3276 it now also looks for '\r' so that it will work unmodified on Windows.
3277
3278 21. Added Google's contributed C++ wrapper to the distribution.
3279
3280 22. Added some untidy missing memory free() calls in pcretest, to keep
3281 Electric Fence happy when testing.
3282
3283
3284
3285 Version 5.0 13-Sep-04
3286 ---------------------
3287
3288 1. Internal change: literal characters are no longer packed up into items
3289 containing multiple characters in a single byte-string. Each character
3290 is now matched using a separate opcode. However, there may be more than one
3291 byte in the character in UTF-8 mode.
3292
3293 2. The pcre_callout_block structure has two new fields: pattern_position and
3294 next_item_length. These contain the offset in the pattern to the next match
3295 item, and its length, respectively.
3296
3297 3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
3298 insertion of callouts before each pattern item. Added the /C option to
3299 pcretest to make use of this.
3300
3301 4. On the advice of a Windows user, the lines
3302
3303 #if defined(_WIN32) || defined(WIN32)
3304 _setmode( _fileno( stdout ), 0x8000 );
3305 #endif /* defined(_WIN32) || defined(WIN32) */
3306
3307 have been added to the source of pcretest. This apparently does useful
3308 magic in relation to line terminators.
3309
3310 5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
3311 for the benefit of those environments where the "b" makes a difference.
3312
3313 6. The icc compiler has the same options as gcc, but "configure" doesn't seem
3314 to know about it. I have put a hack into configure.in that adds in code
3315 to set GCC=yes if CC=icc. This seems to end up at a point in the
3316 generated configure script that is early enough to affect the setting of
3317 compiler options, which is what is needed, but I have no means of testing
3318 whether it really works. (The user who reported this had patched the
3319 generated configure script, which of course I cannot do.)
3320
3321 LATER: After change 22 below (new libtool files), the configure script
3322 seems to know about icc (and also ecc). Therefore, I have commented out
3323 this hack in configure.in.
3324
3325 7. Added support for pkg-config (2 patches were sent in).
3326
3327 8. Negated POSIX character classes that used a combination of internal tables
3328 were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
3329 [[:^ascii]]. Typically, they would match almost any characters. The other
3330 POSIX classes were not broken in this way.
3331
3332 9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
3333 to find the match, as PCRE was deluded into thinking that the match had to
3334 start at the start point or following a newline. The same bug applied to
3335 patterns with negative forward assertions or any backward assertions
3336 preceding ".*" at the start, unless the pattern required a fixed first
3337 character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
3338
3339 10. In UTF-8 mode, when moving forwards in the subject after a failed match
3340 starting at the last subject character, bytes beyond the end of the subject
3341 string were read.
3342
3343 11. Renamed the variable "class" as "classbits" to make life easier for C++
3344 users. (Previously there was a macro definition, but it apparently wasn't
3345 enough.)
3346
3347 12. Added the new field "tables" to the extra data so that tables can be passed
3348 in at exec time, or the internal tables can be re-selected. This allows
3349 a compiled regex to be saved and re-used at a later time by a different
3350 program that might have everything at different addresses.
3351
3352 13. Modified the pcre-config script so that, when run on Solaris, it shows a
3353 -R library as well as a -L library.
3354
3355 14. The debugging options of pcretest (-d on the command line or D on a
3356 pattern) showed incorrect output for anything following an extended class
3357 that contained multibyte characters and which was followed by a quantifier.
3358
3359 15. Added optional support for general category Unicode character properties
3360 via the \p, \P, and \X escapes. Unicode property support implies UTF-8
3361 support. It adds about 90K to the size of the library. The meanings of the
3362 inbuilt class escapes such as \d and \s have NOT been changed.
3363
3364 16. Updated pcredemo.c to include calls to free() to release the memory for the
3365 compiled pattern.
3366
3367 17. The generated file chartables.c was being created in the source directory
3368 instead of in the building directory. This caused the build to fail if the
3369 source directory was different from the building directory, and was
3370 read-only.
3371
3372 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
3373 file. No doubt somebody will tell me if they don't make sense... Also added
3374 Dan Mooney's comments about building on OpenVMS.
3375
3376 19. Added support for partial matching via the PCRE_PARTIAL option for
3377 pcre_exec() and the \P data escape in pcretest.
3378
3379 20. Extended pcretest with 3 new pattern features:
3380
3381 (i) A pattern option of the form ">rest-of-line" causes pcretest to
3382 write the compiled pattern to the file whose name is "rest-of-line".
3383 This is a straight binary dump of the data, with the saved pointer to
3384 the character tables forced to be NULL. The study data, if any, is
3385 written too. After writing, pcretest reads a new pattern.
3386
3387 (ii) If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
3388 compiled pattern from the given file. There must not be any
3389 occurrences of "<" in the file name (pretty unlikely); if there are,
3390 pcretest will instead treat the initial "<" as a pattern delimiter.
3391 After reading in the pattern, pcretest goes on to read data lines as
3392 usual.
3393
3394 (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
3395 and 16-bit fields in a compiled pattern, to simulate a pattern that
3396 was compiled on a host of opposite endianness.
3397
3398 21. The pcre-exec() function can now cope with patterns that were compiled on
3399 hosts of opposite endianness, with this restriction:
3400
3401 As for any compiled expression that is saved and used later, the tables
3402 pointer field cannot be preserved; the extra_data field in the arguments
3403 to pcre_exec() should be used to pass in a tables address if a value
3404 other than the default internal tables were used at compile time.
3405
3406 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
3407 now diagnosed as an error. Previously, most of the time, a negative number
3408 would have been treated as zero, but if in addition "ovector" was passed as
3409 NULL, a crash could occur.
3410
3411 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
3412 new versions from the libtool 1.5 distribution (the last one is a copy of
3413 a file called libtool.m4). This seems to have fixed the need to patch
3414 "configure" to support Darwin 1.3 (which I used to do). However, I still
3415 had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
3416 workstation).
3417
3418 24. Changed the PCRE licence to be the more standard "BSD" licence.
3419
3420
3421 Version 4.5 01-Dec-03
3422 ---------------------
3423
3424 1. There has been some re-arrangement of the code for the match() function so
3425 that it can be compiled in a version that does not call itself recursively.
3426 Instead, it keeps those local variables that need separate instances for
3427 each "recursion" in a frame on the heap, and gets/frees frames whenever it
3428 needs to "recurse". Keeping track of where control must go is done by means
3429 of setjmp/longjmp. The whole thing is implemented by a set of macros that
3430 hide most of the details from the main code, and operates only if
3431 NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
3432 "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
3433 operating.
3434
3435 To make it easier for callers to provide specially tailored get/free
3436 functions for this usage, two new functions, pcre_stack_malloc, and
3437 pcre_stack_free, are used. They are always called in strict stacking order,
3438 and the size of block requested is always the same.
3439
3440 The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
3441 PCRE has been compiled to use the stack or the heap for recursion. The
3442 -C option of pcretest uses this to show which version is compiled.
3443
3444 A new data escape \S, is added to pcretest; it causes the amounts of store
3445 obtained and freed by both kinds of malloc/free at match time to be added
3446 to the output.
3447
3448 2. Changed the locale test to use "fr_FR" instead of "fr" because that's
3449 what's available on my current Linux desktop machine.
3450
3451 3. When matching a UTF-8 string, the test for a valid string at the start has
3452 been extended. If start_offset is not zero, PCRE now checks that it points
3453 to a byte that is the start of a UTF-8 character. If not, it returns
3454 PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
3455 this is necessary because there may be backward assertions in the pattern.
3456 When matching the same subject several times, it may save resources to use
3457 PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
3458
3459 4. The code for checking the validity of UTF-8 strings has been tightened so
3460 that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
3461 containing "overlong sequences".
3462
3463 5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
3464 I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
3465 should have been "&", but it just so happened that all the cases this let
3466 through by mistake were picked up later in the function.
3467
3468 6. I had used a variable called "isblank" - this is a C99 function, causing
3469 some compilers to warn. To avoid this, I renamed it (as "blankclass").
3470
3471 7. Cosmetic: (a) only output another newline at the end of pcretest if it is
3472 prompting; (b) run "./pcretest /dev/null" at the start of the test script
3473 so the version is shown; (c) stop "make test" echoing "./RunTest".
3474
3475 8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
3476
3477 9. The prototype for memmove() for systems that don't have it was using
3478 size_t, but the inclusion of the header that defines size_t was later. I've
3479 moved the #includes for the C headers earlier to avoid this.
3480
3481 10. Added some adjustments to the code to make it easier to compiler on certain
3482 special systems:
3483
3484 (a) Some "const" qualifiers were missing.
3485 (b) Added the macro EXPORT before all exported functions; by default this
3486 is defined to be empty.
3487 (c) Changed the dftables auxiliary program (that builds chartables.c) so
3488 that it reads its output file name as an argument instead of writing
3489 to the standard output and assuming this can be redirected.
3490
3491 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
3492 class containing characters with values greater than 255, PCRE compilation
3493 went into a loop.
3494
3495 12. A recursive reference to a subpattern that was within another subpattern
3496 that had a minimum quantifier of zero caused PCRE to crash. For example,
3497 (x(y(?2))z)? provoked this bug with a subject that got as far as the
3498 recursion. If the recursively-called subpattern itself had a zero repeat,
3499 that was OK.
3500
3501 13. In pcretest, the buffer for reading a data line was set at 30K, but the
3502 buffer into which it was copied (for escape processing) was still set at
3503 1024, so long lines caused crashes.
3504
3505 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
3506 "internal error: code overflow...". This applied to any character class
3507 that was followed by a possessive quantifier.
3508
3509 15. Modified the Makefile to add libpcre.la as a prerequisite for
3510 libpcreposix.la because I was told this is needed for a parallel build to
3511 work.
3512
3513 16. If a pattern that contained .* following optional items at the start was
3514 studied, the wrong optimizing data was generated, leading to matching
3515 errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
3516 matching string must start with a or b or c. The correct conclusion for
3517 this pattern is that a match can start with any character.
3518
3519
3520 Version 4.4 13-Aug-03
3521 ---------------------
3522
3523 1. In UTF-8 mode, a character class containing characters with values between
3524 127 and 255 was not handled correctly if the compiled pattern was studied.
3525 In fixing this, I have also improved the studying algorithm for such
3526 classes (slightly).
3527
3528 2. Three internal functions had redundant arguments passed to them. Removal
3529 might give a very teeny performance improvement.
3530
3531 3. Documentation bug: the value of the capture_top field in a callout is *one
3532 more than* the number of the hightest numbered captured substring.
3533
3534 4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
3535 in incorrectly linking with a previously installed version. They now link
3536 explicitly with libpcre.la.
3537
3538 5. configure.in no longer needs to recognize Cygwin specially.
3539
3540 6. A problem in pcre.in for Windows platforms is fixed.
3541
3542 7. If a pattern was successfully studied, and the -d (or /D) flag was given to
3543 pcretest, it used to include the size of the study block as part of its
3544 output. Unfortunately, the structure contains a field that has a different
3545 size on different hardware architectures. This meant that the tests that
3546 showed this size failed. As the block is currently always of a fixed size,
3547 this information isn't actually particularly useful in pcretest output, so
3548 I have just removed it.
3549
3550 8. Three pre-processor statements accidentally did not start in column 1.
3551 Sadly, there are *still* compilers around that complain, even though
3552 standard C has not required this for well over a decade. Sigh.
3553
3554 9. In pcretest, the code for checking callouts passed small integers in the
3555 callout_data field, which is a void * field. However, some picky compilers
3556 complained about the casts involved for this on 64-bit systems. Now
3557 pcretest passes the address of the small integer instead, which should get
3558 rid of the warnings.
3559
3560 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
3561 both compile and run time, and gives an error if an invalid UTF-8 sequence
3562 is found. There is a option for disabling this check in cases where the
3563 string is known to be correct and/or the maximum performance is wanted.
3564
3565 11. In response to a bug report, I changed one line in Makefile.in from
3566
3567 -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
3568 to
3569 -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
3570
3571 to look similar to other lines, but I have no way of telling whether this
3572 is the right thing to do, as I do not use Windows. No doubt I'll get told
3573 if it's wrong...
3574
3575
3576 Version 4.3 21-May-03
3577 ---------------------
3578
3579 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
3580 Makefile.
3581
3582 2. Some refactoring to improve the quality of the code:
3583
3584 (i) The utf8_table... variables are now declared "const".
3585
3586 (ii) The code for \cx, which used the "case flipping" table to upper case
3587 lower case letters, now just substracts 32. This is ASCII-specific,
3588 but the whole concept of \cx is ASCII-specific, so it seems
3589 reasonable.
3590
3591 (iii) PCRE was using its character types table to recognize decimal and
3592 hexadecimal digits in the pattern. This is silly, because it handles
3593 only 0-9, a-f, and A-F, but the character types table is locale-
3594 specific, which means strange things might happen. A private
3595 table is now used for this - though it costs 256 bytes, a table is
3596 much faster than multiple explicit tests. Of course, the standard
3597 character types table is still used for matching digits in subject
3598 strings against \d.
3599
3600 (iv) Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
3601 ending in _t are). So I've renamed it as ESC_tee.
3602
3603 3. The first argument for regexec() in the POSIX wrapper should have been
3604 defined as "const".
3605
3606 4. Changed pcretest to use malloc() for its buffers so that they can be
3607 Electric Fenced for debugging.
3608
3609 5. There were several places in the code where, in UTF-8 mode, PCRE would try
3610 to read one or more bytes before the start of the subject string. Often this
3611 had no effect on PCRE's behaviour, but in some circumstances it could
3612 provoke a segmentation fault.
3613
3614 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
3615 to try to read one or more bytes before the start of the subject string.
3616
3617 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
3618 UTF-8 support could misbehave in various ways if the subject string
3619 contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
3620 area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
3621 back over UTF-8 characters.)
3622
3623
3624 Version 4.2 14-Apr-03
3625 ---------------------
3626
3627 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
3628
3629 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
3630 [ON_WINDOWS]: new variable, "#" on non-Windows platforms
3631 [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
3632 [WIN_PREFIX]: new variable, "cyg" for Cygwin
3633 * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
3634 and BUILD_EXEEXT
3635 Note: automatic setting of the BUILD variables is not yet working
3636 set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
3637 compile-time but not at link-time
3638 [LINK]: use for linking executables only
3639 make different versions for Windows and non-Windows
3640 [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
3641 libraries
3642 [LINK_FOR_BUILD]: new variable
3643 [OBJEXT]: use throughout
3644 [EXEEXT]: use throughout
3645 <winshared>: new target
3646 <wininstall>: new target
3647 <dftables.o>: use native compiler
3648 <dftables>: use native linker
3649 <install>: handle Windows platform correctly
3650 <clean>: ditto
3651 <check>: ditto
3652 copy DLL to top builddir before testing
3653
3654 As part of these changes, -no-undefined was removed again. This was reported
3655 to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
3656 in any case.
3657
3658 3. Some tidies to get rid of compiler warnings:
3659
3660 . In the match_data structure, match_limit was an unsigned long int, whereas
3661 match_call_count was an int. I've made them both unsigned long ints.
3662
3663 . In pcretest the fact that a const uschar * doesn't automatically cast to
3664 a void * provoked a warning.
3665
3666 . Turning on some more compiler warnings threw up some "shadow" variables
3667 and a few more missing casts.
3668
3669 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
3670 option, a class that contained a single character with a value between 128
3671 and 255 (e.g. /[\xFF]/) caused PCRE to crash.
3672
3673 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
3674 option, a class that contained several characters, but with at least one
3675 whose value was between 128 and 255 caused PCRE to crash.
3676
3677
3678 Version 4.1 12-Mar-03
3679 ---------------------
3680
3681 1. Compiling with gcc -pedantic found a couple of places where casts were
3682 needed, and a string in dftables.c that was longer than standard compilers are
3683 required to support.
3684
3685 2. Compiling with Sun's compiler found a few more places where the code could
3686 be tidied up in order to avoid warnings.
3687
3688 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
3689 first of these names is deprecated in the latest Autoconf in favour of the name
3690 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
3691 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
3692 analogy I have changed it to CFLAGS_FOR_BUILD.
3693
3694 4. Added -no-undefined to the linking command in the Makefile, because this is
3695 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
3696 linking step for the pcreposix library.
3697
3698 5. PCRE was failing to diagnose the case of two named groups with the same
3699 name.
3700
3701 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
3702 literal character that is needed in the subject for a match, and scans along to
3703 ensure that it is present before embarking on the full matching process. This
3704 saves time in cases of nested unlimited repeats that are never going to match.
3705 Problem: the scan can take a lot of time if the subject is very long (e.g.
3706 megabytes), thus penalizing straightforward matches. It is now done only if the
3707 amount of subject to be scanned is less than 1000 bytes.
3708
3709 7. A lesser problem with the same optimization is that it was recording the
3710 first character of an anchored pattern as "needed", thus provoking a search
3711 right along the subject, even when the first match of the pattern was going to
3712 fail. The "needed" character is now not set for anchored patterns, unless it
3713 follows something in the pattern that is of non-fixed length. Thus, it still
3714 fulfils its original purpose of finding quick non-matches in cases of nested
3715 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
3716
3717
3718 Version 4.0 17-Feb-03
3719 ---------------------
3720
3721 1. If a comment in an extended regex that started immediately after a meta-item
3722 extended to the end of string, PCRE compiled incorrect data. This could lead to
3723 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
3724
3725 2. Moved to autoconf 2.53 and libtool 1.4.2.
3726
3727 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
3728 the special perltest8 script is no longer needed - all the tests can be run
3729 from a single perltest script.
3730
3731 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
3732 by \s. It has now been removed in PCRE. This means it isn't recognized as
3733 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
3734 class [:space:] *does* include VT, thereby creating a mess.
3735
3736 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
3737 space and tab.
3738
3739 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
3740 its new features into the main test script, reducing the number of scripts.
3741
3742 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
3743 were backward compatible, and made the (?i) apply to the whole pattern, as if
3744 /i were given. Now it behaves more logically, and applies the option setting
3745 only to what follows. PCRE has been changed to follow suit. However, if it
3746 finds options settings right at the start of the pattern, it extracts them into
3747 the global options, as before. Thus, they show up in the info data.
3748
3749 8. Added support for the \Q...\E escape sequence. Characters in between are
3750 treated as literals. This is slightly different from Perl in that $ and @ are
3751 also handled as literals inside the quotes. In Perl, they will cause variable
3752 interpolation. Note the following examples:
3753
3754 Pattern PCRE matches Perl matches
3755
3756 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
3757 \Qabc\$xyz\E abc\$xyz abc\$xyz
3758 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
3759
3760 For compatibility with Perl, \Q...\E sequences are recognized inside character
3761 classes as well as outside them.
3762
3763 9. Re-organized 3 code statements in pcretest to avoid "overflow in
3764 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
3765 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
3766 signed/unsigned warnings.
3767
3768 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
3769 option for pcretest, so I've replaced it by a simple function that does just
3770 that job.
3771
3772 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
3773 "pcregrep -".
3774
3775 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
3776 Java package. This provides some syntactic sugar for simple cases of what my
3777 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
3778 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
3779 item, you can use this simplified notation. Note that only makes sense with
3780 greedy quantifiers. Consequently, the use of the possessive quantifier forces
3781 greediness, whatever the setting of the PCRE_UNGREEDY option.
3782
3783 13. A change of greediness default within a pattern was not taking effect at
3784 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
3785 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
3786 was abstracted outside.
3787
3788 14. PCRE now supports the \G assertion. It is true when the current matching
3789 position is at the start point of the match. This differs from \A when the
3790 starting offset is non-zero. Used with the /g option of pcretest (or similar
3791 code), it works in the same way as it does for Perl's /g option. If all
3792 alternatives of a regex begin with \G, the expression is anchored to the start
3793 match position, and the "anchored" flag is set in the compiled expression.
3794
3795 15. Some bugs concerning the handling of certain option changes within patterns
3796 have been fixed. These applied to options other than (?ims). For example,
3797 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
3798 been the other way round. Some of this was related to change 7 above.
3799
3800 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
3801 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
3802 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
3803 POSIX classes only within a class (e.g. /[[:alpha:]]/).
3804
3805 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
3806 mode. Unlike ".", it always matches newline, whatever the setting of
3807 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
3808 assertions. Perl allows it, but it doesn't (in general) work because it can't
3809 calculate the length of the lookbehind. At least, that's the case for Perl
3810 5.8.0 - I've been told they are going to document that it doesn't work in
3811 future.
3812
3813 18. Added an error diagnosis for escapes that PCRE does not support: these are
3814 \L, \l, \N, \P, \p, \U, \u, and \X.
3815
3816 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
3817 reading past the end of the pattern in cases such as /[abcd/.
3818
3819 20. PCRE was getting more memory than necessary for patterns with classes that
3820 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
3821
3822 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
3823 compiling PCRE for use with Virtual Pascal.
3824
3825 22. Small fix to the Makefile to make it work properly if the build is done
3826 outside the source tree.
3827
3828 23. Added a new extension: a condition to go with recursion. If a conditional
3829 subpattern starts with (?(R) the "true" branch is used if recursion has
3830 happened, whereas the "false" branch is used only at the top level.
3831
3832 24. When there was a very long string of literal characters (over 255 bytes
3833 without UTF support, over 250 bytes with UTF support), the computation of how
3834 much memory was required could be incorrect, leading to segfaults or other
3835 strange effects.
3836
3837 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
3838 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
3839 there was a subsequent back reference to those brackets. This meant that, for
3840 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
3841 possible to check for precisely this case. All we can do is abandon the
3842 optimization if .* occurs inside capturing brackets when there are any back
3843 references whatsoever. (See below for a better fix that came later.)
3844
3845 26. The handling of the optimization for finding the first character of a
3846 non-anchored pattern, and for finding a character that is required later in the
3847 match were failing in some cases. This didn't break the matching; it just
3848 failed to optimize when it could. The way this is done has been re-implemented.
3849
3850 27. Fixed typo in error message for invalid (?R item (it said "(?p").
3851
3852 28. Added a new feature that provides some of the functionality that Perl
3853 provides with (?{...}). The facility is termed a "callout". The way it is done
3854 in PCRE is for the caller to provide an optional function, by setting
3855 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
3856 global variable. By default it is unset, which disables all calling out. To get
3857 the function called, the regex must include (?C) at appropriate points. This
3858 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
3859 This provides a means of identifying different callout points. When PCRE
3860 reaches such a point in the regex, if pcre_callout has been set, the external
3861 function is called. It is provided with data in a structure called
3862 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
3863 matching continues; if it returns a non-zero value, the match at the current
3864 point fails. However, backtracking will occur if possible. [This was changed
3865 later and other features added - see item 49 below.]
3866
3867 29. pcretest is upgraded to test the callout functionality. It provides a
3868 callout function that displays information. By default, it shows the start of
3869 the match and the current position in the text. There are some new data escapes
3870 to vary what happens:
3871
3872 \C+ in addition, show current contents of captured substrings
3873 \C- do not supply a callout function
3874 \C!n return 1 when callout number n is reached
3875 \C!n!m return 1 when callout number n is reached for the mth time
3876
3877 30. If pcregrep was called with the -l option and just a single file name, it
3878 output "<stdin>" if a match was found, instead of the file name.
3879
3880 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
3881 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
3882 pcre_exec(). This saves a malloc/free per call. The default value of
3883 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
3884 when configuring.
3885
3886 32. The default maximum size of a compiled pattern is 64K. There have been a
3887 few cases of people hitting this limit. The code now uses macros to handle the
3888 storing of links as offsets within the compiled pattern. It defaults to 2-byte
3889 links, but this can be changed to 3 or 4 bytes by --with-link-size when
3890 configuring. Tests 2 and 5 work only with 2-byte links because they output
3891 debugging information about compiled patterns.
3892
3893 33. Internal code re-arrangements:
3894
3895 (a) Moved the debugging function for printing out a compiled regex into
3896 its own source file (printint.c) and used #include to pull it into
3897 pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
3898 separate copies.
3899
3900 (b) Defined the list of op-code names for debugging as a macro in
3901 internal.h so that it is next to the definition of the opcodes.
3902
3903 (c) Defined a table of op-code lengths for simpler skipping along compiled
3904 code. This is again a macro in internal.h so that it is next to the
3905 definition of the opcodes.
3906
3907 34. Added support for recursive calls to individual subpatterns, along the
3908 lines of Robin Houston's patch (but implemented somewhat differently).
3909
3910 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
3911 allow it to read and process whole directories in Win32. This code was
3912 contributed by Lionel Fourquaux; it has not been tested by me.
3913
3914 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
3915 used to name a group. Names consist of alphanumerics and underscores, and must
3916 be unique. Back references use the syntax (?P=name) and recursive calls use
3917 (?P>name) which is a PCRE extension to the Python extension. Groups still have
3918 numbers. The function pcre_fullinfo() can be used after compilation to extract
3919 a name/number map. There are three relevant calls:
3920
3921 PCRE_INFO_NAMEENTRYSIZE yields the size of each entry in the map
3922 PCRE_INFO_NAMECOUNT yields the number of entries
3923 PCRE_INFO_NAMETABLE yields a pointer to the map.
3924
3925 The map is a vector of fixed-size entries. The size of each entry depends on
3926 the length of the longest name used. The first two bytes of each entry are the
3927 group number, most significant byte first. There follows the corresponding
3928 name, zero terminated. The names are in alphabetical order.
3929
3930 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
3931 case instead of 255. Making it the same both with and without UTF-8 support
3932 means that the same test output works with both.
3933
3934 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
3935 calling malloc() with a zero argument.
3936
3937 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
3938 optimization. I've improved things by keeping a bitmap of backreferences with
3939 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
3940 fact referenced, the optimization can be applied. It is unlikely that a
3941 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
3942 the match to follow \n) will appear inside brackets with a number greater than
3943 31, but if it does, any back reference > 31 suppresses the optimization.
3944
3945 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
3946 of disabling numbered capturing parentheses. Any opening parenthesis that is
3947 not followed by ? behaves as if it were followed by ?: but named parentheses
3948 can still be used for capturing (and they will acquire numbers in the usual
3949 way).
3950
3951 41. Redesigned the return codes from the match() function into yes/no/error so
3952 that errors can be passed back from deep inside the nested calls. A malloc
3953 failure while inside a recursive subpattern call now causes the
3954 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
3955
3956 42. It is now possible to set a limit on the number of times the match()
3957 function is called in a call to pcre_exec(). This facility makes it possible to
3958 limit the amount of recursion and backtracking, though not in a directly
3959 obvious way, because the match() function is used in a number of different
3960 circumstances. The count starts from zero for each position in the subject
3961 string (for non-anchored patterns). The default limit is, for compatibility, a
3962 large number, namely 10 000 000. You can change this in two ways:
3963
3964 (a) When configuring PCRE before making, you can use --with-match-limit=n
3965 to set a default value for the compiled library.
3966
3967 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
3968 a different value is set. See 45 below.
3969
3970 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
3971
3972 43. Added a new function pcre_config(int, void *) to enable run-time extraction
3973 of things that can be changed at compile time. The first argument specifies
3974 what is wanted and the second points to where the information is to be placed.
3975 The current list of available information is:
3976
3977 PCRE_CONFIG_UTF8
3978
3979 The output is an integer that is set to one if UTF-8 support is available;
3980 otherwise it is set to zero.
3981
3982 PCRE_CONFIG_NEWLINE
3983
3984 The output is an integer that it set to the value of the code that is used for
3985 newline. It is either LF (10) or CR (13).
3986
3987 PCRE_CONFIG_LINK_SIZE
3988
3989 The output is an integer that contains the number of bytes used for internal
3990 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
3991
3992 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
3993
3994 The output is an integer that contains the threshold above which the POSIX
3995 interface uses malloc() for output vectors. See item 31 above.
3996
3997 PCRE_CONFIG_MATCH_LIMIT
3998
3999 The output is an unsigned integer that contains the default limit of the number
4000 of match() calls in a pcre_exec() execution. See 42 above.
4001
4002 44. pcretest has been upgraded by the addition of the -C option. This causes it
4003 to extract all the available output from the new pcre_config() function, and to
4004 output it. The program then exits immediately.
4005
4006 45. A need has arisen to pass over additional data with calls to pcre_exec() in
4007 order to support additional features. One way would have been to define
4008 pcre_exec2() (for example) with extra arguments, but this would not have been
4009 extensible, and would also have required all calls to the original function to
4010 be mapped to the new one. Instead, I have chosen to extend the mechanism that
4011 is used for passing in "extra" data from pcre_study().
4012
4013 The pcre_extra structure is now exposed and defined in pcre.h. It currently
4014 contains the following fields:
4015
4016 flags a bitmap indicating which of the following fields are set
4017 study_data opaque data from pcre_study()
4018 match_limit a way of specifying a limit on match() calls for a specific
4019 call to pcre_exec()
4020 callout_data data for callouts (see 49 below)
4021
4022 The flag bits are also defined in pcre.h, and are
4023
4024 PCRE_EXTRA_STUDY_DATA
4025 PCRE_EXTRA_MATCH_LIMIT
4026 PCRE_EXTRA_CALLOUT_DATA
4027
4028 The pcre_study() function now returns one of these new pcre_extra blocks, with
4029 the actual study data pointed to by the study_data field, and the
4030 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
4031 before. That is, this change is entirely upwards-compatible and requires no
4032 change to existing code.
4033
4034 If you want to pass in additional data to pcre_exec(), you can either place it
4035 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
4036 block.
4037
4038 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
4039 data string contains the escape sequence \M, pcretest calls pcre_exec() several
4040 times with different match limits, until it finds the minimum value needed for
4041 pcre_exec() to complete. The value is then output. This can be instructive; for
4042 most simple matches the number is quite small, but for pathological cases it
4043 gets very large very quickly.
4044
4045 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
4046 returns the size of the data block pointed to by the study_data field in a
4047 pcre_extra block, that is, the value that was passed as the argument to
4048 pcre_malloc() when PCRE was getting memory in which to place the information
4049 created by pcre_study(). The fourth argument should point to a size_t variable.
4050 pcretest has been extended so that this information is shown after a successful
4051 pcre_study() call when information about the compiled regex is being displayed.
4052
4053 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
4054 because what follows is always an absolute path. (Later: it turns out that this
4055 is more than cosmetic for MinGW, because it doesn't like empty path
4056 components.)
4057
4058 49. Some changes have been made to the callout feature (see 28 above):
4059
4060 (i) A callout function now has three choices for what it returns:
4061
4062 0 => success, carry on matching
4063 > 0 => failure at this point, but backtrack if possible
4064 < 0 => serious error, return this value from pcre_exec()
4065
4066 Negative values should normally be chosen from the set of PCRE_ERROR_xxx
4067 values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
4068 "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
4069 use by callout functions. It will never be used by PCRE itself.
4070
4071 (ii) The pcre_extra structure (see 45 above) has a void * field called
4072 callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
4073 pcre_callout_block structure has a field of the same name. The contents of
4074 the field passed in the pcre_extra structure are passed to the callout
4075 function in the corresponding field in the callout block. This makes it
4076 easier to use the same callout-containing regex from multiple threads. For
4077 testing, the pcretest program has a new data escape
4078
4079 \C*n pass the number n (may be negative) as callout_data
4080
4081 If the callout function in pcretest receives a non-zero value as
4082 callout_data, it returns that value.
4083
4084 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
4085 there were some redundant $(CFLAGS) in commands that are now specified as
4086 $(LINK), which already includes $(CFLAGS).
4087
4088 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
4089 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
4090 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
4091 one-byte characters throughout. Note that case-insensitive matching applies
4092 only to characters whose values are less than 256. PCRE doesn't support the
4093 notion of cases for higher-valued characters.
4094
4095 (i) A character class whose characters are all within 0-255 is handled as
4096 a bit map, and the map is inverted for negative classes. Previously, a
4097 character > 255 always failed to match such a class; however it should
4098 match if the class was a negative one (e.g. [^ab]). This has been fixed.
4099
4100 (ii) A negated character class with a single character < 255 is coded as
4101 "not this character" (OP_NOT). This wasn't working properly when the test
4102 character was multibyte, either singly or repeated.
4103
4104 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
4105 mode, for example: \x{100}{2,3}.
4106
4107 (iv) The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
4108 singly or repeated) now correctly test multibyte characters. However,
4109 PCRE doesn't recognize any characters with values greater than 255 as
4110 digits, spaces, or word characters. Such characters always match \D, \S,
4111 and \W, and never match \d, \s, or \w.
4112
4113 (v) Classes may now contain characters and character ranges with values
4114 greater than 255. For example: [ab\x{100}-\x{400}].
4115
4116 (vi) pcregrep now has a --utf-8 option (synonym -u) which makes it call
4117 PCRE in UTF-8 mode.
4118
4119 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
4120 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
4121 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
4122 value.)
4123
4124 53. The single man page has become too large. I have therefore split it up into
4125 a number of separate man pages. These also give rise to individual HTML pages;
4126 these are now put in a separate directory, and there is an index.html page that
4127 lists them all. Some hyperlinking between the pages has been installed.
4128
4129 54. Added convenience functions for handling named capturing parentheses.
4130
4131 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
4132 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
4133 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
4134 are faulted.
4135
4136 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
4137 calling configure. These values are used when compiling the dftables.c program
4138 which is run to generate the source of the default character tables. They
4139 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
4140 you will need to set these values.
4141
4142 57. Updated the building process for Windows DLL, as provided by Fred Cox.
4143
4144
4145 Version 3.9 02-Jan-02
4146 ---------------------
4147
4148 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
4149
4150 2. If --disable-static was given, the building process failed when trying to
4151 build pcretest and pcregrep. (For some reason it was using libtool to compile
4152 them, which is not right, as they aren't part of the library.)
4153
4154
4155 Version 3.8 18-Dec-01
4156 ---------------------
4157
4158 1. The experimental UTF-8 code was completely screwed up. It was packing the
4159 bytes in the wrong order. How dumb can you get?
4160
4161
4162 Version 3.7 29-Oct-01
4163 ---------------------
4164
4165 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
4166 This caused pcretest, when used on the test data, to segfault. Unfortunately,
4167 this didn't happen under Solaris 8, where I normally test things.
4168
4169 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
4170 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
4171 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
4172 this fix an hour or so after the initial 3.7 release.)
4173
4174
4175 Version 3.6 23-Oct-01
4176 ---------------------
4177
4178 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
4179 offsets passed as NULL with zero offset count.
4180
4181 2. The config.guess and config.sub files had not been updated when I moved to
4182 the latest autoconf.
4183
4184
4185 Version 3.5 15-Aug-01
4186 ---------------------
4187
4188 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
4189 had been forgotten.
4190
4191 2. By using declared but undefined structures, we can avoid using "void"
4192 definitions in pcre.h while keeping the internal definitions of the structures
4193 private.
4194
4195 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
4196 user point of view, this means that both static and shared libraries are built
4197 by default, but this can be individually controlled. More of the work of
4198 handling this static/shared cases is now inside libtool instead of PCRE's make
4199 file.
4200
4201 4. The pcretest utility is now installed along with pcregrep because it is
4202 useful for users (to test regexs) and by doing this, it automatically gets
4203 relinked by libtool. The documentation has been turned into a man page, so
4204 there are now .1, .txt, and .html versions in /doc.
4205
4206 5. Upgrades to pcregrep:
4207 (i) Added long-form option names like gnu grep.
4208 (ii) Added --help to list all options with an explanatory phrase.
4209 (iii) Added -r, --recursive to recurse into sub-directories.
4210 (iv) Added -f, --file to read patterns from a file.
4211
4212 6. pcre_exec() was referring to its "code" argument before testing that
4213 argument for NULL (and giving an error if it was NULL).
4214
4215 7. Upgraded Makefile.in to allow for compiling in a different directory from
4216 the source directory.
4217
4218 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
4219 options bits, the pointer it was passed was to an int instead of to an unsigned
4220 long int. This mattered only on 64-bit systems.
4221
4222 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
4223 generated) instead of pcre.in, which it its source. Also made the same change
4224 in several of the .c files.
4225
4226 10. A new release of gcc defines printf() as a macro, which broke pcretest
4227 because it had an ifdef in the middle of a string argument for printf(). Fixed
4228 by using separate calls to printf().
4229
4230 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
4231 script, to force use of CR or LF instead of \n in the source. On non-Unix
4232 systems, the value can be set in config.h.
4233
4234 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
4235 absolute limit. Changed the text of the error message to make this clear, and
4236 likewise updated the man page.
4237
4238 13. The limit of 99 on the number of capturing subpatterns has been removed.
4239 The new limit is 65535, which I hope will not be a "real" limit.
4240
4241
4242 Version 3.4 22-Aug-00
4243 ---------------------
4244
4245 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
4246
4247 2. Diagnose condition (?(0) as an error instead of crashing on matching.
4248
4249
4250 Version 3.3 01-Aug-00
4251 ---------------------
4252
4253 1. If an octal character was given, but the value was greater than \377, it
4254 was not getting masked to the least significant bits, as documented. This could
4255 lead to crashes in some systems.
4256
4257 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
4258 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
4259
4260 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
4261 These just pass their arguments on to (pcre_free)(), but they are provided
4262 because some uses of PCRE bind it to non-C systems that can call its functions,
4263 but cannot call free() or pcre_free() directly.
4264
4265 4. Add "make test" as a synonym for "make check". Corrected some comments in
4266 the Makefile.
4267
4268 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
4269 Makefile.
4270
4271 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
4272 command called pgrep for grepping around the active processes.
4273
4274 7. Added the beginnings of support for UTF-8 character strings.
4275
4276 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
4277 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
4278 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
4279 out for the ar command.)
4280
4281
4282 Version 3.2 12-May-00
4283 ---------------------
4284
4285 This is purely a bug fixing release.
4286
4287 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
4288 of ZA. This was just one example of several cases that could provoke this bug,
4289 which was introduced by change 9 of version 2.00. The code for breaking
4290 infinite loops after an iteration that matches an empty string was't working
4291 correctly.
4292
4293 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
4294 when matched against abbab (for example). After matching an empty string, it
4295 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
4296 caused it to match further down the string than it should.
4297
4298 3. The code contained an inclusion of sys/types.h. It isn't clear why this
4299 was there because it doesn't seem to be needed, and it causes trouble on some
4300 systems, as it is not a Standard C header. It has been removed.
4301
4302 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
4303 were reported on the Macintosh. The changes were from
4304
4305 while ((c = *(++ptr)) != 0 && c != '\n');
4306 to
4307 while ((c = *(++ptr)) != 0 && c != '\n') ;
4308
4309 Totally extraordinary, but if that's what it takes...
4310
4311 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
4312 available. Added HAVE_BCOPY and an autoconf test for it; if neither
4313 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
4314 assumes the way PCRE uses memmove() (always moving upwards).
4315
4316 6. PCRE is being used in one environment where strchr() is not available. There
4317 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
4318 faster code anyway.
4319
4320
4321 Version 3.1 09-Feb-00
4322 ---------------------
4323
4324 The only change in this release is the fixing of some bugs in Makefile.in for
4325 the "install" target:
4326
4327 (1) It was failing to install pcreposix.h.
4328
4329 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
4330
4331
4332 Version 3.0 01-Feb-00
4333 ---------------------
4334
4335 1. Add support for the /+ modifier to perltest (to output $` like it does in
4336 pcretest).
4337
4338 2. Add support for the /g modifier to perltest.
4339
4340 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
4341 matches null strings.
4342
4343 4. Fix perltest so that it doesn't do unwanted things when fed an empty
4344 pattern. Perl treats empty patterns specially - it reuses the most recent
4345 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
4346 effect.
4347
4348 5. The POSIX interface was broken in that it was just handing over the POSIX
4349 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
4350 required a bigger vector, with some working space on the end. This means that
4351 the POSIX wrapper now has to get and free some memory, and copy the results.
4352
4353 6. Added some simple autoconf support, placing the test data and the
4354 documentation in separate directories, re-organizing some of the
4355 information files, and making it build pcre-config (a GNU standard). Also added
4356 libtool support for building PCRE as a shared library, which is now the
4357 default.
4358
4359 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
4360 09 are not valid octal constants. Single digits will be used for minor values
4361 less than 10.
4362
4363 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
4364 existing programs that set these in the POSIX interface can use PCRE without
4365 modification.
4366
4367 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
4368 return all that pcre_info() returns, plus additional data. The pcre_info()
4369 function is retained for compatibility, but is considered to be obsolete.
4370
4371 10. Added experimental recursion feature (?R) to handle one common case that
4372 Perl 5.6 will be able to do with (?p{...}).
4373
4374 11. Added support for POSIX character classes like [:alpha:], which Perl is
4375 adopting.
4376
4377
4378 Version 2.08 31-Aug-99
4379 ----------------------
4380
4381 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
4382 trying to match at the startoffset position, but instead was moving forward to
4383 the next newline as if a previous match had failed.
4384
4385 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
4386 and could get into a loop if a null string was matched other than at the start
4387 of the subject.
4388
4389 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
4390 be distinguished at compile time, and for completeness also added PCRE_DATE.
4391
4392 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
4393 in GnuWin32 environments.
4394
4395
4396 Version 2.07 29-Jul-99
4397 ----------------------
4398
4399 1. The documentation is now supplied in plain text form and HTML as well as in
4400 the form of man page sources.
4401
4402 2. C++ compilers don't like assigning (void *) values to other pointer types.
4403 In particular this affects malloc(). Although there is no problem in Standard
4404 C, I've put in casts to keep C++ compilers happy.
4405
4406 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
4407 should be (const char *).
4408
4409 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
4410 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
4411 However, I haven't made this a standard facility. The documentation doesn't
4412 mention it, and the Makefile doesn't support it.
4413
4414 5. The Makefile now contains an "install" target, with editable destinations at
4415 the top of the file. The pcretest program is not installed.
4416
4417 6. pgrep -V now gives the PCRE version number and date.
4418
4419 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
4420 causing the entire string to be ignored, instead of just the last character.
4421
4422 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
4423 non-matching string, it can take a very, very long time, even for strings of
4424 quite modest length, because of the nested recursion. PCRE now does better in
4425 some of these cases. It does this by remembering the last required literal
4426 character in the pattern, and pre-searching the subject to ensure it is present
4427 before running the real match. In other words, it applies a heuristic to detect
4428 some types of certain failure quickly, and in the above example, if presented
4429 with a string that has no trailing " it gives "no match" very quickly.
4430
4431 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
4432 other alternatives are tried instead.
4433
4434
4435 Version 2.06 09-Jun-99
4436 ----------------------
4437
4438 1. Change pcretest's output for amount of store used to show just the code
4439 space, because the remainder (the data block) varies in size between 32-bit and
4440 64-bit systems.
4441
4442 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
4443 start matching at. This allows lookbehinds to work when searching for multiple
4444 occurrences in a string.
4445
4446 3. Added additional options to pcretest for testing multiple occurrences:
4447
4448 /+ outputs the rest of the string that follows a match
4449 /g loops for multiple occurrences, using the new startoffset argument
4450 /G loops for multiple occurrences by passing an incremented pointer
4451
4452 4. PCRE wasn't doing the "first character" optimization for patterns starting
4453 with \b or \B, though it was doing it for other lookbehind assertions. That is,
4454 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
4455 the letter 'x'. On long subject strings, this gives a significant speed-up.
4456
4457
4458 Version 2.05 21-Apr-99
4459 ----------------------
4460
4461 1. Changed the type of magic_number from int to long int so that it works
4462 properly on 16-bit systems.
4463
4464 2. Fixed a bug which caused patterns starting with .* not to work correctly
4465 when the subject string contained newline characters. PCRE was assuming
4466 anchoring for such patterns in all cases, which is not correct because .* will
4467 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
4468 DOTALL is set at top level; otherwise it knows that patterns starting with .*
4469 must be retried after every newline in the subject.
4470
4471
4472 Version 2.04 18-Feb-99
4473 ----------------------
4474
4475 1. For parenthesized subpatterns with repeats whose minimum was zero, the
4476 computation of the store needed to hold the pattern was incorrect (too large).
4477 If such patterns were nested a few deep, this could multiply and become a real
4478 problem.
4479
4480 2. Added /M option to pcretest to show the memory requirement of a specific
4481 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
4482
4483 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
4484 compiled in such a way that the backtracking after subsequent failure was
4485 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
4486 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
4487
4488
4489 Version 2.03 02-Feb-99
4490 ----------------------
4491
4492 1. Fixed typo and small mistake in man page.
4493
4494 2. Added 4th condition (GPL supersedes if conflict) and created separate
4495 LICENCE file containing the conditions.
4496
4497 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
4498 Perl, that is the internal \ allows the delimiter to be included in the
4499 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
4500 the final delimiter, add \ to the end of the pattern (to test the error).
4501
4502 4. Added the convenience functions for extracting substrings after a successful
4503 match. Updated pcretest to make it able to test these functions.
4504
4505
4506 Version 2.02 14-Jan-99
4507 ----------------------
4508
4509 1. Initialized the working variables associated with each extraction so that
4510 their saving and restoring doesn't refer to uninitialized store.
4511
4512 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
4513 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
4514 fix the problem.
4515
4516 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
4517 calls, and wasn't printing the correct value for compiling calls. Increased the
4518 default value of LOOPREPEAT, and the number of significant figures in the
4519 times.
4520
4521 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
4522
4523 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
4524 a building problem on Windows NT with a FAT file system.
4525
4526
4527 Version 2.01 21-Oct-98
4528 ----------------------
4529
4530 1. Changed the API for pcre_compile() to allow for the provision of a pointer
4531 to character tables built by pcre_maketables() in the current locale. If NULL
4532 is passed, the default tables are used.
4533
4534
4535 Version 2.00 24-Sep-98
4536 ----------------------
4537
4538 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
4539 it any more.
4540
4541 2. Allow quantification of (?>) groups, and make it work correctly.
4542
4543 3. The first character computation wasn't working for (?>) groups.
4544
4545 4. Correct the implementation of \Z (it is permitted to match on the \n at the
4546 end of the subject) and add 5.005's \z, which really does match only at the
4547 very end of the subject.
4548
4549 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
4550
4551 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
4552 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
4553 localized options. All options to pcre_study() were also removed.
4554
4555 7. Add other new features from 5.005:
4556
4557 $(?<= positive lookbehind
4558 $(?<! negative lookbehind
4559 (?imsx-imsx) added the unsetting capability
4560 such a setting is global if at outer level; local otherwise
4561 (?imsx-imsx:) non-capturing groups with option setting
4562 (?(cond)re|re) conditional pattern matching
4563
4564 A backreference to itself in a repeated group matches the previous
4565 captured string.
4566
4567 8. General tidying up of studying (both automatic and via "study")
4568 consequential on the addition of new assertions.
4569
4570 9. As in 5.005, unlimited repeated groups that could match an empty substring
4571 are no longer faulted at compile time. Instead, the loop is forcibly broken at
4572 runtime if any iteration does actually match an empty substring.
4573
4574 10. Include the RunTest script in the distribution.
4575
4576 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
4577 discrepancies, some of which were old and were also with respect to 5.004. They
4578 have now been fixed.
4579
4580
4581 Version 1.09 28-Apr-98
4582 ----------------------
4583
4584 1. A negated single character class followed by a quantifier with a minimum
4585 value of one (e.g. [^x]{1,6} ) was not compiled correctly. This could lead to
4586 program crashes, or just wrong answers. This did not apply to negated classes
4587 containing more than one character, or to minima other than one.
4588
4589
4590 Version 1.08 27-Mar-98
4591 ----------------------
4592
4593 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
4594
4595 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
4596 latter must appear before anything that relies on it in the pattern.
4597
4598
4599 Version 1.07 16-Feb-98
4600 ----------------------
4601
4602 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
4603 repeat of a potentially empty string).
4604
4605
4606 Version 1.06 23-Jan-98
4607 ----------------------
4608
4609 1. Added Markus Oberhumer's little patches for C++.
4610
4611 2. Literal strings longer than 255 characters were broken.
4612
4613
4614 Version 1.05 23-Dec-97
4615 ----------------------
4616
4617 1. Negated character classes containing more than one character were failing if
4618 PCRE_CASELESS was set at run time.
4619
4620
4621 Version 1.04 19-Dec-97
4622 ----------------------
4623
4624 1. Corrected the man page, where some "const" qualifiers had been omitted.
4625
4626 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
4627 input syntax.
4628
4629 3. Fixed memory leak which occurred when a regex with back references was
4630 matched with an offsets vector that wasn't big enough. The temporary memory
4631 that is used in this case wasn't being freed if the match failed.
4632
4633 4. Tidied pcretest to ensure it frees memory that it gets.
4634
4635 5. Temporary memory was being obtained in the case where the passed offsets
4636 vector was exactly big enough.
4637
4638 6. Corrected definition of offsetof() from change 5 below.
4639
4640 7. I had screwed up change 6 below and broken the rules for the use of
4641 setjmp(). Now fixed.
4642
4643
4644 Version 1.03 18-Dec-97
4645 ----------------------
4646
4647 1. A erroneous regex with a missing opening parenthesis was correctly
4648 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
4649 on some systems.
4650
4651 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
4652 it was reported that one broken compiler failed on the former because "code" is
4653 also an independent variable.
4654
4655 3. The erroneous regex a[]b caused an array overrun reference.
4656
4657 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
4658 fail on data ending with that character. (It was going on too far, and checking
4659 the next character, typically a binary zero.) This was specific to the
4660 optimized code for single-character negative classes.
4661
4662 5. Added a contributed patch from the TIN world which does the following:
4663
4664 + Add an undef for memmove, in case the the system defines a macro for it.
4665
4666 + Add a definition of offsetof(), in case there isn't one. (I don't know
4667 the reason behind this - offsetof() is part of the ANSI standard - but
4668 it does no harm).
4669
4670 + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
4671 most of the places where whitespace preceded '#'. I have given up and
4672 allowed the remaining 2 cases to be at the margin.
4673
4674 + Rename some variables in pcre to eliminate shadowing. This seems very
4675 pedantic, but does no harm, of course.
4676
4677 6. Moved the call to setjmp() into its own function, to get rid of warnings
4678 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
4679
4680 7. Constructs such as \d{8,} were compiling into the equivalent of
4681 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
4682 outcome, but in this particular case used more store than had been allocated,
4683 which caused the bug to be discovered because it threw up an internal error.
4684
4685 8. The debugging code in both pcre and pcretest for outputting the compiled
4686 form of a regex was going wrong in the case of back references followed by
4687 curly-bracketed repeats.
4688
4689
4690 Version 1.02 12-Dec-97
4691 ----------------------
4692
4693 1. Typos in pcre.3 and comments in the source fixed.
4694
4695 2. Applied a contributed patch to get rid of places where it used to remove
4696 'const' from variables, and fixed some signed/unsigned and uninitialized
4697 variable warnings.
4698
4699 3. Added the "runtest" target to Makefile.
4700
4701 4. Set default compiler flag to -O2 rather than just -O.
4702
4703
4704 Version 1.01 19-Nov-97
4705 ----------------------
4706
4707 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
4708 like /([ab]*)*/, that is, for classes with more than one character in them.
4709
4710 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
4711 as /((?>a*))*/ (a PCRE_EXTRA facility).
4712
4713
4714 Version 1.00 18-Nov-97
4715 ----------------------
4716
4717 1. Added compile-time macros to support systems such as SunOS4 which don't have
4718 memmove() or strerror() but have other things that can be used instead.
4719
4720 2. Arranged that "make clean" removes the executables.
4721
4722
4723 Version 0.99 27-Oct-97
4724 ----------------------
4725
4726 1. Fixed bug in code for optimizing classes with only one character. It was
4727 initializing a 32-byte map regardless, which could cause it to run off the end
4728 of the memory it had got.
4729
4730 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
4731
4732
4733 Version 0.98 22-Oct-97
4734 ----------------------
4735
4736 1. Fixed bug in code for handling temporary memory usage when there are more
4737 back references than supplied space in the ovector. This could cause segfaults.
4738
4739
4740 Version 0.97 21-Oct-97
4741 ----------------------
4742
4743 1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
4744
4745 2. Optimized negated single characters not to use a bit map.
4746
4747 3. Brought error texts together as macro definitions; clarified some of them;
4748 fixed one that was wrong - it said "range out of order" when it meant "invalid
4749 escape sequence".
4750
4751 4. Changed some char * arguments to const char *.
4752
4753 5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
4754
4755 6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
4756 pcretest.
4757
4758
4759 Version 0.96 16-Oct-97
4760 ----------------------
4761
4762 1. Added a simple "pgrep" utility to the distribution.
4763
4764 2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
4765 unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
4766 where "ddd" means "one or more decimal digits".
4767
4768 3. Fixed serious bug. If a pattern had a back reference, but the call to
4769 pcre_exec() didn't supply a large enough ovector to record the related
4770 identifying subpattern, the match always failed. PCRE now remembers the number
4771 of the largest back reference, and gets some temporary memory in which to save
4772 the offsets during matching if necessary, in order to ensure that
4773 backreferences always work.
4774
4775 4. Increased the compatibility with Perl in a number of ways:
4776
4777 (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
4778 to request this handling. The option can be set at compile or exec time.
4779
4780 (b) $ matches before a terminating newline by default; an option
4781 PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
4782 mode). The option can be set at compile or exec time.
4783
4784 (c) The handling of \ followed by a digit other than 0 is now supposed to be
4785 the same as Perl's. If the decimal number it represents is less than 10
4786 or there aren't that many previous left capturing parentheses, an octal
4787 escape is read. Inside a character class, it's always an octal escape,
4788 even if it is a single digit.
4789
4790 (d) An escaped but undefined alphabetic character is taken as a literal,
4791 unless PCRE_EXTRA is set. Currently this just reserves the remaining
4792 escapes.
4793
4794 (e) {0} is now permitted. (The previous item is removed from the compiled
4795 pattern).
4796
4797 5. Changed all the names of code files so that the basic parts are no longer
4798 than 10 characters, and abolished the teeny "globals.c" file.
4799
4800 6. Changed the handling of character classes; they are now done with a 32-byte
4801 bit map always.
4802
4803 7. Added the -d and /D options to pcretest to make it possible to look at the
4804 internals of compilation without having to recompile pcre.
4805
4806
4807 Version 0.95 23-Sep-97
4808 ----------------------
4809
4810 1. Fixed bug in pre-pass concerning escaped "normal" characters such as \x5c or
4811 \x20 at the start of a run of normal characters. These were being treated as
4812 real characters, instead of the source characters being re-checked.
4813
4814
4815 Version 0.94 18-Sep-97
4816 ----------------------
4817
4818 1. The functions are now thread-safe, with the caveat that the global variables
4819 containing pointers to malloc() and free() or alternative functions are the
4820 same for all threads.
4821
4822 2. Get pcre_study() to generate a bitmap of initial characters for non-
4823 anchored patterns when this is possible, and use it if passed to pcre_exec().
4824
4825
4826 Version 0.93 15-Sep-97
4827 ----------------------
4828
4829 1. /(b)|(:+)/ was computing an incorrect first character.
4830
4831 2. Add pcre_study() to the API and the passing of pcre_extra to pcre_exec(),
4832 but not actually doing anything yet.
4833
4834 3. Treat "-" characters in classes that cannot be part of ranges as literals,
4835 as Perl does (e.g. [-az] or [az-]).
4836
4837 4. Set the anchored flag if a branch starts with .* or .*? because that tests
4838 all possible positions.
4839
4840 5. Split up into different modules to avoid including unneeded functions in a
4841 compiled binary. However, compile and exec are still in one module. The "study"
4842 function is split off.
4843
4844 6. The character tables are now in a separate module whose source is generated
4845 by an auxiliary program - but can then be edited by hand if required. There are
4846 now no calls to isalnum(), isspace(), isdigit(), isxdigit(), tolower() or
4847 toupper() in the code.
4848
4849 7. Turn the malloc/free funtions variables into pcre_malloc and pcre_free and
4850 make them global. Abolish the function for setting them, as the caller can now
4851 set them directly.
4852
4853
4854 Version 0.92 11-Sep-97
4855 ----------------------
4856
4857 1. A repeat with a fixed maximum and a minimum of 1 for an ordinary character
4858 (e.g. /a{1,3}/) was broken (I mis-optimized it).
4859
4860 2. Caseless matching was not working in character classes if the characters in
4861 the pattern were in upper case.
4862
4863 3. Make ranges like [W-c] work in the same way as Perl for caseless matching.
4864
4865 4. Make PCRE_ANCHORED public and accept as a compile option.
4866
4867 5. Add an options word to pcre_exec() and accept PCRE_ANCHORED and
4868 PCRE_CASELESS at run time. Add escapes \A and \I to pcretest to cause it to
4869 pass them.
4870
4871 6. Give an error if bad option bits passed at compile or run time.
4872
4873 7. Add PCRE_MULTILINE at compile and exec time, and (?m) as well. Add \M to
4874 pcretest to cause it to pass that flag.
4875
4876 8. Add pcre_info(), to get the number of identifying subpatterns, the stored
4877 options, and the first character, if set.
4878
4879 9. Recognize C+ or C{n,m} where n >= 1 as providing a fixed starting character.
4880
4881
4882 Version 0.91 10-Sep-97
4883 ----------------------
4884
4885 1. PCRE was failing to diagnose unlimited repeats of subpatterns that could
4886 match the empty string as in /(a*)*/. It was looping and ultimately crashing.
4887
4888 2. PCRE was looping on encountering an indefinitely repeated back reference to
4889 a subpattern that had matched an empty string, e.g. /(a|)\1*/. It now does what
4890 Perl does - treats the match as successful.
4891
4892 ****

Properties

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

  ViewVC Help
Powered by ViewVC 1.1.5