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

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1234 - (show annotations)
Sat Dec 8 16:35:42 2012 UTC (7 years, 2 months ago) by ph10
File size: 240249 byte(s)
Error occurred while calculating annotation data.
Applied patch to quash MSVC warnings.
1 ChangeLog for PCRE
2 ------------------
3
4 Version 8.33 xx-xxxx-201x
5 -------------------------
6
7 1. Added 'U' to some constants that are compared to unsigned integers, to
8 avoid compiler signed/unsigned warnings. Added (int) casts to unsigned
9 variables that are added to signed variables, to ensure the result is
10 signed and can be negated.
11
12 2. Applied patch by Daniel Richard G for quashing MSVC warnings to the
13 CMake config files.
14
15
16 Version 8.32 30-November-2012
17 -----------------------------
18
19 1. Improved JIT compiler optimizations for first character search and single
20 character iterators.
21
22 2. Supporting IBM XL C compilers for PPC architectures in the JIT compiler.
23 Patch by Daniel Richard G.
24
25 3. Single character iterator optimizations in the JIT compiler.
26
27 4. Improved JIT compiler optimizations for character ranges.
28
29 5. Rename the "leave" variable names to "quit" to improve WinCE compatibility.
30 Reported by Giuseppe D'Angelo.
31
32 6. The PCRE_STARTLINE bit, indicating that a match can occur only at the start
33 of a line, was being set incorrectly in cases where .* appeared inside
34 atomic brackets at the start of a pattern, or where there was a subsequent
35 *PRUNE or *SKIP.
36
37 7. Improved instruction cache flush for POWER/PowerPC.
38 Patch by Daniel Richard G.
39
40 8. Fixed a number of issues in pcregrep, making it more compatible with GNU
41 grep:
42
43 (a) There is now no limit to the number of patterns to be matched.
44
45 (b) An error is given if a pattern is too long.
46
47 (c) Multiple uses of --exclude, --exclude-dir, --include, and --include-dir
48 are now supported.
49
50 (d) --exclude-from and --include-from (multiple use) have been added.
51
52 (e) Exclusions and inclusions now apply to all files and directories, not
53 just to those obtained from scanning a directory recursively.
54
55 (f) Multiple uses of -f and --file-list are now supported.
56
57 (g) In a Windows environment, the default for -d has been changed from
58 "read" (the GNU grep default) to "skip", because otherwise the presence
59 of a directory in the file list provokes an error.
60
61 (h) The documentation has been revised and clarified in places.
62
63 9. Improve the matching speed of capturing brackets.
64
65 10. Changed the meaning of \X so that it now matches a Unicode extended
66 grapheme cluster.
67
68 11. Patch by Daniel Richard G to the autoconf files to add a macro for sorting
69 out POSIX threads when JIT support is configured.
70
71 12. Added support for PCRE_STUDY_EXTRA_NEEDED.
72
73 13. In the POSIX wrapper regcomp() function, setting re_nsub field in the preg
74 structure could go wrong in environments where size_t is not the same size
75 as int.
76
77 14. Applied user-supplied patch to pcrecpp.cc to allow PCRE_NO_UTF8_CHECK to be
78 set.
79
80 15. The EBCDIC support had decayed; later updates to the code had included
81 explicit references to (e.g.) \x0a instead of CHAR_LF. There has been a
82 general tidy up of EBCDIC-related issues, and the documentation was also
83 not quite right. There is now a test that can be run on ASCII systems to
84 check some of the EBCDIC-related things (but is it not a full test).
85
86 16. The new PCRE_STUDY_EXTRA_NEEDED option is now used by pcregrep, resulting
87 in a small tidy to the code.
88
89 17. Fix JIT tests when UTF is disabled and both 8 and 16 bit mode are enabled.
90
91 18. If the --only-matching (-o) option in pcregrep is specified multiple
92 times, each one causes appropriate output. For example, -o1 -o2 outputs the
93 substrings matched by the 1st and 2nd capturing parentheses. A separating
94 string can be specified by --om-separator (default empty).
95
96 19. Improving the first n character searches.
97
98 20. Turn case lists for horizontal and vertical white space into macros so that
99 they are defined only once.
100
101 21. This set of changes together give more compatible Unicode case-folding
102 behaviour for characters that have more than one other case when UCP
103 support is available.
104
105 (a) The Unicode property table now has offsets into a new table of sets of
106 three or more characters that are case-equivalent. The MultiStage2.py
107 script that generates these tables (the pcre_ucd.c file) now scans
108 CaseFolding.txt instead of UnicodeData.txt for character case
109 information.
110
111 (b) The code for adding characters or ranges of characters to a character
112 class has been abstracted into a generalized function that also handles
113 case-independence. In UTF-mode with UCP support, this uses the new data
114 to handle characters with more than one other case.
115
116 (c) A bug that is fixed as a result of (b) is that codepoints less than 256
117 whose other case is greater than 256 are now correctly matched
118 caselessly. Previously, the high codepoint matched the low one, but not
119 vice versa.
120
121 (d) The processing of \h, \H, \v, and \ in character classes now makes use
122 of the new class addition function, using character lists defined as
123 macros alongside the case definitions of 20 above.
124
125 (e) Caseless back references now work with characters that have more than
126 one other case.
127
128 (f) General caseless matching of characters with more than one other case
129 is supported.
130
131 22. Unicode character properties were updated from Unicode 6.2.0
132
133 23. Improved CMake support under Windows. Patch by Daniel Richard G.
134
135 24. Add support for 32-bit character strings, and UTF-32
136
137 25. Major JIT compiler update (code refactoring and bugfixing).
138 Experimental Sparc 32 support is added.
139
140 26. Applied a modified version of Daniel Richard G's patch to create
141 pcre.h.generic and config.h.generic by "make" instead of in the
142 PrepareRelease script.
143
144 27. Added a definition for CHAR_NULL (helpful for the z/OS port), and use it in
145 pcre_compile.c when checking for a zero character.
146
147 28. Introducing a native interface for JIT. Through this interface, the compiled
148 machine code can be directly executed. The purpose of this interface is to
149 provide fast pattern matching, so several sanity checks are not performed.
150 However, feature tests are still performed. The new interface provides
151 1.4x speedup compared to the old one.
152
153 29. If pcre_exec() or pcre_dfa_exec() was called with a negative value for
154 the subject string length, the error given was PCRE_ERROR_BADOFFSET, which
155 was confusing. There is now a new error PCRE_ERROR_BADLENGTH for this case.
156
157 30. In 8-bit UTF-8 mode, pcretest failed to give an error for data codepoints
158 greater than 0x7fffffff (which cannot be represented in UTF-8, even under
159 the "old" RFC 2279). Instead, it ended up passing a negative length to
160 pcre_exec().
161
162 31. Add support for GCC's visibility feature to hide internal functions.
163
164 32. Running "pcretest -C pcre8" or "pcretest -C pcre16" gave a spurious error
165 "unknown -C option" after outputting 0 or 1.
166
167 33. There is now support for generating a code coverage report for the test
168 suite in environments where gcc is the compiler and lcov is installed. This
169 is mainly for the benefit of the developers.
170
171 34. If PCRE is built with --enable-valgrind, certain memory regions are marked
172 unaddressable using valgrind annotations, allowing valgrind to detect
173 invalid memory accesses. This is mainly for the benefit of the developers.
174
175 25. (*UTF) can now be used to start a pattern in any of the three libraries.
176
177 26. Give configure error if --enable-cpp but no C++ compiler found.
178
179
180 Version 8.31 06-July-2012
181 -------------------------
182
183 1. Fixing a wrong JIT test case and some compiler warnings.
184
185 2. Removed a bashism from the RunTest script.
186
187 3. Add a cast to pcre_exec.c to fix the warning "unary minus operator applied
188 to unsigned type, result still unsigned" that was given by an MS compiler
189 on encountering the code "-sizeof(xxx)".
190
191 4. Partial matching support is added to the JIT compiler.
192
193 5. Fixed several bugs concerned with partial matching of items that consist
194 of more than one character:
195
196 (a) /^(..)\1/ did not partially match "aba" because checking references was
197 done on an "all or nothing" basis. This also applied to repeated
198 references.
199
200 (b) \R did not give a hard partial match if \r was found at the end of the
201 subject.
202
203 (c) \X did not give a hard partial match after matching one or more
204 characters at the end of the subject.
205
206 (d) When newline was set to CRLF, a pattern such as /a$/ did not recognize
207 a partial match for the string "\r".
208
209 (e) When newline was set to CRLF, the metacharacter "." did not recognize
210 a partial match for a CR character at the end of the subject string.
211
212 6. If JIT is requested using /S++ or -s++ (instead of just /S+ or -s+) when
213 running pcretest, the text "(JIT)" added to the output whenever JIT is
214 actually used to run the match.
215
216 7. Individual JIT compile options can be set in pcretest by following -s+[+]
217 or /S+[+] with a digit between 1 and 7.
218
219 8. OP_NOT now supports any UTF character not just single-byte ones.
220
221 9. (*MARK) control verb is now supported by the JIT compiler.
222
223 10. The command "./RunTest list" lists the available tests without actually
224 running any of them. (Because I keep forgetting what they all are.)
225
226 11. Add PCRE_INFO_MAXLOOKBEHIND.
227
228 12. Applied a (slightly modified) user-supplied patch that improves performance
229 when the heap is used for recursion (compiled with --disable-stack-for-
230 recursion). Instead of malloc and free for each heap frame each time a
231 logical recursion happens, frames are retained on a chain and re-used where
232 possible. This sometimes gives as much as 30% improvement.
233
234 13. As documented, (*COMMIT) is now confined to within a recursive subpattern
235 call.
236
237 14. As documented, (*COMMIT) is now confined to within a positive assertion.
238
239 15. It is now possible to link pcretest with libedit as an alternative to
240 libreadline.
241
242 16. (*COMMIT) control verb is now supported by the JIT compiler.
243
244 17. The Unicode data tables have been updated to Unicode 6.1.0.
245
246 18. Added --file-list option to pcregrep.
247
248 19. Added binary file support to pcregrep, including the -a, --binary-files,
249 -I, and --text options.
250
251 20. The madvise function is renamed for posix_madvise for QNX compatibility
252 reasons. Fixed by Giuseppe D'Angelo.
253
254 21. Fixed a bug for backward assertions with REVERSE 0 in the JIT compiler.
255
256 22. Changed the option for creating symbolic links for 16-bit man pages from
257 -s to -sf so that re-installing does not cause issues.
258
259 23. Support PCRE_NO_START_OPTIMIZE in JIT as (*MARK) support requires it.
260
261 24. Fixed a very old bug in pcretest that caused errors with restarted DFA
262 matches in certain environments (the workspace was not being correctly
263 retained). Also added to pcre_dfa_exec() a simple plausibility check on
264 some of the workspace data at the beginning of a restart.
265
266 25. \s*\R was auto-possessifying the \s* when it should not, whereas \S*\R
267 was not doing so when it should - probably a typo introduced by SVN 528
268 (change 8.10/14).
269
270 26. When PCRE_UCP was not set, \w+\x{c4} was incorrectly auto-possessifying the
271 \w+ when the character tables indicated that \x{c4} was a word character.
272 There were several related cases, all because the tests for doing a table
273 lookup were testing for characters less than 127 instead of 255.
274
275 27. If a pattern contains capturing parentheses that are not used in a match,
276 their slots in the ovector are set to -1. For those that are higher than
277 any matched groups, this happens at the end of processing. In the case when
278 there were back references that the ovector was too small to contain
279 (causing temporary malloc'd memory to be used during matching), and the
280 highest capturing number was not used, memory off the end of the ovector
281 was incorrectly being set to -1. (It was using the size of the temporary
282 memory instead of the true size.)
283
284 28. To catch bugs like 27 using valgrind, when pcretest is asked to specify an
285 ovector size, it uses memory at the end of the block that it has got.
286
287 29. Check for an overlong MARK name and give an error at compile time. The
288 limit is 255 for the 8-bit library and 65535 for the 16-bit library.
289
290 30. JIT compiler update.
291
292 31. JIT is now supported on jailbroken iOS devices. Thanks for Ruiger
293 Rill for the patch.
294
295 32. Put spaces around SLJIT_PRINT_D in the JIT compiler. Required by CXX11.
296
297 33. Variable renamings in the PCRE-JIT compiler. No functionality change.
298
299 34. Fixed typos in pcregrep: in two places there was SUPPORT_LIBZ2 instead of
300 SUPPORT_LIBBZ2. This caused a build problem when bzip2 but not gzip (zlib)
301 was enabled.
302
303 35. Improve JIT code generation for greedy plus quantifier.
304
305 36. When /((?:a?)*)*c/ or /((?>a?)*)*c/ was matched against "aac", it set group
306 1 to "aa" instead of to an empty string. The bug affected repeated groups
307 that could potentially match an empty string.
308
309 37. Optimizing single character iterators in JIT.
310
311 38. Wide characters specified with \uxxxx in JavaScript mode are now subject to
312 the same checks as \x{...} characters in non-JavaScript mode. Specifically,
313 codepoints that are too big for the mode are faulted, and in a UTF mode,
314 disallowed codepoints are also faulted.
315
316 39. If PCRE was compiled with UTF support, in three places in the DFA
317 matcher there was code that should only have been obeyed in UTF mode, but
318 was being obeyed unconditionally. In 8-bit mode this could cause incorrect
319 processing when bytes with values greater than 127 were present. In 16-bit
320 mode the bug would be provoked by values in the range 0xfc00 to 0xdc00. In
321 both cases the values are those that cannot be the first data item in a UTF
322 character. The three items that might have provoked this were recursions,
323 possessively repeated groups, and atomic groups.
324
325 40. Ensure that libpcre is explicitly listed in the link commands for pcretest
326 and pcregrep, because some OS require shared objects to be explicitly
327 passed to ld, causing the link step to fail if they are not.
328
329 41. There were two incorrect #ifdefs in pcre_study.c, meaning that, in 16-bit
330 mode, patterns that started with \h* or \R* might be incorrectly matched.
331
332
333 Version 8.30 04-February-2012
334 -----------------------------
335
336 1. Renamed "isnumber" as "is_a_number" because in some Mac environments this
337 name is defined in ctype.h.
338
339 2. Fixed a bug in fixed-length calculation for lookbehinds that would show up
340 only in quite long subpatterns.
341
342 3. Removed the function pcre_info(), which has been obsolete and deprecated
343 since it was replaced by pcre_fullinfo() in February 2000.
344
345 4. For a non-anchored pattern, if (*SKIP) was given with a name that did not
346 match a (*MARK), and the match failed at the start of the subject, a
347 reference to memory before the start of the subject could occur. This bug
348 was introduced by fix 17 of release 8.21.
349
350 5. A reference to an unset group with zero minimum repetition was giving
351 totally wrong answers (in non-JavaScript-compatibility mode). For example,
352 /(another)?(\1?)test/ matched against "hello world test". This bug was
353 introduced in release 8.13.
354
355 6. Add support for 16-bit character strings (a large amount of work involving
356 many changes and refactorings).
357
358 7. RunGrepTest failed on msys because \r\n was replaced by whitespace when the
359 command "pattern=`printf 'xxx\r\njkl'`" was run. The pattern is now taken
360 from a file.
361
362 8. Ovector size of 2 is also supported by JIT based pcre_exec (the ovector size
363 rounding is not applied in this particular case).
364
365 9. The invalid Unicode surrogate codepoints U+D800 to U+DFFF are now rejected
366 if they appear, or are escaped, in patterns.
367
368 10. Get rid of a number of -Wunused-but-set-variable warnings.
369
370 11. The pattern /(?=(*:x))(q|)/ matches an empty string, and returns the mark
371 "x". The similar pattern /(?=(*:x))((*:y)q|)/ did not return a mark at all.
372 Oddly, Perl behaves the same way. PCRE has been fixed so that this pattern
373 also returns the mark "x". This bug applied to capturing parentheses,
374 non-capturing parentheses, and atomic parentheses. It also applied to some
375 assertions.
376
377 12. Stephen Kelly's patch to CMakeLists.txt allows it to parse the version
378 information out of configure.ac instead of relying on pcre.h.generic, which
379 is not stored in the repository.
380
381 13. Applied Dmitry V. Levin's patch for a more portable method for linking with
382 -lreadline.
383
384 14. ZH added PCRE_CONFIG_JITTARGET; added its output to pcretest -C.
385
386 15. Applied Graycode's patch to put the top-level frame on the stack rather
387 than the heap when not using the stack for recursion. This gives a
388 performance improvement in many cases when recursion is not deep.
389
390 16. Experimental code added to "pcretest -C" to output the stack frame size.
391
392
393 Version 8.21 12-Dec-2011
394 ------------------------
395
396 1. Updating the JIT compiler.
397
398 2. JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
399 are added as well.
400
401 3. Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
402 PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
403 calling _pcre_jit_exec. Some extra comments are added.
404
405 4. (*MARK) settings inside atomic groups that do not contain any capturing
406 parentheses, for example, (?>a(*:m)), were not being passed out. This bug
407 was introduced by change 18 for 8.20.
408
409 5. Supporting of \x, \U and \u in JavaScript compatibility mode based on the
410 ECMA-262 standard.
411
412 6. Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
413 erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
414 This bug was probably introduced by change 9 of 8.13.
415
416 7. While fixing 6 above, I noticed that a number of other items were being
417 incorrectly rejected as "not fixed length". This arose partly because newer
418 opcodes had not been added to the fixed-length checking code. I have (a)
419 corrected the bug and added tests for these items, and (b) arranged for an
420 error to occur if an unknown opcode is encountered while checking for fixed
421 length instead of just assuming "not fixed length". The items that were
422 rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
423 (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
424 repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
425
426 8. A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
427 being incorrectly compiled and would have given unpredicatble results.
428
429 9. A possessively repeated subpattern with minimum repeat count greater than
430 one behaved incorrectly. For example, (A){2,}+ behaved as if it was
431 (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
432 the first (A) could occur when it should not.
433
434 10. Add a cast and remove a redundant test from the code.
435
436 11. JIT should use pcre_malloc/pcre_free for allocation.
437
438 12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
439 best practice nowadays, and helps with cross-compiling. (If the exec_prefix
440 is anything other than /usr, -L is still shown).
441
442 13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
443
444 14. Perl does not support \N without a following name in a [] class; PCRE now
445 also gives an error.
446
447 15. If a forward reference was repeated with an upper limit of around 2000,
448 it caused the error "internal error: overran compiling workspace". The
449 maximum number of forward references (including repeats) was limited by the
450 internal workspace, and dependent on the LINK_SIZE. The code has been
451 rewritten so that the workspace expands (via pcre_malloc) if necessary, and
452 the default depends on LINK_SIZE. There is a new upper limit (for safety)
453 of around 200,000 forward references. While doing this, I also speeded up
454 the filling in of repeated forward references.
455
456 16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
457 incorrectly expecting the subject to contain another "a" after the start.
458
459 17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
460 in the match, the SKIP should be ignored. This was not happening; instead
461 the SKIP was being treated as NOMATCH. For patterns such as
462 /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
463 tested.
464
465 18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
466 now much more compatible with Perl, in particular in cases where the result
467 is a non-match for a non-anchored pattern. For example, if
468 /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
469 "m", where previously it did not return a name. A side effect of this
470 change is that for partial matches, the last encountered mark name is
471 returned, as for non matches. A number of tests that were previously not
472 Perl-compatible have been moved into the Perl-compatible test files. The
473 refactoring has had the pleasing side effect of removing one argument from
474 the match() function, thus reducing its stack requirements.
475
476 19. If the /S+ option was used in pcretest to study a pattern using JIT,
477 subsequent uses of /S (without +) incorrectly behaved like /S+.
478
479 21. Retrieve executable code size support for the JIT compiler and fixing
480 some warnings.
481
482 22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
483 not work when the shorter character appeared right at the end of the
484 subject string.
485
486 23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
487 systems.
488
489 24. Added PCRE_INFO_JITSIZE to pass on the value from (21) above, and also
490 output it when the /M option is used in pcretest.
491
492 25. The CheckMan script was not being included in the distribution. Also, added
493 an explicit "perl" to run Perl scripts from the PrepareRelease script
494 because this is reportedly needed in Windows.
495
496 26. If study data was being save in a file and studying had not found a set of
497 "starts with" bytes for the pattern, the data written to the file (though
498 never used) was taken from uninitialized memory and so caused valgrind to
499 complain.
500
501 27. Updated RunTest.bat as provided by Sheri Pierce.
502
503 28. Fixed a possible uninitialized memory bug in pcre_jit_compile.c.
504
505 29. Computation of memory usage for the table of capturing group names was
506 giving an unnecessarily large value.
507
508
509 Version 8.20 21-Oct-2011
510 ------------------------
511
512 1. Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
513 a POSIX class. After further experiments with Perl, which convinced me that
514 Perl has bugs and confusions, a closing square bracket is no longer allowed
515 in a POSIX name. This bug also affected patterns with classes that started
516 with full stops.
517
518 2. If a pattern such as /(a)b|ac/ is matched against "ac", there is no
519 captured substring, but while checking the failing first alternative,
520 substring 1 is temporarily captured. If the output vector supplied to
521 pcre_exec() was not big enough for this capture, the yield of the function
522 was still zero ("insufficient space for captured substrings"). This cannot
523 be totally fixed without adding another stack variable, which seems a lot
524 of expense for a edge case. However, I have improved the situation in cases
525 such as /(a)(b)x|abc/ matched against "abc", where the return code
526 indicates that fewer than the maximum number of slots in the ovector have
527 been set.
528
529 3. Related to (2) above: when there are more back references in a pattern than
530 slots in the output vector, pcre_exec() uses temporary memory during
531 matching, and copies in the captures as far as possible afterwards. It was
532 using the entire output vector, but this conflicts with the specification
533 that only 2/3 is used for passing back captured substrings. Now it uses
534 only the first 2/3, for compatibility. This is, of course, another edge
535 case.
536
537 4. Zoltan Herczeg's just-in-time compiler support has been integrated into the
538 main code base, and can be used by building with --enable-jit. When this is
539 done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
540 runtime --no-jit option is given.
541
542 5. When the number of matches in a pcre_dfa_exec() run exactly filled the
543 ovector, the return from the function was zero, implying that there were
544 other matches that did not fit. The correct "exactly full" value is now
545 returned.
546
547 6. If a subpattern that was called recursively or as a subroutine contained
548 (*PRUNE) or any other control that caused it to give a non-standard return,
549 invalid errors such as "Error -26 (nested recursion at the same subject
550 position)" or even infinite loops could occur.
551
552 7. If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
553 computing the minimum length on reaching *ACCEPT, and so ended up with the
554 wrong value of 1 rather than 0. Further investigation indicates that
555 computing a minimum subject length in the presence of *ACCEPT is difficult
556 (think back references, subroutine calls), and so I have changed the code
557 so that no minimum is registered for a pattern that contains *ACCEPT.
558
559 8. If (*THEN) was present in the first (true) branch of a conditional group,
560 it was not handled as intended. [But see 16 below.]
561
562 9. Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
563 Sheri Pierce.
564
565 10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
566 the first byte in a match must be "a".
567
568 11. Change 17 for 8.13 increased the recursion depth for patterns like
569 /a(?:.)*?a/ drastically. I've improved things by remembering whether a
570 pattern contains any instances of (*THEN). If it does not, the old
571 optimizations are restored. It would be nice to do this on a per-group
572 basis, but at the moment that is not feasible.
573
574 12. In some environments, the output of pcretest -C is CRLF terminated. This
575 broke RunTest's code that checks for the link size. A single white space
576 character after the value is now allowed for.
577
578 13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
579 For "fr", it uses the Windows-specific input and output files.
580
581 14. If (*THEN) appeared in a group that was called recursively or as a
582 subroutine, it did not work as intended. [But see next item.]
583
584 15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
585 pattern fragments (but not containing any | characters). If A and B are
586 matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
587 was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
588 D. In other words, Perl considers parentheses that do not contain any |
589 characters to be part of a surrounding alternative, whereas PCRE was
590 treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
591 differently. PCRE now behaves in the same way as Perl, except in the case
592 of subroutine/recursion calls such as (?1) which have in any case always
593 been different (but PCRE had them first :-).
594
595 16. Related to 15 above: Perl does not treat the | in a conditional group as
596 creating alternatives. Such a group is treated in the same way as an
597 ordinary group without any | characters when processing (*THEN). PCRE has
598 been changed to match Perl's behaviour.
599
600 17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
601 RunGrepTest script failed.
602
603 18. Change 22 for version 13 caused atomic groups to use more stack. This is
604 inevitable for groups that contain captures, but it can lead to a lot of
605 stack use in large patterns. The old behaviour has been restored for atomic
606 groups that do not contain any capturing parentheses.
607
608 19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
609 suppress the check for a minimum subject length at run time. (If it was
610 given to pcre_exec() or pcre_dfa_exec() it did work.)
611
612 20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
613 fail to work when decoding hex characters in data strings in EBCDIC
614 environments.
615
616 21. It appears that in at least one Mac OS environment, the isxdigit() function
617 is implemented as a macro that evaluates to its argument more than once,
618 contravening the C 90 Standard (I haven't checked a later standard). There
619 was an instance in pcretest which caused it to go wrong when processing
620 \x{...} escapes in subject strings. The has been rewritten to avoid using
621 things like p++ in the argument of isxdigit().
622
623
624 Version 8.13 16-Aug-2011
625 ------------------------
626
627 1. The Unicode data tables have been updated to Unicode 6.0.0.
628
629 2. Two minor typos in pcre_internal.h have been fixed.
630
631 3. Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
632 pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
633 in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
634
635 4. There were a number of related bugs in the code for matching backrefences
636 caselessly in UTF-8 mode when codes for the characters concerned were
637 different numbers of bytes. For example, U+023A and U+2C65 are an upper
638 and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
639 (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
640 code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
641 2-byte code at the end of the subject (it thought there wasn't enough data
642 left).
643
644 5. Comprehensive information about what went wrong is now returned by
645 pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
646 as the output vector has at least 2 elements. The offset of the start of
647 the failing character and a reason code are placed in the vector.
648
649 6. When the UTF-8 string check fails for pcre_compile(), the offset that is
650 now returned is for the first byte of the failing character, instead of the
651 last byte inspected. This is an incompatible change, but I hope it is small
652 enough not to be a problem. It makes the returned offset consistent with
653 pcre_exec() and pcre_dfa_exec().
654
655 7. pcretest now gives a text phrase as well as the error number when
656 pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
657 failure, the offset and reason code are output.
658
659 8. When \R was used with a maximizing quantifier it failed to skip backwards
660 over a \r\n pair if the subsequent match failed. Instead, it just skipped
661 back over a single character (\n). This seems wrong (because it treated the
662 two characters as a single entity when going forwards), conflicts with the
663 documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
664 behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
665 has been changed.
666
667 9. Some internal refactoring has changed the processing so that the handling
668 of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
669 time (the PCRE_DOTALL option was changed this way some time ago: version
670 7.7 change 16). This has made it possible to abolish the OP_OPT op code,
671 which was always a bit of a fudge. It also means that there is one less
672 argument for the match() function, which reduces its stack requirements
673 slightly. This change also fixes an incompatibility with Perl: the pattern
674 (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
675
676 10. More internal refactoring has drastically reduced the number of recursive
677 calls to match() for possessively repeated groups such as (abc)++ when
678 using pcre_exec().
679
680 11. While implementing 10, a number of bugs in the handling of groups were
681 discovered and fixed:
682
683 (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
684 (a|)*(?1) gave a compile-time internal error.
685 ((a|)+)+ did not notice that the outer group could match an empty string.
686 (^a|^)+ was not marked as anchored.
687 (.*a|.*)+ was not marked as matching at start or after a newline.
688
689 12. Yet more internal refactoring has removed another argument from the match()
690 function. Special calls to this function are now indicated by setting a
691 value in a variable in the "match data" data block.
692
693 13. Be more explicit in pcre_study() instead of relying on "default" for
694 opcodes that mean there is no starting character; this means that when new
695 ones are added and accidentally left out of pcre_study(), testing should
696 pick them up.
697
698 14. The -s option of pcretest has been documented for ages as being an old
699 synonym of -m (show memory usage). I have changed it to mean "force study
700 for every regex", that is, assume /S for every regex. This is similar to -i
701 and -d etc. It's slightly incompatible, but I'm hoping nobody is still
702 using it. It makes it easier to run collections of tests with and without
703 study enabled, and thereby test pcre_study() more easily. All the standard
704 tests are now run with and without -s (but some patterns can be marked as
705 "never study" - see 20 below).
706
707 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
708 restoration of the capturing data to the outer values was not happening
709 correctly.
710
711 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
712 empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
713 pattern had matched an empty string, and so incorrectly returned a no
714 match.
715
716 17. There was optimizing code for the last branch of non-capturing parentheses,
717 and also for the obeyed branch of a conditional subexpression, which used
718 tail recursion to cut down on stack usage. Unfortunately, now that there is
719 the possibility of (*THEN) occurring in these branches, tail recursion is
720 no longer possible because the return has to be checked for (*THEN). These
721 two optimizations have therefore been removed. [But see 8.20/11 above.]
722
723 18. If a pattern containing \R was studied, it was assumed that \R always
724 matched two bytes, thus causing the minimum subject length to be
725 incorrectly computed because \R can also match just one byte.
726
727 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
728 was incorrectly computed.
729
730 20. If /S is present twice on a test pattern in pcretest input, it now
731 *disables* studying, thereby overriding the use of -s on the command line
732 (see 14 above). This is necessary for one or two tests to keep the output
733 identical in both cases.
734
735 21. When (*ACCEPT) was used in an assertion that matched an empty string and
736 PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
737
738 22. When an atomic group that contained a capturing parenthesis was
739 successfully matched, but the branch in which it appeared failed, the
740 capturing was not being forgotten if a higher numbered group was later
741 captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
742 group 1 to "a", when in fact it should be unset. This applied to multi-
743 branched capturing and non-capturing groups, repeated or not, and also to
744 positive assertions (capturing in negative assertions does not happen
745 in PCRE) and also to nested atomic groups.
746
747 23. Add the ++ qualifier feature to pcretest, to show the remainder of the
748 subject after a captured substring, to make it easier to tell which of a
749 number of identical substrings has been captured.
750
751 24. The way atomic groups are processed by pcre_exec() has been changed so that
752 if they are repeated, backtracking one repetition now resets captured
753 values correctly. For example, if ((?>(a+)b)+aabab) is matched against
754 "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
755 "aaa". Previously, it would have been "a". As part of this code
756 refactoring, the way recursive calls are handled has also been changed.
757
758 25. If an assertion condition captured any substrings, they were not passed
759 back unless some other capturing happened later. For example, if
760 (?(?=(a))a) was matched against "a", no capturing was returned.
761
762 26. When studying a pattern that contained subroutine calls or assertions,
763 the code for finding the minimum length of a possible match was handling
764 direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
765 group 1 called group 2 while simultaneously a separate group 2 called group
766 1). A stack overflow occurred in this case. I have fixed this by limiting
767 the recursion depth to 10.
768
769 27. Updated RunTest.bat in the distribution to the version supplied by Tom
770 Fortmann. This supports explicit test numbers on the command line, and has
771 argument validation and error reporting.
772
773 28. An instance of \X with an unlimited repeat could fail if at any point the
774 first character it looked at was a mark character.
775
776 29. Some minor code refactoring concerning Unicode properties and scripts
777 should reduce the stack requirement of match() slightly.
778
779 30. Added the '=' option to pcretest to check the setting of unused capturing
780 slots at the end of the pattern, which are documented as being -1, but are
781 not included in the return count.
782
783 31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
784 compiled something random. Now it gives a compile-time error (as does
785 Perl).
786
787 32. A *MARK encountered during the processing of a positive assertion is now
788 recorded and passed back (compatible with Perl).
789
790 33. If --only-matching or --colour was set on a pcregrep call whose pattern
791 had alternative anchored branches, the search for a second match in a line
792 was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
793 matched the line "0102" twice. The same bug affected patterns that started
794 with a backwards assertion. For example /\b01|\b02/ also matched "0102"
795 twice.
796
797 34. Previously, PCRE did not allow quantification of assertions. However, Perl
798 does, and because of capturing effects, quantifying parenthesized
799 assertions may at times be useful. Quantifiers are now allowed for
800 parenthesized assertions.
801
802 35. A minor code tidy in pcre_compile() when checking options for \R usage.
803
804 36. \g was being checked for fancy things in a character class, when it should
805 just be a literal "g".
806
807 37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
808 appearance of a nested POSIX class supersedes an apparent external class.
809 For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
810 unescaped square brackets may also appear as part of class names. For
811 example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
812 more like Perl. (But see 8.20/1 above.)
813
814 38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
815 was because it thought it was \N{name}, which is not supported).
816
817 39. Add minix to OS list not supporting the -S option in pcretest.
818
819 40. PCRE tries to detect cases of infinite recursion at compile time, but it
820 cannot analyze patterns in sufficient detail to catch mutual recursions
821 such as ((?1))((?2)). There is now a runtime test that gives an error if a
822 subgroup is called recursively as a subpattern for a second time at the
823 same position in the subject string. In previous releases this might have
824 been caught by the recursion limit, or it might have run out of stack.
825
826 41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
827 happen only once. PCRE was, however incorrectly giving a compile time error
828 "recursive call could loop indefinitely" because it cannot analyze the
829 pattern in sufficient detail. The compile time test no longer happens when
830 PCRE is compiling a conditional subpattern, but actual runaway loops are
831 now caught at runtime (see 40 above).
832
833 42. It seems that Perl allows any characters other than a closing parenthesis
834 to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
835 has been changed to be the same.
836
837 43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
838 as not to get warnings when autogen.sh is called. Also changed
839 AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
840
841 44. To help people who use pcregrep to scan files containing exceedingly long
842 lines, the following changes have been made:
843
844 (a) The default value of the buffer size parameter has been increased from
845 8K to 20K. (The actual buffer used is three times this size.)
846
847 (b) The default can be changed by ./configure --with-pcregrep-bufsize when
848 PCRE is built.
849
850 (c) A --buffer-size=n option has been added to pcregrep, to allow the size
851 to be set at run time.
852
853 (d) Numerical values in pcregrep options can be followed by K or M, for
854 example --buffer-size=50K.
855
856 (e) If a line being scanned overflows pcregrep's buffer, an error is now
857 given and the return code is set to 2.
858
859 45. Add a pointer to the latest mark to the callout data block.
860
861 46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
862 partial match of an empty string instead of no match. This was specific to
863 the use of ".".
864
865 47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
866 complete match instead of a partial match. This bug was dependent on both
867 the PCRE_UTF8 and PCRE_DOTALL options being set.
868
869 48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
870 starting byte set, because \b was not being ignored.
871
872
873 Version 8.12 15-Jan-2011
874 ------------------------
875
876 1. Fixed some typos in the markup of the man pages, and wrote a script that
877 checks for such things as part of the documentation building process.
878
879 2. On a big-endian 64-bit system, pcregrep did not correctly process the
880 --match-limit and --recursion-limit options (added for 8.11). In
881 particular, this made one of the standard tests fail. (The integer value
882 went into the wrong half of a long int.)
883
884 3. If the --colour option was given to pcregrep with -v (invert match), it
885 did strange things, either producing crazy output, or crashing. It should,
886 of course, ignore a request for colour when reporting lines that do not
887 match.
888
889 4. Another pcregrep bug caused similar problems if --colour was specified with
890 -M (multiline) and the pattern match finished with a line ending.
891
892 5. In pcregrep, when a pattern that ended with a literal newline sequence was
893 matched in multiline mode, the following line was shown as part of the
894 match. This seems wrong, so I have changed it.
895
896 6. Another pcregrep bug in multiline mode, when --colour was specified, caused
897 the check for further matches in the same line (so they could be coloured)
898 to overrun the end of the current line. If another match was found, it was
899 incorrectly shown (and then shown again when found in the next line).
900
901 7. If pcregrep was compiled under Windows, there was a reference to the
902 function pcregrep_exit() before it was defined. I am assuming this was
903 the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
904 reported by a user. I've moved the definition above the reference.
905
906
907 Version 8.11 10-Dec-2010
908 ------------------------
909
910 1. (*THEN) was not working properly if there were untried alternatives prior
911 to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
912 backtracked to try for "b" instead of moving to the next alternative branch
913 at the same level (in this case, to look for "c"). The Perl documentation
914 is clear that when (*THEN) is backtracked onto, it goes to the "next
915 alternative in the innermost enclosing group".
916
917 2. (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
918 such as (A(*COMMIT)B(*THEN)C|D) any failure after matching A should
919 result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
920 (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
921 (*THEN).
922
923 3. If \s appeared in a character class, it removed the VT character from
924 the class, even if it had been included by some previous item, for example
925 in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
926 of \s, but is part of the POSIX "space" class.)
927
928 4. A partial match never returns an empty string (because you can always
929 match an empty string at the end of the subject); however the checking for
930 an empty string was starting at the "start of match" point. This has been
931 changed to the "earliest inspected character" point, because the returned
932 data for a partial match starts at this character. This means that, for
933 example, /(?<=abc)def/ gives a partial match for the subject "abc"
934 (previously it gave "no match").
935
936 5. Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
937 of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
938 previously a full match would be given. However, setting PCRE_PARTIAL_HARD
939 has an implication that the given string is incomplete (because a partial
940 match is preferred over a full match). For this reason, these items now
941 give a partial match in this situation. [Aside: previously, the one case
942 /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
943 match rather than a full match, which was wrong by the old rules, but is
944 now correct.]
945
946 6. There was a bug in the handling of #-introduced comments, recognized when
947 PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
948 If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
949 UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
950 scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
951 but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
952 places in pcre_compile().
953
954 7. Related to (6) above, when pcre_compile() was skipping #-introduced
955 comments when looking ahead for named forward references to subpatterns,
956 the only newline sequence it recognized was NL. It now handles newlines
957 according to the set newline convention.
958
959 8. SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
960 former, but used strtoul(), whereas pcretest avoided strtoul() but did not
961 cater for a lack of strerror(). These oversights have been fixed.
962
963 9. Added --match-limit and --recursion-limit to pcregrep.
964
965 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
966
967 11. When the -o option was used, pcregrep was setting a return code of 1, even
968 when matches were found, and --line-buffered was not being honoured.
969
970 12. Added an optional parentheses number to the -o and --only-matching options
971 of pcregrep.
972
973 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
974 can match an empty string. The code to do it in pcretest and pcredemo
975 needed fixing:
976
977 (a) When the newline convention was "crlf", pcretest got it wrong, skipping
978 only one byte after an empty string match just before CRLF (this case
979 just got forgotten; "any" and "anycrlf" were OK).
980
981 (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
982 mode when an empty string match preceded an ASCII character followed by
983 a non-ASCII character. (The code for advancing by one character rather
984 than one byte was nonsense.)
985
986 (c) The pcredemo.c sample program did not have any code at all to handle
987 the cases when CRLF is a valid newline sequence.
988
989 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
990 as a starting offset was within the subject string. There is now a new
991 error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
992 negative or greater than the length of the string. In order to test this,
993 pcretest is extended to allow the setting of negative starting offsets.
994
995 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
996 starting offset points to the beginning of a UTF-8 character was
997 unnecessarily clumsy. I tidied it up.
998
999 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
1000 bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
1001
1002 17. Nobody had reported that the --include_dir option, which was added in
1003 release 7.7 should have been called --include-dir (hyphen, not underscore)
1004 for compatibility with GNU grep. I have changed it to --include-dir, but
1005 left --include_dir as an undocumented synonym, and the same for
1006 --exclude-dir, though that is not available in GNU grep, at least as of
1007 release 2.5.4.
1008
1009 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
1010 characters from a string of bytes) have been redefined so as not to use
1011 loops, in order to improve performance in some environments. At the same
1012 time, I abstracted some of the common code into auxiliary macros to save
1013 repetition (this should not affect the compiled code).
1014
1015 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
1016 compile-time error is now given if \c is not followed by an ASCII
1017 character, that is, a byte less than 128. (In EBCDIC mode, the code is
1018 different, and any byte value is allowed.)
1019
1020 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
1021 START_OPTIMIZE option, which is now allowed at compile time - but just
1022 passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
1023 to pcregrep and other applications that have no direct access to PCRE
1024 options. The new /Y option in pcretest sets this option when calling
1025 pcre_compile().
1026
1027 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
1028 back references. Groups containing recursive back references were forced to
1029 be atomic by that change, but in the case of named groups, the amount of
1030 memory required was incorrectly computed, leading to "Failed: internal
1031 error: code overflow". This has been fixed.
1032
1033 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
1034 pcretest.c, to avoid build problems in some Borland environments.
1035
1036
1037 Version 8.10 25-Jun-2010
1038 ------------------------
1039
1040 1. Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
1041 THEN.
1042
1043 2. (*ACCEPT) was not working when inside an atomic group.
1044
1045 3. Inside a character class, \B is treated as a literal by default, but
1046 faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
1047 causes the error). The code is unchanged, but I tidied the documentation.
1048
1049 4. Inside a character class, PCRE always treated \R and \X as literals,
1050 whereas Perl faults them if its -w option is set. I have changed PCRE so
1051 that it faults them when PCRE_EXTRA is set.
1052
1053 5. Added support for \N, which always matches any character other than
1054 newline. (It is the same as "." when PCRE_DOTALL is not set.)
1055
1056 6. When compiling pcregrep with newer versions of gcc which may have
1057 FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
1058 declared with attribute warn_unused_result" were given. Just casting the
1059 result to (void) does not stop the warnings; a more elaborate fudge is
1060 needed. I've used a macro to implement this.
1061
1062 7. Minor change to pcretest.c to avoid a compiler warning.
1063
1064 8. Added four artifical Unicode properties to help with an option to make
1065 \s etc use properties (see next item). The new properties are: Xan
1066 (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
1067
1068 9. Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
1069 use Unicode properties. (*UCP) at the start of a pattern can be used to set
1070 this option. Modified pcretest to add /W to test this facility. Added
1071 REG_UCP to make it available via the POSIX interface.
1072
1073 10. Added --line-buffered to pcregrep.
1074
1075 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
1076 studied, and the match started with a letter with a code point greater than
1077 127 whose first byte was different to the first byte of the other case of
1078 the letter, the other case of this starting letter was not recognized
1079 (#976).
1080
1081 12. If a pattern that was studied started with a repeated Unicode property
1082 test, for example, \p{Nd}+, there was the theoretical possibility of
1083 setting up an incorrect bitmap of starting bytes, but fortunately it could
1084 not have actually happened in practice until change 8 above was made (it
1085 added property types that matched character-matching opcodes).
1086
1087 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
1088 possible starting bytes for non-anchored patterns.
1089
1090 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
1091 \R, and also a number of cases that involve Unicode properties, both
1092 explicit and implicit when PCRE_UCP is set.
1093
1094 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
1095 input, it could crash or give wrong results if characters with values
1096 greater than 0xc0 were present in the subject string. (Detail: it assumed
1097 UTF-8 input when processing these items.)
1098
1099 16. Added a lot of (int) casts to avoid compiler warnings in systems where
1100 size_t is 64-bit (#991).
1101
1102 17. Added a check for running out of memory when PCRE is compiled with
1103 --disable-stack-for-recursion (#990).
1104
1105 18. If the last data line in a file for pcretest does not have a newline on
1106 the end, a newline was missing in the output.
1107
1108 19. The default pcre_chartables.c file recognizes only ASCII characters (values
1109 less than 128) in its various bitmaps. However, there is a facility for
1110 generating tables according to the current locale when PCRE is compiled. It
1111 turns out that in some environments, 0x85 and 0xa0, which are Unicode space
1112 characters, are recognized by isspace() and therefore were getting set in
1113 these tables, and indeed these tables seem to approximate to ISO 8859. This
1114 caused a problem in UTF-8 mode when pcre_study() was used to create a list
1115 of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
1116 which of course cannot start UTF-8 characters. I have changed the code so
1117 that only real ASCII characters (less than 128) and the correct starting
1118 bytes for UTF-8 encodings are set for characters greater than 127 when in
1119 UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
1120 altogether.)
1121
1122 20. Added the /T option to pcretest so as to be able to run tests with non-
1123 standard character tables, thus making it possible to include the tests
1124 used for 19 above in the standard set of tests.
1125
1126 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
1127 reference to a subpattern the other side of a comment that contains an
1128 opening parenthesis caused either an internal compiling error, or a
1129 reference to the wrong subpattern.
1130
1131
1132 Version 8.02 19-Mar-2010
1133 ------------------------
1134
1135 1. The Unicode data tables have been updated to Unicode 5.2.0.
1136
1137 2. Added the option --libs-cpp to pcre-config, but only when C++ support is
1138 configured.
1139
1140 3. Updated the licensing terms in the pcregexp.pas file, as agreed with the
1141 original author of that file, following a query about its status.
1142
1143 4. On systems that do not have stdint.h (e.g. Solaris), check for and include
1144 inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
1145
1146 5. A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
1147 quantifier applied to a forward-referencing subroutine call, could compile
1148 incorrect code or give the error "internal error: previously-checked
1149 referenced subpattern not found".
1150
1151 6. Both MS Visual Studio and Symbian OS have problems with initializing
1152 variables to point to external functions. For these systems, therefore,
1153 pcre_malloc etc. are now initialized to local functions that call the
1154 relevant global functions.
1155
1156 7. There were two entries missing in the vectors called coptable and poptable
1157 in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
1158 I've fixed the data, and added a kludgy way of testing at compile time that
1159 the lengths are correct (equal to the number of opcodes).
1160
1161 8. Following on from 7, I added a similar kludge to check the length of the
1162 eint vector in pcreposix.c.
1163
1164 9. Error texts for pcre_compile() are held as one long string to avoid too
1165 much relocation at load time. To find a text, the string is searched,
1166 counting zeros. There was no check for running off the end of the string,
1167 which could happen if a new error number was added without updating the
1168 string.
1169
1170 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
1171
1172 11. \K was not working if it appeared in an atomic group or in a group that
1173 was called as a "subroutine", or in an assertion. Perl 5.11 documents that
1174 \K is "not well defined" if used in an assertion. PCRE now accepts it if
1175 the assertion is positive, but not if it is negative.
1176
1177 12. Change 11 fortuitously reduced the size of the stack frame used in the
1178 "match()" function of pcre_exec.c by one pointer. Forthcoming
1179 implementation of support for (*MARK) will need an extra pointer on the
1180 stack; I have reserved it now, so that the stack frame size does not
1181 decrease.
1182
1183 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
1184 item in branch that calls a recursion is a subroutine call - as in the
1185 second branch in the above example - was incorrectly given the compile-
1186 time error "recursive call could loop indefinitely" because pcre_compile()
1187 was not correctly checking the subroutine for matching a non-empty string.
1188
1189 14. The checks for overrunning compiling workspace could trigger after an
1190 overrun had occurred. This is a "should never occur" error, but it can be
1191 triggered by pathological patterns such as hundreds of nested parentheses.
1192 The checks now trigger 100 bytes before the end of the workspace.
1193
1194 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
1195
1196
1197 Version 8.01 19-Jan-2010
1198 ------------------------
1199
1200 1. If a pattern contained a conditional subpattern with only one branch (in
1201 particular, this includes all (*DEFINE) patterns), a call to pcre_study()
1202 computed the wrong minimum data length (which is of course zero for such
1203 subpatterns). This could cause incorrect "no match" results.
1204
1205 2. For patterns such as (?i)a(?-i)b|c where an option setting at the start of
1206 the pattern is reset in the first branch, pcre_compile() failed with
1207 "internal error: code overflow at offset...". This happened only when
1208 the reset was to the original external option setting. (An optimization
1209 abstracts leading options settings into an external setting, which was the
1210 cause of this.)
1211
1212 3. A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
1213 of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
1214 assertion pattern did not match (meaning that the assertion was true), it
1215 was incorrectly treated as false if the SKIP had been reached during the
1216 matching. This also applied to assertions used as conditions.
1217
1218 4. If an item that is not supported by pcre_dfa_exec() was encountered in an
1219 assertion subpattern, including such a pattern used as a condition,
1220 unpredictable results occurred, instead of the error return
1221 PCRE_ERROR_DFA_UITEM.
1222
1223 5. The C++ GlobalReplace function was not working like Perl for the special
1224 situation when an empty string is matched. It now does the fancy magic
1225 stuff that is necessary.
1226
1227 6. In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
1228 removed. (These were left over from very, very early versions of PCRE.)
1229
1230 7. Some cosmetic changes to the code to make life easier when compiling it
1231 as part of something else:
1232
1233 (a) Change DEBUG to PCRE_DEBUG.
1234
1235 (b) In pcre_compile(), rename the member of the "branch_chain" structure
1236 called "current" as "current_branch", to prevent a collision with the
1237 Linux macro when compiled as a kernel module.
1238
1239 (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
1240 prevent a collision with the Linux macro when compiled as a kernel
1241 module.
1242
1243 8. In pcre_compile() there are some checks for integer overflows that used to
1244 cast potentially large values to (double). This has been changed to that
1245 when building, a check for int64_t is made, and if it is found, it is used
1246 instead, thus avoiding the use of floating point arithmetic. (There is no
1247 other use of FP in PCRE.) If int64_t is not found, the fallback is to
1248 double.
1249
1250 9. Added two casts to avoid signed/unsigned warnings from VS Studio Express
1251 2005 (difference between two addresses compared to an unsigned value).
1252
1253 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
1254 custom one, because of the following reported problem in Windows:
1255
1256 - libbz2 uses the Pascal calling convention (WINAPI) for the functions
1257 under Win32.
1258 - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
1259 therefore missing the function definition.
1260 - The compiler thus generates a "C" signature for the test function.
1261 - The linker fails to find the "C" function.
1262 - PCRE fails to configure if asked to do so against libbz2.
1263
1264 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
1265 messages were output:
1266
1267 Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
1268 rerunning libtoolize, to keep the correct libtool macros in-tree.
1269 Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
1270
1271 I have done both of these things.
1272
1273 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
1274 most of the time, it *can* run out if it is given a pattern that contains a
1275 runaway infinite recursion. I updated the discussion in the pcrestack man
1276 page.
1277
1278 13. Now that we have gone to the x.xx style of version numbers, the minor
1279 version may start with zero. Using 08 or 09 is a bad idea because users
1280 might check the value of PCRE_MINOR in their code, and 08 or 09 may be
1281 interpreted as invalid octal numbers. I've updated the previous comment in
1282 configure.ac, and also added a check that gives an error if 08 or 09 are
1283 used.
1284
1285 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
1286 causing partial matching to fail when the end of the subject matched \W
1287 in a UTF-8 pattern where \W was quantified with a minimum of 3.
1288
1289 15. There were some discrepancies between the declarations in pcre_internal.h
1290 of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
1291 their definitions. The declarations used "const uschar *" and the
1292 definitions used USPTR. Even though USPTR is normally defined as "const
1293 unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
1294 reported that: "This difference in casting confuses some C++ compilers, for
1295 example, SunCC recognizes above declarations as different functions and
1296 generates broken code for hbpcre." I have changed the declarations to use
1297 USPTR.
1298
1299 16. GNU libtool is named differently on some systems. The autogen.sh script now
1300 tries several variants such as glibtoolize (MacOSX) and libtoolize1x
1301 (FreeBSD).
1302
1303 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
1304 (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
1305 comment: "Figure out how to create a longlong from a string: strtoll and
1306 equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
1307 instance, but it only takes 2 args instead of 3!"
1308
1309 18. A subtle bug concerned with back references has been fixed by a change of
1310 specification, with a corresponding code fix. A pattern such as
1311 ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
1312 refers, was giving matches when it shouldn't. For example, xa=xaaa would
1313 match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
1314 same bug. Such groups have to be quantified to be useful, or contained
1315 inside another quantified group. (If there's no repetition, the reference
1316 can never match.) The problem arises because, having left the group and
1317 moved on to the rest of the pattern, a later failure that backtracks into
1318 the group uses the captured value from the final iteration of the group
1319 rather than the correct earlier one. I have fixed this in PCRE by forcing
1320 any group that contains a reference to itself to be an atomic group; that
1321 is, there cannot be any backtracking into it once it has completed. This is
1322 similar to recursive and subroutine calls.
1323
1324
1325 Version 8.00 19-Oct-09
1326 ----------------------
1327
1328 1. The table for translating pcre_compile() error codes into POSIX error codes
1329 was out-of-date, and there was no check on the pcre_compile() error code
1330 being within the table. This could lead to an OK return being given in
1331 error.
1332
1333 2. Changed the call to open a subject file in pcregrep from fopen(pathname,
1334 "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
1335 in a Windows environment.
1336
1337 3. The pcregrep --count option prints the count for each file even when it is
1338 zero, as does GNU grep. However, pcregrep was also printing all files when
1339 --files-with-matches was added. Now, when both options are given, it prints
1340 counts only for those files that have at least one match. (GNU grep just
1341 prints the file name in this circumstance, but including the count seems
1342 more useful - otherwise, why use --count?) Also ensured that the
1343 combination -clh just lists non-zero counts, with no names.
1344
1345 4. The long form of the pcregrep -F option was incorrectly implemented as
1346 --fixed_strings instead of --fixed-strings. This is an incompatible change,
1347 but it seems right to fix it, and I didn't think it was worth preserving
1348 the old behaviour.
1349
1350 5. The command line items --regex=pattern and --regexp=pattern were not
1351 recognized by pcregrep, which required --regex pattern or --regexp pattern
1352 (with a space rather than an '='). The man page documented the '=' forms,
1353 which are compatible with GNU grep; these now work.
1354
1355 6. No libpcreposix.pc file was created for pkg-config; there was just
1356 libpcre.pc and libpcrecpp.pc. The omission has been rectified.
1357
1358 7. Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
1359 when UCP support is not needed, by modifying the Python script that
1360 generates it from Unicode data files. This should not matter if the module
1361 is correctly used as a library, but I received one complaint about 50K of
1362 unwanted data. My guess is that the person linked everything into his
1363 program rather than using a library. Anyway, it does no harm.
1364
1365 8. A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
1366 was a minimum greater than 1 for a wide character in a possessive
1367 repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
1368 which had an unlimited repeat of a nested, fixed maximum repeat of a wide
1369 character. Chaos in the form of incorrect output or a compiling loop could
1370 result.
1371
1372 9. The restrictions on what a pattern can contain when partial matching is
1373 requested for pcre_exec() have been removed. All patterns can now be
1374 partially matched by this function. In addition, if there are at least two
1375 slots in the offset vector, the offset of the earliest inspected character
1376 for the match and the offset of the end of the subject are set in them when
1377 PCRE_ERROR_PARTIAL is returned.
1378
1379 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
1380 synonymous with PCRE_PARTIAL, for backwards compatibility, and
1381 PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
1382 and may be more useful for multi-segment matching.
1383
1384 11. Partial matching with pcre_exec() is now more intuitive. A partial match
1385 used to be given if ever the end of the subject was reached; now it is
1386 given only if matching could not proceed because another character was
1387 needed. This makes a difference in some odd cases such as Z(*FAIL) with the
1388 string "Z", which now yields "no match" instead of "partial match". In the
1389 case of pcre_dfa_exec(), "no match" is given if every matching path for the
1390 final character ended with (*FAIL).
1391
1392 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
1393 if the pattern had a "must contain" character that was already found in the
1394 earlier partial match, unless partial matching was again requested. For
1395 example, with the pattern /dog.(body)?/, the "must contain" character is
1396 "g". If the first part-match was for the string "dog", restarting with
1397 "sbody" failed. This bug has been fixed.
1398
1399 13. The string returned by pcre_dfa_exec() after a partial match has been
1400 changed so that it starts at the first inspected character rather than the
1401 first character of the match. This makes a difference only if the pattern
1402 starts with a lookbehind assertion or \b or \B (\K is not supported by
1403 pcre_dfa_exec()). It's an incompatible change, but it makes the two
1404 matching functions compatible, and I think it's the right thing to do.
1405
1406 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
1407 so that the demonstration program is easily available in environments where
1408 PCRE has not been installed from source.
1409
1410 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
1411 libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
1412 library.
1413
1414 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
1415 It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
1416 is not the first non-POSIX option to be added. Clearly some people find
1417 these options useful.
1418
1419 17. If a caller to the POSIX matching function regexec() passes a non-zero
1420 value for nmatch with a NULL value for pmatch, the value of
1421 nmatch is forced to zero.
1422
1423 18. RunGrepTest did not have a test for the availability of the -u option of
1424 the diff command, as RunTest does. It now checks in the same way as
1425 RunTest, and also checks for the -b option.
1426
1427 19. If an odd number of negated classes containing just a single character
1428 interposed, within parentheses, between a forward reference to a named
1429 subpattern and the definition of the subpattern, compilation crashed with
1430 an internal error, complaining that it could not find the referenced
1431 subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
1432 [The bug was that it was starting one character too far in when skipping
1433 over the character class, thus treating the ] as data rather than
1434 terminating the class. This meant it could skip too much.]
1435
1436 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
1437 /g option in pcretest when the pattern contains \K, which makes it possible
1438 to have an empty string match not at the start, even when the pattern is
1439 anchored. Updated pcretest and pcredemo to use this option.
1440
1441 21. If the maximum number of capturing subpatterns in a recursion was greater
1442 than the maximum at the outer level, the higher number was returned, but
1443 with unset values at the outer level. The correct (outer level) value is
1444 now given.
1445
1446 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
1447 PCRE did not set those parentheses (unlike Perl). I have now found a way to
1448 make it do so. The string so far is captured, making this feature
1449 compatible with Perl.
1450
1451 23. The tests have been re-organized, adding tests 11 and 12, to make it
1452 possible to check the Perl 5.10 features against Perl 5.10.
1453
1454 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
1455 pattern matches a fixed length string. PCRE did not allow this; now it
1456 does. Neither allows recursion.
1457
1458 25. I finally figured out how to implement a request to provide the minimum
1459 length of subject string that was needed in order to match a given pattern.
1460 (It was back references and recursion that I had previously got hung up
1461 on.) This code has now been added to pcre_study(); it finds a lower bound
1462 to the length of subject needed. It is not necessarily the greatest lower
1463 bound, but using it to avoid searching strings that are too short does give
1464 some useful speed-ups. The value is available to calling programs via
1465 pcre_fullinfo().
1466
1467 26. While implementing 25, I discovered to my embarrassment that pcretest had
1468 not been passing the result of pcre_study() to pcre_dfa_exec(), so the
1469 study optimizations had never been tested with that matching function.
1470 Oops. What is worse, even when it was passed study data, there was a bug in
1471 pcre_dfa_exec() that meant it never actually used it. Double oops. There
1472 were also very few tests of studied patterns with pcre_dfa_exec().
1473
1474 27. If (?| is used to create subpatterns with duplicate numbers, they are now
1475 allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
1476 on the other side of the coin, they are no longer allowed to have different
1477 names, because these cannot be distinguished in PCRE, and this has caused
1478 confusion. (This is a difference from Perl.)
1479
1480 28. When duplicate subpattern names are present (necessarily with different
1481 numbers, as required by 27 above), and a test is made by name in a
1482 conditional pattern, either for a subpattern having been matched, or for
1483 recursion in such a pattern, all the associated numbered subpatterns are
1484 tested, and the overall condition is true if the condition is true for any
1485 one of them. This is the way Perl works, and is also more like the way
1486 testing by number works.
1487
1488
1489 Version 7.9 11-Apr-09
1490 ---------------------
1491
1492 1. When building with support for bzlib/zlib (pcregrep) and/or readline
1493 (pcretest), all targets were linked against these libraries. This included
1494 libpcre, libpcreposix, and libpcrecpp, even though they do not use these
1495 libraries. This caused unwanted dependencies to be created. This problem
1496 has been fixed, and now only pcregrep is linked with bzlib/zlib and only
1497 pcretest is linked with readline.
1498
1499 2. The "typedef int BOOL" in pcre_internal.h that was included inside the
1500 "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
1501 moved outside it again, because FALSE and TRUE are already defined in AIX,
1502 but BOOL is not.
1503
1504 3. The pcre_config() function was treating the PCRE_MATCH_LIMIT and
1505 PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
1506
1507 4. The pcregrep documentation said spaces were inserted as well as colons (or
1508 hyphens) following file names and line numbers when outputting matching
1509 lines. This is not true; no spaces are inserted. I have also clarified the
1510 wording for the --colour (or --color) option.
1511
1512 5. In pcregrep, when --colour was used with -o, the list of matching strings
1513 was not coloured; this is different to GNU grep, so I have changed it to be
1514 the same.
1515
1516 6. When --colo(u)r was used in pcregrep, only the first matching substring in
1517 each matching line was coloured. Now it goes on to look for further matches
1518 of any of the test patterns, which is the same behaviour as GNU grep.
1519
1520 7. A pattern that could match an empty string could cause pcregrep to loop; it
1521 doesn't make sense to accept an empty string match in pcregrep, so I have
1522 locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
1523 seems to be how GNU grep behaves.
1524
1525 8. The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
1526 start or after a newline", because the conditional assertion was not being
1527 correctly handled. The rule now is that both the assertion and what follows
1528 in the first alternative must satisfy the test.
1529
1530 9. If auto-callout was enabled in a pattern with a conditional group whose
1531 condition was an assertion, PCRE could crash during matching, both with
1532 pcre_exec() and pcre_dfa_exec().
1533
1534 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
1535 used for matching.
1536
1537 11. Unicode property support in character classes was not working for
1538 characters (bytes) greater than 127 when not in UTF-8 mode.
1539
1540 12. Added the -M command line option to pcretest.
1541
1542 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
1543
1544 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
1545
1546 16. Added comments and documentation about mis-use of no_arg in the C++
1547 wrapper.
1548
1549 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
1550 from Martin Jerabek that uses macro names for all relevant character and
1551 string constants.
1552
1553 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
1554 SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
1555 SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
1556 these, but not everybody uses configure.
1557
1558 19. A conditional group that had only one branch was not being correctly
1559 recognized as an item that could match an empty string. This meant that an
1560 enclosing group might also not be so recognized, causing infinite looping
1561 (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
1562 with the subject "ab", where knowledge that the repeated group can match
1563 nothing is needed in order to break the loop.
1564
1565 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
1566 exec(), but without supplying a callout function, matching went wrong.
1567
1568 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
1569 leak if the size of the offset vector was greater than 30. When the vector
1570 is smaller, the saved offsets during recursion go onto a local stack
1571 vector, but for larger vectors malloc() is used. It was failing to free
1572 when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
1573 error, in fact).
1574
1575 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
1576 heapframe that is used only when UTF-8 support is enabled. This caused no
1577 problem, but was untidy.
1578
1579 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
1580 CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
1581 included within another project.
1582
1583 24. Steven Van Ingelgem's patches to add more options to the CMake support,
1584 slightly modified by me:
1585
1586 (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
1587 not building pcregrep.
1588
1589 (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
1590 if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
1591
1592 25. Forward references, both numeric and by name, in patterns that made use of
1593 duplicate group numbers, could behave incorrectly or give incorrect errors,
1594 because when scanning forward to find the reference group, PCRE was not
1595 taking into account the duplicate group numbers. A pattern such as
1596 ^X(?3)(a)(?|(b)|(q))(Y) is an example.
1597
1598 26. Changed a few more instances of "const unsigned char *" to USPTR, making
1599 the feature of a custom pointer more persuasive (as requested by a user).
1600
1601 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
1602 pcretest.c, inside #ifndefs, because it seems they are sometimes already
1603 pre-defined.
1604
1605 28. Added support for (*UTF8) at the start of a pattern.
1606
1607 29. Arrange for flags added by the "release type" setting in CMake to be shown
1608 in the configuration summary.
1609
1610
1611 Version 7.8 05-Sep-08
1612 ---------------------
1613
1614 1. Replaced UCP searching code with optimized version as implemented for Ad
1615 Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
1616 stage table and inline lookup instead of a function, giving speed ups of 2
1617 to 5 times on some simple patterns that I tested. Permission was given to
1618 distribute the MultiStage2.py script that generates the tables (it's not in
1619 the tarball, but is in the Subversion repository).
1620
1621 2. Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
1622 scripts.
1623
1624 3. Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
1625 a group with a zero qualifier. The result of the study could be incorrect,
1626 or the function might crash, depending on the pattern.
1627
1628 4. Caseless matching was not working for non-ASCII characters in back
1629 references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
1630 It now works when Unicode Property Support is available.
1631
1632 5. In pcretest, an escape such as \x{de} in the data was always generating
1633 a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
1634 non-UTF-8 mode. If the value is greater than 255, it gives a warning about
1635 truncation.
1636
1637 6. Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
1638
1639 7. Added two (int) casts to pcregrep when printing the difference of two
1640 pointers, in case they are 64-bit values.
1641
1642 8. Added comments about Mac OS X stack usage to the pcrestack man page and to
1643 test 2 if it fails.
1644
1645 9. Added PCRE_CALL_CONVENTION just before the names of all exported functions,
1646 and a #define of that name to empty if it is not externally set. This is to
1647 allow users of MSVC to set it if necessary.
1648
1649 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
1650 the convenience functions in the pcre_get.c source file.
1651
1652 11. An option change at the start of a pattern that had top-level alternatives
1653 could cause overwriting and/or a crash. This command provoked a crash in
1654 some environments:
1655
1656 printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
1657
1658 This potential security problem was recorded as CVE-2008-2371.
1659
1660 12. For a pattern where the match had to start at the beginning or immediately
1661 after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
1662 pcre_dfa_exec() could read past the end of the passed subject if there was
1663 no match. To help with detecting such bugs (e.g. with valgrind), I modified
1664 pcretest so that it places the subject at the end of its malloc-ed buffer.
1665
1666 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
1667 exec() might read past the end of the data buffer in UTF-8 mode.
1668
1669 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
1670 the data contained the byte 0x85 as part of a UTF-8 character within its
1671 first line. This applied both to normal and DFA matching.
1672
1673 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
1674 /^[^d]*?$/8 failed to match "abc".
1675
1676 16. Added a missing copyright notice to pcrecpp_internal.h.
1677
1678 17. Make it more clear in the documentation that values returned from
1679 pcre_exec() in ovector are byte offsets, not character counts.
1680
1681 18. Tidied a few places to stop certain compilers from issuing warnings.
1682
1683 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
1684 supplied by Stefan Weber. I made a further small update for 7.8 because
1685 there is a change of source arrangements: the pcre_searchfuncs.c module is
1686 replaced by pcre_ucd.c.
1687
1688
1689 Version 7.7 07-May-08
1690 ---------------------
1691
1692 1. Applied Craig's patch to sort out a long long problem: "If we can't convert
1693 a string to a long long, pretend we don't even have a long long." This is
1694 done by checking for the strtoq, strtoll, and _strtoi64 functions.
1695
1696 2. Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
1697 pre-7.6 versions, which defined a global no_arg variable instead of putting
1698 it in the RE class. (See also #8 below.)
1699
1700 3. Remove a line of dead code, identified by coverity and reported by Nuno
1701 Lopes.
1702
1703 4. Fixed two related pcregrep bugs involving -r with --include or --exclude:
1704
1705 (1) The include/exclude patterns were being applied to the whole pathnames
1706 of files, instead of just to the final components.
1707
1708 (2) If there was more than one level of directory, the subdirectories were
1709 skipped unless they satisfied the include/exclude conditions. This is
1710 inconsistent with GNU grep (and could even be seen as contrary to the
1711 pcregrep specification - which I improved to make it absolutely clear).
1712 The action now is always to scan all levels of directory, and just
1713 apply the include/exclude patterns to regular files.
1714
1715 5. Added the --include_dir and --exclude_dir patterns to pcregrep, and used
1716 --exclude_dir in the tests to avoid scanning .svn directories.
1717
1718 6. Applied Craig's patch to the QuoteMeta function so that it escapes the
1719 NUL character as backslash + 0 rather than backslash + NUL, because PCRE
1720 doesn't support NULs in patterns.
1721
1722 7. Added some missing "const"s to declarations of static tables in
1723 pcre_compile.c and pcre_dfa_exec.c.
1724
1725 8. Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
1726 caused by fix #2 above. (Subsequently also a second patch to fix the
1727 first patch. And a third patch - this was a messy problem.)
1728
1729 9. Applied Craig's patch to remove the use of push_back().
1730
1731 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
1732 matching function regexec().
1733
1734 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
1735 which, however, unlike Perl's \g{...}, are subroutine calls, not back
1736 references. PCRE supports relative numbers with this syntax (I don't think
1737 Oniguruma does).
1738
1739 12. Previously, a group with a zero repeat such as (...){0} was completely
1740 omitted from the compiled regex. However, this means that if the group
1741 was called as a subroutine from elsewhere in the pattern, things went wrong
1742 (an internal error was given). Such groups are now left in the compiled
1743 pattern, with a new opcode that causes them to be skipped at execution
1744 time.
1745
1746 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
1747 to the way PCRE behaves:
1748
1749 (a) A lone ] character is dis-allowed (Perl treats it as data).
1750
1751 (b) A back reference to an unmatched subpattern matches an empty string
1752 (Perl fails the current match path).
1753
1754 (c) A data ] in a character class must be notated as \] because if the
1755 first data character in a class is ], it defines an empty class. (In
1756 Perl it is not possible to have an empty class.) The empty class []
1757 never matches; it forces failure and is equivalent to (*FAIL) or (?!).
1758 The negative empty class [^] matches any one character, independently
1759 of the DOTALL setting.
1760
1761 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
1762 non-existent subpattern following a character class starting with ']' and
1763 containing () gave an internal compiling error instead of "reference to
1764 non-existent subpattern". Fortunately, when the pattern did exist, the
1765 compiled code was correct. (When scanning forwards to check for the
1766 existencd of the subpattern, it was treating the data ']' as terminating
1767 the class, so got the count wrong. When actually compiling, the reference
1768 was subsequently set up correctly.)
1769
1770 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
1771 it was being rejected as not supported by pcre_dfa_exec(), even though
1772 other assertions are supported. I have made pcre_dfa_exec() support
1773 (*FAIL).
1774
1775 16. The implementation of 13c above involved the invention of a new opcode,
1776 OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
1777 cannot be changed at match time, I realized I could make a small
1778 improvement to matching performance by compiling OP_ALLANY instead of
1779 OP_ANY for "." when DOTALL was set, and then removing the runtime tests
1780 on the OP_ANY path.
1781
1782 17. Compiling pcretest on Windows with readline support failed without the
1783 following two fixes: (1) Make the unistd.h include conditional on
1784 HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
1785
1786 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
1787 ncurses library to be included for pcretest when ReadLine support is
1788 requested, but also to allow for it to be overridden. This patch came from
1789 Daniel Bergström.
1790
1791 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
1792 as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
1793 any errors with the current Unicode tables. Thanks to Peter Kankowski for
1794 spotting this.
1795
1796
1797 Version 7.6 28-Jan-08
1798 ---------------------
1799
1800 1. A character class containing a very large number of characters with
1801 codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
1802 overflow.
1803
1804 2. Patch to cut out the "long long" test in pcrecpp_unittest when
1805 HAVE_LONG_LONG is not defined.
1806
1807 3. Applied Christian Ehrlicher's patch to update the CMake build files to
1808 bring them up to date and include new features. This patch includes:
1809
1810 - Fixed PH's badly added libz and libbz2 support.
1811 - Fixed a problem with static linking.
1812 - Added pcredemo. [But later removed - see 7 below.]
1813 - Fixed dftables problem and added an option.
1814 - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
1815 HAVE_LONG_LONG.
1816 - Added readline support for pcretest.
1817 - Added an listing of the option settings after cmake has run.
1818
1819 4. A user submitted a patch to Makefile that makes it easy to create
1820 "pcre.dll" under mingw when using Configure/Make. I added stuff to
1821 Makefile.am that cause it to include this special target, without
1822 affecting anything else. Note that the same mingw target plus all
1823 the other distribution libraries and programs are now supported
1824 when configuring with CMake (see 6 below) instead of with
1825 Configure/Make.
1826
1827 5. Applied Craig's patch that moves no_arg into the RE class in the C++ code.
1828 This is an attempt to solve the reported problem "pcrecpp::no_arg is not
1829 exported in the Windows port". It has not yet been confirmed that the patch
1830 solves the problem, but it does no harm.
1831
1832 6. Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
1833 NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
1834 with CMake, and also correct the comment about stack recursion.
1835
1836 7. Remove the automatic building of pcredemo from the ./configure system and
1837 from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
1838 of a program that users should build themselves after PCRE is installed, so
1839 building it automatically is not really right. What is more, it gave
1840 trouble in some build environments.
1841
1842 8. Further tidies to CMakeLists.txt from Sheri and Christian.
1843
1844
1845 Version 7.5 10-Jan-08
1846 ---------------------
1847
1848 1. Applied a patch from Craig: "This patch makes it possible to 'ignore'
1849 values in parens when parsing an RE using the C++ wrapper."
1850
1851 2. Negative specials like \S did not work in character classes in UTF-8 mode.
1852 Characters greater than 255 were excluded from the class instead of being
1853 included.
1854
1855 3. The same bug as (2) above applied to negated POSIX classes such as
1856 [:^space:].
1857
1858 4. PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
1859 defined or documented. It seems to have been a typo for PCRE_STATIC, so
1860 I have changed it.
1861
1862 5. The construct (?&) was not diagnosed as a syntax error (it referenced the
1863 first named subpattern) and a construct such as (?&a) would reference the
1864 first named subpattern whose name started with "a" (in other words, the
1865 length check was missing). Both these problems are fixed. "Subpattern name
1866 expected" is now given for (?&) (a zero-length name), and this patch also
1867 makes it give the same error for \k'' (previously it complained that that
1868 was a reference to a non-existent subpattern).
1869
1870 6. The erroneous patterns (?+-a) and (?-+a) give different error messages;
1871 this is right because (?- can be followed by option settings as well as by
1872 digits. I have, however, made the messages clearer.
1873
1874 7. Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
1875 than the number used in the conditional) now cause a compile-time error.
1876 This is actually not compatible with Perl, which accepts such patterns, but
1877 treats the conditional as always being FALSE (as PCRE used to), but it
1878 seems to me that giving a diagnostic is better.
1879
1880 8. Change "alphameric" to the more common word "alphanumeric" in comments
1881 and messages.
1882
1883 9. Fix two occurrences of "backslash" in comments that should have been
1884 "backspace".
1885
1886 10. Remove two redundant lines of code that can never be obeyed (their function
1887 was moved elsewhere).
1888
1889 11. The program that makes PCRE's Unicode character property table had a bug
1890 which caused it to generate incorrect table entries for sequences of
1891 characters that have the same character type, but are in different scripts.
1892 It amalgamated them into a single range, with the script of the first of
1893 them. In other words, some characters were in the wrong script. There were
1894 thirteen such cases, affecting characters in the following ranges:
1895
1896 U+002b0 - U+002c1
1897 U+0060c - U+0060d
1898 U+0061e - U+00612
1899 U+0064b - U+0065e
1900 U+0074d - U+0076d
1901 U+01800 - U+01805
1902 U+01d00 - U+01d77
1903 U+01d9b - U+01dbf
1904 U+0200b - U+0200f
1905 U+030fc - U+030fe
1906 U+03260 - U+0327f
1907 U+0fb46 - U+0fbb1
1908 U+10450 - U+1049d
1909
1910 12. The -o option (show only the matching part of a line) for pcregrep was not
1911 compatible with GNU grep in that, if there was more than one match in a
1912 line, it showed only the first of them. It now behaves in the same way as
1913 GNU grep.
1914
1915 13. If the -o and -v options were combined for pcregrep, it printed a blank
1916 line for every non-matching line. GNU grep prints nothing, and pcregrep now
1917 does the same. The return code can be used to tell if there were any
1918 non-matching lines.
1919
1920 14. Added --file-offsets and --line-offsets to pcregrep.
1921
1922 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
1923 infinitely looping recursion. The bug was that positive lookaheads were not
1924 being skipped when checking for a possible empty match (negative lookaheads
1925 and both kinds of lookbehind were skipped).
1926
1927 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
1928 inclusion of <windows.h> to before rather than after the definition of
1929 INVALID_FILE_ATTRIBUTES (patch from David Byron).
1930
1931 17. Specifying a possessive quantifier with a specific limit for a Unicode
1932 character property caused pcre_compile() to compile bad code, which led at
1933 runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
1934 are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
1935 caused the error; without that there was no problem.
1936
1937 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
1938
1939 19. Added --enable-pcretest-libreadline.
1940
1941 20. In pcrecpp.cc, the variable 'count' was incremented twice in
1942 RE::GlobalReplace(). As a result, the number of replacements returned was
1943 double what it should be. I removed one of the increments, but Craig sent a
1944 later patch that removed the other one (the right fix) and added unit tests
1945 that check the return values (which was not done before).
1946
1947 21. Several CMake things:
1948
1949 (1) Arranged that, when cmake is used on Unix, the libraries end up with
1950 the names libpcre and libpcreposix, not just pcre and pcreposix.
1951
1952 (2) The above change means that pcretest and pcregrep are now correctly
1953 linked with the newly-built libraries, not previously installed ones.
1954
1955 (3) Added PCRE_SUPPORT_LIBREADLINE, PCRE_SUPPORT_LIBZ, PCRE_SUPPORT_LIBBZ2.
1956
1957 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
1958 crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
1959 UTF-8 newline character). The key issue is that the pattern starts .*;
1960 this means that the match must be either at the beginning, or after a
1961 newline. The bug was in the code for advancing after a failed match and
1962 checking that the new position followed a newline. It was not taking
1963 account of UTF-8 characters correctly.
1964
1965 23. PCRE was behaving differently from Perl in the way it recognized POSIX
1966 character classes. PCRE was not treating the sequence [:...:] as a
1967 character class unless the ... were all letters. Perl, however, seems to
1968 allow any characters between [: and :], though of course it rejects as
1969 unknown any "names" that contain non-letters, because all the known class
1970 names consist only of letters. Thus, Perl gives an error for [[:1234:]],
1971 for example, whereas PCRE did not - it did not recognize a POSIX character
1972 class. This seemed a bit dangerous, so the code has been changed to be
1973 closer to Perl. The behaviour is not identical to Perl, because PCRE will
1974 diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
1975 treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
1976 Perl does, and where it didn't before.
1977
1978 24. Rewrite so as to remove the single use of %n from pcregrep because in some
1979 Windows environments %n is disabled by default.
1980
1981
1982 Version 7.4 21-Sep-07
1983 ---------------------
1984
1985 1. Change 7.3/28 was implemented for classes by looking at the bitmap. This
1986 means that a class such as [\s] counted as "explicit reference to CR or
1987 LF". That isn't really right - the whole point of the change was to try to
1988 help when there was an actual mention of one of the two characters. So now
1989 the change happens only if \r or \n (or a literal CR or LF) character is
1990 encountered.
1991
1992 2. The 32-bit options word was also used for 6 internal flags, but the numbers
1993 of both had grown to the point where there were only 3 bits left.
1994 Fortunately, there was spare space in the data structure, and so I have
1995 moved the internal flags into a new 16-bit field to free up more option
1996 bits.
1997
1998 3. The appearance of (?J) at the start of a pattern set the DUPNAMES option,
1999 but did not set the internal JCHANGED flag - either of these is enough to
2000 control the way the "get" function works - but the PCRE_INFO_JCHANGED
2001 facility is supposed to tell if (?J) was ever used, so now (?J) at the
2002 start sets both bits.
2003
2004 4. Added options (at build time, compile time, exec time) to change \R from
2005 matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
2006
2007 5. doc/pcresyntax.html was missing from the distribution.
2008
2009 6. Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
2010 compatibility, even though it is no longer used.
2011
2012 7. Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
2013 strtoull to pcrecpp.cc to select the available functions in WIN32 when the
2014 windows.h file is present (where different names are used). [This was
2015 reversed later after testing - see 16 below.]
2016
2017 8. Changed all #include <config.h> to #include "config.h". There were also
2018 some further <pcre.h> cases that I changed to "pcre.h".
2019
2020 9. When pcregrep was used with the --colour option, it missed the line ending
2021 sequence off the lines that it output.
2022
2023 10. It was pointed out to me that arrays of string pointers cause lots of
2024 relocations when a shared library is dynamically loaded. A technique of
2025 using a single long string with a table of offsets can drastically reduce
2026 these. I have refactored PCRE in four places to do this. The result is
2027 dramatic:
2028
2029 Originally: 290
2030 After changing UCP table: 187
2031 After changing error message table: 43
2032 After changing table of "verbs" 36
2033 After changing table of Posix names 22
2034
2035 Thanks to the folks working on Gregex for glib for this insight.
2036
2037 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
2038 unicode-properties was also set.
2039
2040 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
2041
2042 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
2043 checked only for CRLF.
2044
2045 14. Added casts to pcretest.c to avoid compiler warnings.
2046
2047 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
2048
2049 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
2050 and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
2051 entirely. This removes changes made in 7 above.
2052
2053 17. The CMake files have been updated, and there is now more information about
2054 building with CMake in the NON-UNIX-USE document.
2055
2056
2057 Version 7.3 28-Aug-07
2058 ---------------------
2059
2060 1. In the rejigging of the build system that eventually resulted in 7.1, the
2061 line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
2062 brackets there is not right, since it causes compilers to look for an
2063 installed pcre.h, not the version that is in the source that is being
2064 compiled (which of course may be different). I have changed it back to:
2065
2066 #include "pcre.h"
2067
2068 I have a vague recollection that the change was concerned with compiling in
2069 different directories, but in the new build system, that is taken care of
2070 by the VPATH setting the Makefile.
2071
2072 2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
2073 when the subject happened to end in the byte 0x85 (e.g. if the last
2074 character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
2075 characters but of course it shouldn't be taken as a newline when it is part
2076 of another character. The bug was that, for an unlimited repeat of . in
2077 not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
2078 characters when looking for a newline.
2079
2080 3. A small performance improvement in the DOTALL UTF-8 mode .* case.
2081
2082 4. Debugging: adjusted the names of opcodes for different kinds of parentheses
2083 in debug output.
2084
2085 5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
2086 long printing in the pcrecpp unittest when running under MinGW.
2087
2088 6. ESC_K was left out of the EBCDIC table.
2089
2090 7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
2091 parentheses; I made it 1000, which seemed large enough. Unfortunately, the
2092 limit also applies to "virtual nesting" when a pattern is recursive, and in
2093 this case 1000 isn't so big. I have been able to remove this limit at the
2094 expense of backing off one optimization in certain circumstances. Normally,
2095 when pcre_exec() would call its internal match() function recursively and
2096 immediately return the result unconditionally, it uses a "tail recursion"
2097 feature to save stack. However, when a subpattern that can match an empty
2098 string has an unlimited repetition quantifier, it no longer makes this
2099 optimization. That gives it a stack frame in which to save the data for
2100 checking that an empty string has been matched. Previously this was taken
2101 from the 1000-entry workspace that had been reserved. So now there is no
2102 explicit limit, but more stack is used.
2103
2104 8. Applied Daniel's patches to solve problems with the import/export magic
2105 syntax that is required for Windows, and which was going wrong for the
2106 pcreposix and pcrecpp parts of the library. These were overlooked when this
2107 problem was solved for the main library.
2108
2109 9. There were some crude static tests to avoid integer overflow when computing
2110 the size of patterns that contain repeated groups with explicit upper
2111 limits. As the maximum quantifier is 65535, the maximum group length was
2112 set at 30,000 so that the product of these two numbers did not overflow a
2113 32-bit integer. However, it turns out that people want to use groups that
2114 are longer than 30,000 bytes (though not repeat them that many times).
2115 Change 7.0/17 (the refactoring of the way the pattern size is computed) has
2116 made it possible to implement the integer overflow checks in a much more
2117 dynamic way, which I have now done. The artificial limitation on group
2118 length has been removed - we now have only the limit on the total length of
2119 the compiled pattern, which depends on the LINK_SIZE setting.
2120
2121 10. Fixed a bug in the documentation for get/copy named substring when
2122 duplicate names are permitted. If none of the named substrings are set, the
2123 functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
2124 empty string.
2125
2126 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
2127 instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
2128 because the ] is interpreted as the first data character and the
2129 terminating ] is not found. PCRE has been made compatible with Perl in this
2130 regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
2131 cause memory overwriting.
2132
2133 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
2134 string has been matched (to stop an infinite loop). It was not recognizing
2135 a conditional subpattern that could match an empty string if that
2136 subpattern was within another subpattern. For example, it looped when
2137 trying to match (((?(1)X|))*) but it was OK with ((?(1)X|)*) where the
2138 condition was not nested. This bug has been fixed.
2139
2140 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
2141 past the start of the subject in the presence of bytes with the top bit
2142 set, for example "\x8aBCD".
2143
2144 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
2145 (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
2146
2147 14. Optimized (?!) to (*FAIL).
2148
2149 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
2150 This restricts code points to be within the range 0 to 0x10FFFF, excluding
2151 the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
2152 full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
2153 does: it's just the validity check that is more restrictive.
2154
2155 16. Inserted checks for integer overflows during escape sequence (backslash)
2156 processing, and also fixed erroneous offset values for syntax errors during
2157 backslash processing.
2158
2159 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
2160 for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
2161
2162 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
2163 caused an overrun.
2164
2165 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
2166 something other than just ASCII characters) inside a group that had an
2167 unlimited repeat caused a loop at compile time (while checking to see
2168 whether the group could match an empty string).
2169
2170 20. Debugging a pattern containing \p or \P could cause a crash. For example,
2171 [\P{Any}] did so. (Error in the code for printing property names.)
2172
2173 21. An orphan \E inside a character class could cause a crash.
2174
2175 22. A repeated capturing bracket such as (A)? could cause a wild memory
2176 reference during compilation.
2177
2178 23. There are several functions in pcre_compile() that scan along a compiled
2179 expression for various reasons (e.g. to see if it's fixed length for look
2180 behind). There were bugs in these functions when a repeated \p or \P was
2181 present in the pattern. These operators have additional parameters compared
2182 with \d, etc, and these were not being taken into account when moving along
2183 the compiled data. Specifically:
2184
2185 (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
2186 length.
2187
2188 (b) An item such as \pL+ within a repeated group could cause crashes or
2189 loops.
2190
2191 (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
2192 "reference to non-existent subpattern" error.
2193
2194 (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
2195
2196 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
2197 characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
2198
2199 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
2200
2201 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
2202 character were causing crashes (broken optimization).
2203
2204 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
2205 \p or \P) caused a compile-time loop.
2206
2207 28. More problems have arisen in unanchored patterns when CRLF is a valid line
2208 break. For example, the unstudied pattern [\r\n]A does not match the string
2209 "\r\nA" because change 7.0/46 below moves the current point on by two
2210 characters after failing to match at the start. However, the pattern \nA
2211 *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
2212 the same is true. There doesn't seem any very clean way out of this, but
2213 what I have chosen to do makes the common cases work: PCRE now takes note
2214 of whether there can be an explicit match for \r or \n anywhere in the
2215 pattern, and if so, 7.0/46 no longer applies. As part of this change,
2216 there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
2217 pattern has explicit CR or LF references.
2218
2219 29. Added (*CR) etc for changing newline setting at start of pattern.
2220
2221
2222 Version 7.2 19-Jun-07
2223 ---------------------
2224
2225 1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
2226 which is apparently normally available under Windows.
2227
2228 2. Re-jig the pcregrep tests with different newline settings in an attempt
2229 to make them independent of the local environment's newline setting.
2230
2231 3. Add code to configure.ac to remove -g from the CFLAGS default settings.
2232
2233 4. Some of the "internals" tests were previously cut out when the link size
2234 was not 2, because the output contained actual offsets. The recent new
2235 "Z" feature of pcretest means that these can be cut out, making the tests
2236 usable with all link sizes.
2237
2238 5. Implemented Stan Switzer's goto replacement for longjmp() when not using
2239 stack recursion. This gives a massive performance boost under BSD, but just
2240 a small improvement under Linux. However, it saves one field in the frame
2241 in all cases.
2242
2243 6. Added more features from the forthcoming Perl 5.10:
2244
2245 (a) (?-n) (where n is a string of digits) is a relative subroutine or
2246 recursion call. It refers to the nth most recently opened parentheses.
2247
2248 (b) (?+n) is also a relative subroutine call; it refers to the nth next
2249 to be opened parentheses.
2250
2251 (c) Conditions that refer to capturing parentheses can be specified
2252 relatively, for example, (?(-2)... or (?(+3)...
2253
2254 (d) \K resets the start of the current match so that everything before
2255 is not part of it.
2256
2257 (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
2258
2259 (f) \g{name} is another synonym - part of Perl 5.10's unification of
2260 reference syntax.
2261
2262 (g) (?| introduces a group in which the numbering of parentheses in each
2263 alternative starts with the same number.
2264
2265 (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
2266
2267 7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
2268 PCRE_INFO_JCHANGED.
2269
2270 8. A pattern such as (.*(.)?)* caused pcre_exec() to fail by either not
2271 terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
2272 for detecting groups that can match an empty string.
2273
2274 9. A pattern with a very large number of alternatives (more than several
2275 hundred) was running out of internal workspace during the pre-compile
2276 phase, where pcre_compile() figures out how much memory will be needed. A
2277 bit of new cunning has reduced the workspace needed for groups with
2278 alternatives. The 1000-alternative test pattern now uses 12 bytes of
2279 workspace instead of running out of the 4096 that are available.
2280
2281 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
2282
2283 11. Applied patch from Google to remove an optimization that didn't quite work.
2284 The report of the bug said:
2285
2286 pcrecpp::RE("a*").FullMatch("aaa") matches, while
2287 pcrecpp::RE("a*?").FullMatch("aaa") does not, and
2288 pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
2289
2290 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
2291 it matched the wrong number of bytes.
2292
2293
2294 Version 7.1 24-Apr-07
2295 ---------------------
2296
2297 1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
2298 that is more "standard", making use of automake and other Autotools. There
2299 is some re-arrangement of the files and adjustment of comments consequent
2300 on this.
2301
2302 2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
2303 for recursive directory scanning broke on some systems because the files
2304 are not scanned in any specific order and on different systems the order
2305 was different. A call to "sort" has been inserted into RunGrepTest for the
2306 approprate test as a short-term fix. In the longer term there may be an
2307 alternative.
2308
2309 3. I had an email from Eric Raymond about problems translating some of PCRE's
2310 man pages to HTML (despite the fact that I distribute HTML pages, some
2311 people do their own conversions for various reasons). The problems
2312 concerned the use of low-level troff macros .br and .in. I have therefore
2313 removed all such uses from the man pages (some were redundant, some could
2314 be replaced by .nf/.fi pairs). The 132html script that I use to generate
2315 HTML has been updated to handle .nf/.fi and to complain if it encounters
2316 .br or .in.
2317
2318 4. Updated comments in configure.ac that get placed in config.h.in and also
2319 arranged for config.h to be included in the distribution, with the name
2320 config.h.generic, for the benefit of those who have to compile without
2321 Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
2322
2323 5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
2324 Weber: (1) pcre_internal.h was missing some function renames; (2) updated
2325 makevp.bat for the current PCRE, using the additional files
2326 makevp_c.txt, makevp_l.txt, and pcregexp.pas.
2327
2328 6. A Windows user reported a minor discrepancy with test 2, which turned out
2329 to be caused by a trailing space on an input line that had got lost in his
2330 copy. The trailing space was an accident, so I've just removed it.
2331
2332 7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
2333 that is needed.
2334
2335 8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
2336 as "const" (a) because they are and (b) because it helps the PHP
2337 maintainers who have recently made a script to detect big data structures
2338 in the php code that should be moved to the .rodata section. I remembered
2339 to update Builducptable as well, so it won't revert if ucptable.h is ever
2340 re-created.
2341
2342 9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
2343 pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
2344 order to be able to cut out the UTF-8 tables in the latter when UTF-8
2345 support is not required. This saves 1.5-2K of code, which is important in
2346 some applications.
2347
2348 Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
2349 so as not to refer to the tables, even though these functions will never be
2350 called when UTF-8 support is disabled. Otherwise there are problems with a
2351 shared library.
2352
2353 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
2354
2355 (a) It was defining its arguments as char * instead of void *.
2356
2357 (b) It was assuming that all moves were upwards in memory; this was true
2358 a long time ago when I wrote it, but is no longer the case.
2359
2360 The emulated memove() is provided for those environments that have neither
2361 memmove() nor bcopy(). I didn't think anyone used it these days, but that
2362 is clearly not the case, as these two bugs were recently reported.
2363
2364 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
2365 and Detrail to create the HTML documentation, the .txt form of the man
2366 pages, and it removes trailing spaces from listed files. It also creates
2367 pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
2368 case, it wraps all the #defines with #ifndefs. This script should be run
2369 before "make dist".
2370
2371 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
2372 with Unicode property support.
2373
2374 (a) For a maximizing quantifier, if the two different cases of the
2375 character were of different lengths in their UTF-8 codings (there are
2376 some cases like this - I found 11), and the matching function had to
2377 back up over a mixture of the two cases, it incorrectly assumed they
2378 were both the same length.
2379
2380 (b) When PCRE was configured to use the heap rather than the stack for
2381 recursion during matching, it was not correctly preserving the data for
2382 the other case of a UTF-8 character when checking ahead for a match
2383 while processing a minimizing repeat. If the check also involved
2384 matching a wide character, but failed, corruption could cause an
2385 erroneous result when trying to check for a repeat of the original
2386 character.
2387
2388 13. Some tidying changes to the testing mechanism:
2389
2390 (a) The RunTest script now detects the internal link size and whether there
2391 is UTF-8 and UCP support by running ./pcretest -C instead of relying on
2392 values substituted by "configure". (The RunGrepTest script already did
2393 this for UTF-8.) The configure.ac script no longer substitutes the
2394 relevant variables.
2395
2396 (b) The debugging options /B and /D in pcretest show the compiled bytecode
2397 with length and offset values. This means that the output is different
2398 for different internal link sizes. Test 2 is skipped for link sizes
2399 other than 2 because of this, bypassing the problem. Unfortunately,
2400 there was also a test in test 3 (the locale tests) that used /B and
2401 failed for link sizes other than 2. Rather than cut the whole test out,
2402 I have added a new /Z option to pcretest that replaces the length and
2403 offset values with spaces. This is now used to make test 3 independent
2404 of link size. (Test 2 will be tidied up later.)
2405
2406 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
2407 segmentation fault instead of returning the appropriate error message.
2408
2409 15. In multiline mode when the newline sequence was set to "any", the pattern
2410 ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
2411 This doesn't seem right; it now treats the CRLF combination as the line
2412 ending, and so does not match in that case. It's only a pattern such as ^$
2413 that would hit this one: something like ^ABC$ would have failed after \r
2414 and then tried again after \r\n.
2415
2416 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
2417 in an attempt to make files that differ only in their line terminators
2418 compare equal. This works on Linux.
2419
2420 17. Under certain error circumstances pcregrep might try to free random memory
2421 as it exited. This is now fixed, thanks to valgrind.
2422
2423 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
2424 "abc\r\n\r\n", it found an unwanted second match after the second \r. This
2425 was because its rules for how to advance for /g after matching an empty
2426 string at the end of a line did not allow for this case. They now check for
2427 it specially.
2428
2429 20. pcretest is supposed to handle patterns and data of any length, by
2430 extending its buffers when necessary. It was getting this wrong when the
2431 buffer for a data line had to be extended.
2432
2433 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
2434 CRLF as a newline sequence.
2435
2436 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
2437 out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
2438 I have nevertheless tidied it up.
2439
2440 23. Added some casts to kill warnings from HP-UX ia64 compiler.
2441
2442 24. Added a man page for pcre-config.
2443
2444
2445 Version 7.0 19-Dec-06
2446 ---------------------
2447
2448 1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
2449 moving to gcc 4.1.1.
2450
2451 2. The -S option for pcretest uses setrlimit(); I had omitted to #include
2452 sys/time.h, which is documented as needed for this function. It doesn't
2453 seem to matter on Linux, but it showed up on some releases of OS X.
2454
2455 3. It seems that there are systems where bytes whose values are greater than
2456 127 match isprint() in the "C" locale. The "C" locale should be the
2457 default when a C program starts up. In most systems, only ASCII printing
2458 characters match isprint(). This difference caused the output from pcretest
2459 to vary, making some of the tests fail. I have changed pcretest so that:
2460
2461 (a) When it is outputting text in the compiled version of a pattern, bytes
2462 other than 32-126 are always shown as hex escapes.
2463
2464 (b) When it is outputting text that is a matched part of a subject string,
2465 it does the same, unless a different locale has been set for the match
2466 (using the /L modifier). In this case, it uses isprint() to decide.
2467
2468 4. Fixed a major bug that caused incorrect computation of the amount of memory
2469 required for a compiled pattern when options that changed within the
2470 pattern affected the logic of the preliminary scan that determines the
2471 length. The relevant options are -x, and -i in UTF-8 mode. The result was
2472 that the computed length was too small. The symptoms of this bug were
2473 either the PCRE error "internal error: code overflow" from pcre_compile(),
2474 or a glibc crash with a message such as "pcretest: free(): invalid next
2475 size (fast)". Examples of patterns that provoked this bug (shown in
2476 pcretest format) are:
2477
2478 /(?-x: )/x
2479 /(?x)(?-x: \s*#\s*)/
2480 /((?i)[\x{c0}])/8
2481 /(?i:[\x{c0}])/8
2482
2483 HOWEVER: Change 17 below makes this fix obsolete as the memory computation
2484 is now done differently.
2485
2486 5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
2487 wrapper classes; (b) implement a new function in the C++ scanner that is
2488 more efficient than the old way of doing things because it avoids levels of
2489 recursion in the regex matching; (c) add a paragraph to the documentation
2490 for the FullMatch() function.
2491
2492 6. The escape sequence \n was being treated as whatever was defined as
2493 "newline". Not only was this contrary to the documentation, which states
2494 that \n is character 10 (hex 0A), but it also went horribly wrong when
2495 "newline" was defined as CRLF. This has been fixed.
2496
2497 7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
2498 was being set to -1 for the "end of line" case (supposedly a value that no
2499 character can have). Though this value is never used (the check for end of
2500 line is "zero bytes in current character"), it caused compiler complaints.
2501 I've changed it to 0xffffffff.
2502
2503 8. In pcre_version.c, the version string was being built by a sequence of
2504 C macros that, in the event of PCRE_PRERELEASE being defined as an empty
2505 string (as it is for production releases) called a macro with an empty
2506 argument. The C standard says the result of this is undefined. The gcc
2507 compiler treats it as an empty string (which was what was wanted) but it is
2508 reported that Visual C gives an error. The source has been hacked around to
2509 avoid this problem.
2510
2511 9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
2512 builds of pcretest, and changed the call to _setmode() to use _O_BINARY
2513 instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
2514 of them did).
2515
2516 10. Originally, pcretest opened its input and output without "b"; then I was
2517 told that "b" was needed in some environments, so it was added for release
2518 5.0 to both the input and output. (It makes no difference on Unix-like
2519 systems.) Later I was told that it is wrong for the input on Windows. I've
2520 now abstracted the modes into two macros, to make it easier to fiddle with
2521 them, and removed "b" from the input mode under Windows.
2522
2523 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
2524
2525 12. Added -help and --help to pcretest as an official way of being reminded
2526 of the options.
2527
2528 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
2529 and pcrecpp.cc because they annoy compilers at high warning levels.
2530
2531 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
2532
2533 15. Fixed an occurrence of == in configure.ac that should have been = (shell
2534 scripts are not C programs :-) and which was not noticed because it works
2535 on Linux.
2536
2537 16. pcretest is supposed to handle any length of pattern and data line (as one
2538 line or as a continued sequence of lines) by extending its input buffer if
2539 necessary. This feature was broken for very long pattern lines, leading to
2540 a string of junk being passed to pcre_compile() if the pattern was longer
2541 than about 50K.
2542
2543 17. I have done a major re-factoring of the way pcre_compile() computes the
2544 amount of memory needed for a compiled pattern. Previously, there was code
2545 that made a preliminary scan of the pattern in order to do this. That was
2546 OK when PCRE was new, but as the facilities have expanded, it has become
2547 harder and harder to keep it in step with the real compile phase, and there
2548 have been a number of bugs (see for example, 4 above). I have now found a
2549 cunning way of running the real compile function in a "fake" mode that
2550 enables it to compute how much memory it would need, while actually only
2551 ever using a few hundred bytes of working memory and without too many
2552 tests of the mode. This should make future maintenance and development
2553 easier. A side effect of this work is that the limit of 200 on the nesting
2554 depth of parentheses has been removed (though this was never a serious
2555 limitation, I suspect). However, there is a downside: pcre_compile() now
2556 runs more slowly than before (30% or more, depending on the pattern). I
2557 hope this isn't a big issue. There is no effect on runtime performance.
2558
2559 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
2560 newline (only possible for the last line of a file) and it was a
2561 pattern that set a locale (followed by /Lsomething), pcretest crashed.
2562
2563 19. Added additional timing features to pcretest. (1) The -tm option now times
2564 matching only, not compiling. (2) Both -t and -tm can be followed, as a
2565 separate command line item, by a number that specifies the number of
2566 repeats to use when timing. The default is 50000; this gives better
2567 precision, but takes uncomfortably long for very large patterns.
2568
2569 20. Extended pcre_study() to be more clever in cases where a branch of a
2570 subpattern has no definite first character. For example, (a*|b*)[cd] would
2571 previously give no result from pcre_study(). Now it recognizes that the
2572 first character must be a, b, c, or d.
2573
2574 21. There was an incorrect error "recursive call could loop indefinitely" if
2575 a subpattern (or the entire pattern) that was being tested for matching an
2576 empty string contained only one non-empty item after a nested subpattern.
2577 For example, the pattern (?>\x{100}*)\d(?R) provoked this error
2578 incorrectly, because the \d was being skipped in the check.
2579
2580 22. The pcretest program now has a new pattern option /B and a command line
2581 option -b, which is equivalent to adding /B to every pattern. This causes
2582 it to show the compiled bytecode, without the additional information that
2583 -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
2584 is the same as /B/I).
2585
2586 23. A new optimization is now able automatically to treat some sequences such
2587 as a*b as a*+b. More specifically, if something simple (such as a character
2588 or a simple class like \d) has an unlimited quantifier, and is followed by
2589 something that cannot possibly match the quantified thing, the quantifier
2590 is automatically "possessified".
2591
2592 24. A recursive reference to a subpattern whose number was greater than 39
2593 went wrong under certain circumstances in UTF-8 mode. This bug could also
2594 have affected the operation of pcre_study().
2595
2596 25. Realized that a little bit of performance could be had by replacing
2597 (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
2598
2599 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
2600
2601 27. Possessive quantifiers such as a++ were previously implemented by turning
2602 them into atomic groups such as ($>a+). Now they have their own opcodes,
2603 which improves performance. This includes the automatically created ones
2604 from 23 above.
2605
2606 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
2607 lookahead was broken if it was not anchored. PCRE was mistakenly expecting
2608 the first matched character to be a colon. This applied both to named and
2609 numbered groups.
2610
2611 29. The ucpinternal.h header file was missing its idempotency #ifdef.
2612
2613 30. I was sent a "project" file called libpcre.a.dev which I understand makes
2614 building PCRE on Windows easier, so I have included it in the distribution.
2615
2616 31. There is now a check in pcretest against a ridiculously large number being
2617 returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
2618 loop, the loop is abandoned.
2619
2620 32. Forward references to subpatterns in conditions such as (?(2)...) where
2621 subpattern 2 is defined later cause pcre_compile() to search forwards in
2622 the pattern for the relevant set of parentheses. This search went wrong
2623 when there were unescaped parentheses in a character class, parentheses
2624 escaped with \Q...\E, or parentheses in a #-comment in /x mode.
2625
2626 33. "Subroutine" calls and backreferences were previously restricted to
2627 referencing subpatterns earlier in the regex. This restriction has now
2628 been removed.
2629
2630 34. Added a number of extra features that are going to be in Perl 5.10. On the
2631 whole, these are just syntactic alternatives for features that PCRE had
2632 previously implemented using the Python syntax or my own invention. The
2633 other formats are all retained for compatibility.
2634
2635 (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
2636 as (?P<name>...). The new forms, as well as being in Perl 5.10, are
2637 also .NET compatible.
2638
2639 (b) A recursion or subroutine call to a named group can now be defined as
2640 (?&name) as well as (?P>name).
2641
2642 (c) A backreference to a named group can now be defined as \k<name> or
2643 \k'name' as well as (?P=name). The new forms, as well as being in Perl
2644 5.10, are also .NET compatible.
2645
2646 (d) A conditional reference to a named group can now use the syntax
2647 (?(<name>) or (?('name') as well as (?(name).
2648
2649 (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
2650 groups (named and numbered) that are never evaluated inline, but can be
2651 called as "subroutines" from elsewhere. In effect, the DEFINE condition
2652 is always false. There may be only one alternative in such a group.
2653
2654 (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
2655 as the simple (?(R). The condition is true only if the most recent
2656 recursion is that of the given number or name. It does not search out
2657 through the entire recursion stack.
2658
2659 (g) The escape \gN or \g{N} has been added, where N is a positive or
2660 negative number, specifying an absolute or relative reference.
2661
2662 35. Tidied to get rid of some further signed/unsigned compiler warnings and
2663 some "unreachable code" warnings.
2664
2665 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
2666 things, this adds five new scripts.
2667
2668 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
2669 There were also incompatibilities regarding the handling of \Q..\E inside
2670 character classes, for example with patterns like [\Qa\E-\Qz\E] where the
2671 hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
2672
2673 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
2674 matches an empty string, and forcibly breaks the loop. There were bugs in
2675 this code in non-simple cases. For a pattern such as ^(a()*)* matched
2676 against aaaa the result was just "a" rather than "aaaa", for example. Two
2677 separate and independent bugs (that affected different cases) have been
2678 fixed.
2679
2680 39. Refactored the code to abolish the use of different opcodes for small
2681 capturing bracket numbers. This is a tidy that I avoided doing when I
2682 removed the limit on the number of capturing brackets for 3.5 back in 2001.
2683 The new approach is not only tidier, it makes it possible to reduce the
2684 memory needed to fix the previous bug (38).
2685
2686 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
2687 sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
2688 processing dot, circumflex, or dollar metacharacters, or #-comments in /x
2689 mode.
2690
2691 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
2692 report.
2693
2694 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
2695 copy construction and assignment in the C++ wrapper.
2696
2697 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
2698 couple of bugs that could have given wrong results in the "--newline=crlf"
2699 case.
2700
2701 44. Added a number of casts and did some reorganization of signed/unsigned int
2702 variables following suggestions from Dair Grant. Also renamed the variable
2703 "this" as "item" because it is a C++ keyword.
2704
2705 45. Arranged for dftables to add
2706
2707 #include "pcre_internal.h"
2708
2709 to pcre_chartables.c because without it, gcc 4.x may remove the array
2710 definition from the final binary if PCRE is built into a static library and
2711 dead code stripping is activated.
2712
2713 46. For an unanchored pattern, if a match attempt fails at the start of a
2714 newline sequence, and the newline setting is CRLF or ANY, and the next two
2715 characters are CRLF, advance by two characters instead of one.
2716
2717
2718 Version 6.7 04-Jul-06
2719 ---------------------
2720
2721 1. In order to handle tests when input lines are enormously long, pcretest has
2722 been re-factored so that it automatically extends its buffers when
2723 necessary. The code is crude, but this _is_ just a test program. The
2724 default size has been increased from 32K to 50K.
2725
2726 2. The code in pcre_study() was using the value of the re argument before
2727 testing it for NULL. (Of course, in any sensible call of the function, it
2728 won't be NULL.)
2729
2730 3. The memmove() emulation function in pcre_internal.h, which is used on
2731 systems that lack both memmove() and bcopy() - that is, hardly ever -
2732 was missing a "static" storage class specifier.
2733
2734 4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
2735 containing an extended class (one that cannot be represented by a bitmap
2736 because it contains high-valued characters or Unicode property items, e.g.
2737 [\pZ]). Almost always one would set UTF-8 mode when processing such a
2738 pattern, but PCRE should not loop if you do not (it no longer does).
2739 [Detail: two cases were found: (a) a repeated subpattern containing an
2740 extended class; (b) a recursive reference to a subpattern that followed a
2741 previous extended class. It wasn't skipping over the extended class
2742 correctly when UTF-8 mode was not set.]
2743
2744 5. A negated single-character class was not being recognized as fixed-length
2745 in lookbehind assertions such as (?<=[^f]), leading to an incorrect
2746 compile error "lookbehind assertion is not fixed length".
2747
2748 6. The RunPerlTest auxiliary script was showing an unexpected difference
2749 between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
2750 write a Perl script that can interpret lines of an input file either as
2751 byte characters or as UTF-8, which is what "perltest" was being required to
2752 do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
2753 can't do is switch easily at run time between having the "use utf8;" pragma
2754 or not. In the end, I fudged it by using the RunPerlTest script to insert
2755 "use utf8;" explicitly for the UTF-8 tests.
2756
2757 7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
2758 the end of the subject string, contrary to the documentation and to what
2759 Perl does. This was true of both matching functions. Now it matches only at
2760 the start of the subject and immediately after *internal* newlines.
2761
2762 8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
2763 a pointer to an int instead of a pointer to an unsigned long int. This
2764 caused problems on 64-bit systems.
2765
2766 9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
2767 instance of the 'standard' template library not being so standard".
2768
2769 10. There was no check on the number of named subpatterns nor the maximum
2770 length of a subpattern name. The product of these values is used to compute
2771 the size of the memory block for a compiled pattern. By supplying a very
2772 long subpattern name and a large number of named subpatterns, the size
2773 computation could be caused to overflow. This is now prevented by limiting
2774 the length of names to 32 characters, and the number of named subpatterns
2775 to 10,000.
2776
2777 11. Subpatterns that are repeated with specific counts have to be replicated in
2778 the compiled pattern. The size of memory for this was computed from the
2779 length of the subpattern and the repeat count. The latter is limited to
2780 65535, but there was no limit on the former, meaning that integer overflow
2781 could in principle occur. The compiled length of a repeated subpattern is
2782 now limited to 30,000 bytes in order to prevent this.
2783
2784 12. Added the optional facility to have named substrings with the same name.
2785
2786 13. Added the ability to use a named substring as a condition, using the
2787 Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
2788 are numbers (not recommended). Forward references are permitted.
2789
2790 14. Added forward references in named backreferences (if you see what I mean).
2791
2792 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
2793 pattern could run off the end of the subject. For example, the pattern
2794 "(?s)(.{1,5})"8 did this with the subject "ab".
2795
2796 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
2797 PCRE_CASELESS was set when matching characters that were quantified with ?
2798 or *.
2799
2800 17. A character class other than a single negated character that had a minimum
2801 but no maximum quantifier - for example [ab]{6,} - was not handled
2802 correctly by pce_dfa_exec(). It would match only one character.
2803
2804 18. A valid (though odd) pattern that looked like a POSIX character
2805 class but used an invalid character after [ (for example [[,abc,]]) caused
2806 pcre_compile() to give the error "Failed: internal error: code overflow" or
2807 in some cases to crash with a glibc free() error. This could even happen if
2808 the pattern terminated after [[ but there just happened to be a sequence of
2809 letters, a binary zero, and a closing ] in the memory that followed.
2810
2811 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
2812 over the years. Originally (before any Unicode support), just the bottom 8
2813 bits were taken. Thus, for example, \500 really meant \100. Nowadays the
2814 output from "man perlunicode" includes this:
2815
2816 The regular expression compiler produces polymorphic opcodes. That
2817 is, the pattern adapts to the data and automatically switches to
2818 the Unicode character scheme when presented with Unicode data--or
2819 instead uses a traditional byte scheme when presented with byte
2820 data.
2821
2822 Sadly, a wide octal escape does not cause a switch, and in a string with
2823 no other multibyte characters, these octal escapes are treated as before.
2824 Thus, in Perl, the pattern /\500/ actually matches \100 but the pattern
2825 /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
2826 Unicode string.
2827
2828 I have not perpetrated such confusion in PCRE. Up till now, it took just
2829 the bottom 8 bits, as in old Perl. I have now made octal escapes with
2830 values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
2831 translate to the appropriate multibyte character.
2832
2833 29. Applied some refactoring to reduce the number of warnings from Microsoft
2834 and Borland compilers. This has included removing the fudge introduced
2835 seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
2836 a warning about an unused variable.
2837
2838 21. PCRE has not included VT (character 0x0b) in the set of whitespace
2839 characters since release 4.0, because Perl (from release 5.004) does not.
2840 [Or at least, is documented not to: some releases seem to be in conflict
2841 with the documentation.] However, when a pattern was studied with
2842 pcre_study() and all its branches started with \s, PCRE still included VT
2843 as a possible starting character. Of course, this did no harm; it just
2844 caused an unnecessary match attempt.
2845
2846 22. Removed a now-redundant internal flag bit that recorded the fact that case
2847 dependency changed within the pattern. This was once needed for "required
2848 byte" processing, but is no longer used. This recovers a now-scarce options
2849 bit. Also moved the least significant internal flag bit to the most-
2850 significant bit of the word, which was not previously used (hangover from
2851 the days when it was an int rather than a uint) to free up another bit for
2852 the future.
2853
2854 23. Added support for CRLF line endings as well as CR and LF. As well as the
2855 default being selectable at build time, it can now be changed at runtime
2856 via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
2857 specify that it is scanning data with non-default line endings.
2858
2859 24. Changed the definition of CXXLINK to make it agree with the definition of
2860 LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
2861
2862 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
2863 recursions. This makes a big different to stack usage for some patterns.
2864
2865 26. If a subpattern containing a named recursion or subroutine reference such
2866 as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
2867 the space required for the compiled pattern went wrong and gave too small a
2868 value. Depending on the environment, this could lead to "Failed: internal
2869 error: code overflow at offset 49" or "glibc detected double free or
2870 corruption" errors.
2871
2872 27. Applied patches from Google (a) to support the new newline modes and (b) to
2873 advance over multibyte UTF-8 characters in GlobalReplace.
2874
2875 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
2876 difference for some implementation of PCRE in some Windows version.
2877
2878 29. Added some extra testing facilities to pcretest:
2879
2880 \q<number> in a data line sets the "match limit" value
2881 \Q<number> in a data line sets the "match recursion limt" value
2882 -S <number> sets the stack size, where <number> is in megabytes
2883
2884 The -S option isn't available for Windows.
2885
2886
2887 Version 6.6 06-Feb-06
2888 ---------------------
2889
2890 1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
2891 in pcreposix.h. I have copied the definition from pcre.h.
2892
2893 2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
2894 because pcre.h is no longer a built file.
2895
2896 3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
2897 not normally included in the compiled code.
2898
2899
2900 Version 6.5 01-Feb-06
2901 ---------------------
2902
2903 1. When using the partial match feature with pcre_dfa_exec(), it was not
2904 anchoring the second and subsequent partial matches at the new starting
2905 point. This could lead to incorrect results. For example, with the pattern
2906 /1234/, partially matching against "123" and then "a4" gave a match.
2907
2908 2. Changes to pcregrep:
2909
2910 (a) All non-match returns from pcre_exec() were being treated as failures
2911 to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
2912 error message is output. Some extra information is given for the
2913 PCRE_ERROR_MATCHLIMIT and PCRE_ERROR_RECURSIONLIMIT errors, which are
2914 probably the only errors that are likely to be caused by users (by
2915 specifying a regex that has nested indefinite repeats, for instance).
2916 If there are more than 20 of these errors, pcregrep is abandoned.
2917
2918 (b) A binary zero was treated as data while matching, but terminated the
2919 output line if it was written out. This has been fixed: binary zeroes
2920 are now no different to any other data bytes.
2921
2922 (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
2923 used to set a locale for matching. The --locale=xxxx long option has
2924 been added (no short equivalent) to specify a locale explicitly on the
2925 pcregrep command, overriding the environment variables.
2926
2927 (d) When -B was used with -n, some line numbers in the output were one less
2928 than they should have been.
2929
2930 (e) Added the -o (--only-matching) option.
2931
2932 (f) If -A or -C was used with -c (count only), some lines of context were
2933 accidentally printed for the final match.
2934
2935 (g) Added the -H (--with-filename) option.
2936
2937 (h) The combination of options -rh failed to suppress file names for files
2938 that were found from directory arguments.
2939
2940 (i) Added the -D (--devices) and -d (--directories) options.
2941
2942 (j) Added the -F (--fixed-strings) option.
2943
2944 (k) Allow "-" to be used as a file name for -f as well as for a data file.
2945
2946 (l) Added the --colo(u)r option.
2947
2948 (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
2949 is not present by default.
2950
2951 3. A nasty bug was discovered in the handling of recursive patterns, that is,
2952 items such as (?R) or (?1), when the recursion could match a number of
2953 alternatives. If it matched one of the alternatives, but subsequently,
2954 outside the recursion, there was a failure, the code tried to back up into
2955 the recursion. However, because of the way PCRE is implemented, this is not
2956 possible, and the result was an incorrect result from the match.
2957
2958 In order to prevent this happening, the specification of recursion has
2959 been changed so that all such subpatterns are automatically treated as
2960 atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
2961
2962 4. I had overlooked the fact that, in some locales, there are characters for
2963 which isalpha() is true but neither isupper() nor islower() are true. In
2964 the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
2965 and ordfeminine) are like this. This affected the treatment of \w and \W
2966 when they appeared in character classes, but not when they appeared outside
2967 a character class. The bit map for "word" characters is now created
2968 separately from the results of isalnum() instead of just taking it from the
2969 upper, lower, and digit maps. (Plus the underscore character, of course.)
2970
2971 5. The above bug also affected the handling of POSIX character classes such as
2972 [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
2973 permanent tables. Instead, the bit maps for such a class were previously
2974 created as the appropriate unions of the upper, lower, and digit bitmaps.
2975 Now they are created by subtraction from the [[:word:]] class, which has
2976 its own bitmap.
2977
2978 6. The [[:blank:]] character class matches horizontal, but not vertical space.
2979 It is created by subtracting the vertical space characters (\x09, \x0a,
2980 \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
2981 subtraction was done in the overall bitmap for a character class, meaning
2982 that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
2983 be recognized. This bug has been fixed.
2984
2985 7. Patches from the folks at Google:
2986
2987 (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
2988 real life, but is still worth protecting against".
2989
2990 (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
2991 regular expressions".
2992
2993 (c) pcre_scanner.cc: avoid use of std::count() because not all systems
2994 have it.
2995
2996 (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
2997 "configure" and the latter not, in order to fix a problem somebody had
2998 with compiling the Arg class on HP-UX.
2999
3000 (e) Improve the error-handling of the C++ wrapper a little bit.
3001
3002 (f) New tests for checking recursion limiting.
3003
3004 8. The pcre_memmove() function, which is used only if the environment does not
3005 have a standard memmove() function (and is therefore rarely compiled),
3006 contained two bugs: (a) use of int instead of size_t, and (b) it was not
3007 returning a result (though PCRE never actually uses the result).
3008
3009 9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
3010 large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
3011 returned instead of calling malloc() with an overflowing number that would
3012 most likely cause subsequent chaos.
3013
3014 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
3015
3016 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
3017 with this option is matched, the nmatch and pmatch options of regexec() are
3018 ignored.
3019
3020 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
3021 provided in case anyone wants to the the POSIX interface with UTF-8
3022 strings.
3023
3024 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
3025 C++ linking (needed for some HP-UX environments).
3026
3027 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
3028 (unused parameter) and in the pcre_printint() function (omitted "default"
3029 switch label when the default is to do nothing).
3030
3031 15. Added some code to make it possible, when PCRE is compiled as a C++
3032 library, to replace subject pointers for pcre_exec() with a smart pointer
3033 class, thus making it possible to process discontinuous strings.
3034
3035 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
3036 much the same function. They were added by different people who were trying
3037 to make PCRE easy to compile on non-Unix systems. It has been suggested
3038 that PCRE_EXPORT be abolished now that there is more automatic apparatus
3039 for compiling on Windows systems. I have therefore replaced it with
3040 PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
3041 defaults to "extern" for C or "extern C" for C++, which works fine on
3042 Unix-like systems. It is now possible to override the value of PCRE_DATA_
3043 SCOPE with something explicit in config.h. In addition:
3044
3045 (a) pcreposix.h still had just "extern" instead of either of these macros;
3046 I have replaced it with PCRE_DATA_SCOPE.
3047
3048 (b) Functions such as _pcre_xclass(), which are internal to the library,
3049 but external in the C sense, all had PCRE_EXPORT in their definitions.
3050 This is apparently wrong for the Windows case, so I have removed it.
3051 (It makes no difference on Unix-like systems.)
3052
3053 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
3054 of recursive calls to match(). This is different to MATCH_LIMIT because
3055 that limits the total number of calls to match(), not all of which increase
3056 the depth of recursion. Limiting the recursion depth limits the amount of
3057 stack (or heap if NO_RECURSE is set) that is used. The default can be set
3058 when PCRE is compiled, and changed at run time. A patch from Google adds
3059 this functionality to the C++ interface.
3060
3061 18. Changes to the handling of Unicode character properties:
3062
3063 (a) Updated the table to Unicode 4.1.0.
3064
3065 (b) Recognize characters that are not in the table as "Cn" (undefined).
3066
3067 (c) I revised the way the table is implemented to a much improved format
3068 which includes recognition of ranges. It now supports the ranges that
3069 are defined in UnicodeData.txt, and it also amalgamates other
3070 characters into ranges. This has reduced the number of entries in the
3071 table from around 16,000 to around 3,000, thus reducing its size
3072 considerably. I realized I did not need to use a tree structure after
3073 all - a binary chop search is just as efficient. Having reduced the
3074 number of entries, I extended their size from 6 bytes to 8 bytes to
3075 allow for more data.
3076
3077 (d) Added support for Unicode script names via properties such as \p{Han}.
3078
3079 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
3080 matching that character.
3081
3082 20. When matching a repeated Unicode property with a minimum greater than zero,
3083 (for example \pL{2,}), PCRE could look past the end of the subject if it
3084 reached it while seeking the minimum number of characters. This could
3085 happen only if some of the characters were more than one byte long, because
3086 there is a check for at least the minimum number of bytes.
3087
3088 21. Refactored the implementation of \p and \P so as to be more general, to
3089 allow for more different types of property in future. This has changed the
3090 compiled form incompatibly. Anybody with saved compiled patterns that use
3091 \p or \P will have to recompile them.
3092
3093 22. Added "Any" and "L&" to the supported property types.
3094
3095 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
3096 but give a compile time error if the value is greater than 0xff.
3097
3098 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
3099 accidentally not being installed or uninstalled.
3100
3101 25. The pcre.h file was built from pcre.h.in, but the only changes that were
3102 made were to insert the current release number. This seemed silly, because
3103 it made things harder for people building PCRE on systems that don't run
3104 "configure". I have turned pcre.h into a distributed file, no longer built
3105 by "configure", with the version identification directly included. There is
3106 no longer a pcre.h.in file.
3107
3108 However, this change necessitated a change to the pcre-config script as
3109 well. It is built from pcre-config.in, and one of the substitutions was the
3110 release number. I have updated configure.ac so that ./configure now finds
3111 the release number by grepping pcre.h.
3112
3113 26. Added the ability to run the tests under valgrind.
3114
3115
3116 Version 6.4 05-Sep-05
3117 ---------------------
3118
3119 1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
3120 "--" to be printed when multiple files were scanned, even when none of the
3121 -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
3122 consider it to be a bug, and have restored the previous behaviour.
3123
3124 2. A couple of code tidies to get rid of compiler warnings.
3125
3126 3. The pcretest program used to cheat by referring to symbols in the library
3127 whose names begin with _pcre_. These are internal symbols that are not
3128 really supposed to be visible externally, and in some environments it is
3129 possible to suppress them. The cheating is now confined to including
3130 certain files from the library's source, which is a bit cleaner.
3131
3132 4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
3133 file's purpose clearer.
3134
3135 5. Reorganized pcre_ucp_findchar().
3136
3137
3138 Version 6.3 15-Aug-05
3139 ---------------------
3140
3141 1. The file libpcre.pc.in did not have general read permission in the tarball.
3142
3143 2. There were some problems when building without C++ support:
3144
3145 (a) If C++ support was not built, "make install" and "make test" still
3146 tried to test it.
3147
3148 (b) There were problems when the value of CXX was explicitly set. Some
3149 changes have been made to try to fix these, and ...
3150
3151 (c) --disable-cpp can now be used to explicitly disable C++ support.
3152
3153 (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
3154 backslash in a target when C++ was disabled. This confuses some
3155 versions of "make", apparently. Using an intermediate variable solves
3156 this. (Same for CPP_LOBJ.)
3157
3158 3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
3159 (non-Windows) now includes $(CFLAGS) because these flags are sometimes
3160 necessary on certain architectures.
3161
3162 4. Added a setting of -export-symbols-regex to the link command to remove
3163 those symbols that are exported in the C sense, but actually are local
3164 within the library, and not documented. Their names all begin with
3165 "_pcre_". This is not a perfect job, because (a) we have to except some
3166 symbols that pcretest ("illegally") uses, and (b) the facility isn't always
3167 available (and never for static libraries). I have made a note to try to
3168 find a way round (a) in the future.
3169
3170
3171 Version 6.2 01-Aug-05
3172 ---------------------
3173
3174 1. There was no test for integer overflow of quantifier values. A construction
3175 such as {1111111111111111} would give undefined results. What is worse, if
3176 a minimum quantifier for a parenthesized subpattern overflowed and became
3177 negative, the calculation of the memory size went wrong. This could have
3178 led to memory overwriting.
3179
3180 2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
3181
3182 3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
3183 operating environments where this matters.
3184
3185 4. Applied Giuseppe Maxia's patch to add additional features for controlling
3186 PCRE options from within the C++ wrapper.
3187
3188 5. Named capturing subpatterns were not being correctly counted when a pattern
3189 was compiled. This caused two problems: (a) If there were more than 100
3190 such subpatterns, the calculation of the memory needed for the whole
3191 compiled pattern went wrong, leading to an overflow error. (b) Numerical
3192 back references of the form \12, where the number was greater than 9, were
3193 not recognized as back references, even though there were sufficient
3194 previous subpatterns.
3195
3196 6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
3197 versions of gcc, e.g. 2.95.4.
3198
3199
3200 Version 6.1 21-Jun-05
3201 ---------------------
3202
3203 1. There was one reference to the variable "posix" in pcretest.c that was not
3204 surrounded by "#if !defined NOPOSIX".
3205
3206 2. Make it possible to compile pcretest without DFA support, UTF8 support, or
3207 the cross-check on the old pcre_info() function, for the benefit of the
3208 cut-down version of PCRE that is currently imported into Exim.
3209
3210 3. A (silly) pattern starting with (?i)(?-i) caused an internal space
3211 allocation error. I've done the easy fix, which wastes 2 bytes for sensible
3212 patterns that start (?i) but I don't think that matters. The use of (?i) is
3213 just an example; this all applies to the other options as well.
3214
3215 4. Since libtool seems to echo the compile commands it is issuing, the output
3216 from "make" can be reduced a bit by putting "@" in front of each libtool
3217 compile command.
3218
3219 5. Patch from the folks at Google for configure.in to be a bit more thorough
3220 in checking for a suitable C++ installation before trying to compile the
3221 C++ stuff. This should fix a reported problem when a compiler was present,
3222 but no suitable headers.
3223
3224 6. The man pages all had just "PCRE" as their title. I have changed them to
3225 be the relevant file name. I have also arranged that these names are
3226 retained in the file doc/pcre.txt, which is a concatenation in text format
3227 of all the man pages except the little individual ones for each function.
3228
3229 7. The NON-UNIX-USE file had not been updated for the different set of source
3230 files that come with release 6. I also added a few comments about the C++
3231 wrapper.
3232
3233
3234 Version 6.0 07-Jun-05
3235 ---------------------
3236
3237 1. Some minor internal re-organization to help with my DFA experiments.
3238
3239 2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
3240 didn't matter for the library itself when fully configured, but did matter
3241 when compiling without UCP support, or within Exim, where the ucp files are
3242 not imported.
3243
3244 3. Refactoring of the library code to split up the various functions into
3245 different source modules. The addition of the new DFA matching code (see
3246 below) to a single monolithic source would have made it really too
3247 unwieldy, quite apart from causing all the code to be include in a
3248 statically linked application, when only some functions are used. This is
3249 relevant even without the DFA addition now that patterns can be compiled in
3250 one application and matched in another.
3251
3252 The downside of splitting up is that there have to be some external
3253 functions and data tables that are used internally in different modules of
3254 the library but which are not part of the API. These have all had their
3255 names changed to start with "_pcre_" so that they are unlikely to clash
3256 with other external names.
3257
3258 4. Added an alternate matching function, pcre_dfa_exec(), which matches using
3259 a different (DFA) algorithm. Although it is slower than the original
3260 function, it does have some advantages for certain types of matching
3261 problem.
3262
3263 5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
3264 including restarting after a partial match.
3265
3266 6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
3267 defined when compiling for Windows was sent to me. I have put it into the
3268 code, though I have no means of testing or verifying it.
3269
3270 7. Added the pcre_refcount() auxiliary function.
3271
3272 8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
3273 match before or at the first newline in the subject string. In pcretest,
3274 the /f option on a pattern can be used to set this.
3275
3276 9. A repeated \w when used in UTF-8 mode with characters greater than 256
3277 would behave wrongly. This has been present in PCRE since release 4.0.
3278
3279 10. A number of changes to the pcregrep command:
3280
3281 (a) Refactored how -x works; insert ^(...)$ instead of setting
3282 PCRE_ANCHORED and checking the length, in preparation for adding
3283 something similar for -w.
3284
3285 (b) Added the -w (match as a word) option.
3286
3287 (c) Refactored the way lines are read and buffered so as to have more
3288 than one at a time available.
3289
3290 (d) Implemented a pcregrep test script.
3291
3292 (e) Added the -M (multiline match) option. This allows patterns to match
3293 over several lines of the subject. The buffering ensures that at least
3294 8K, or the rest of the document (whichever is the shorter) is available
3295 for matching (and similarly the previous 8K for lookbehind assertions).
3296
3297 (f) Changed the --help output so that it now says
3298
3299 -w, --word-regex(p)
3300
3301 instead of two lines, one with "regex" and the other with "regexp"
3302 because that confused at least one person since the short forms are the
3303 same. (This required a bit of code, as the output is generated
3304 automatically from a table. It wasn't just a text change.)
3305
3306 (g) -- can be used to terminate pcregrep options if the next thing isn't an
3307 option but starts with a hyphen. Could be a pattern or a path name
3308 starting with a hyphen, for instance.
3309
3310 (h) "-" can be given as a file name to represent stdin.
3311
3312 (i) When file names are being printed, "(standard input)" is used for
3313 the standard input, for compatibility with GNU grep. Previously
3314 "<stdin>" was used.
3315
3316 (j) The option --label=xxx can be used to supply a name to be used for
3317 stdin when file names are being printed. There is no short form.
3318
3319 (k) Re-factored the options decoding logic because we are going to add
3320 two more options that take data. Such options can now be given in four
3321 different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
3322
3323 (l) Added the -A, -B, and -C options for requesting that lines of context
3324 around matches be printed.
3325
3326 (m) Added the -L option to print the names of files that do not contain
3327 any matching lines, that is, the complement of -l.
3328
3329 (n) The return code is 2 if any file cannot be opened, but pcregrep does
3330 continue to scan other files.
3331
3332 (o) The -s option was incorrectly implemented. For compatibility with other
3333 greps, it now suppresses the error message for a non-existent or non-
3334 accessible file (but not the return code). There is a new option called
3335 -q that suppresses the output of matching lines, which was what -s was
3336 previously doing.
3337
3338 (p) Added --include and --exclude options to specify files for inclusion
3339 and exclusion when recursing.
3340
3341 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
3342 Hopefully, it now does.
3343
3344 12. Missing cast in pcre_study().
3345
3346 13. Added an "uninstall" target to the makefile.
3347
3348 14. Replaced "extern" in the function prototypes in Makefile.in with
3349 "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
3350 world, but is set differently for Windows.
3351
3352 15. Added a second compiling function called pcre_compile2(). The only
3353 difference is that it has an extra argument, which is a pointer to an
3354 integer error code. When there is a compile-time failure, this is set
3355 non-zero, in addition to the error test pointer being set to point to an
3356 error message. The new argument may be NULL if no error number is required
3357 (but then you may as well call pcre_compile(), which is now just a
3358 wrapper). This facility is provided because some applications need a
3359 numeric error indication, but it has also enabled me to tidy up the way
3360 compile-time errors are handled in the POSIX wrapper.
3361
3362 16. Added VPATH=.libs to the makefile; this should help when building with one
3363 prefix path and installing with another. (Or so I'm told by someone who
3364 knows more about this stuff than I do.)
3365
3366 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
3367 passes PCRE_DOTALL to the pcre_compile() function, making the "." character
3368 match everything, including newlines. This is not POSIX-compatible, but
3369 somebody wanted the feature. From pcretest it can be activated by using
3370 both the P and the s flags.
3371
3372 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
3373
3374 19. libpcre.pc was being incorrectly installed as executable.
3375
3376 20. A couple of places in pcretest check for end-of-line by looking for '\n';
3377 it now also looks for '\r' so that it will work unmodified on Windows.
3378
3379 21. Added Google's contributed C++ wrapper to the distribution.
3380
3381 22. Added some untidy missing memory free() calls in pcretest, to keep
3382 Electric Fence happy when testing.
3383
3384
3385
3386 Version 5.0 13-Sep-04
3387 ---------------------
3388
3389 1. Internal change: literal characters are no longer packed up into items
3390 containing multiple characters in a single byte-string. Each character
3391 is now matched using a separate opcode. However, there may be more than one
3392 byte in the character in UTF-8 mode.
3393
3394 2. The pcre_callout_block structure has two new fields: pattern_position and
3395 next_item_length. These contain the offset in the pattern to the next match
3396 item, and its length, respectively.
3397
3398 3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
3399 insertion of callouts before each pattern item. Added the /C option to
3400 pcretest to make use of this.
3401
3402 4. On the advice of a Windows user, the lines
3403
3404 #if defined(_WIN32) || defined(WIN32)
3405 _setmode( _fileno( stdout ), 0x8000 );
3406 #endif /* defined(_WIN32) || defined(WIN32) */
3407
3408 have been added to the source of pcretest. This apparently does useful
3409 magic in relation to line terminators.
3410
3411 5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
3412 for the benefit of those environments where the "b" makes a difference.
3413
3414 6. The icc compiler has the same options as gcc, but "configure" doesn't seem
3415 to know about it. I have put a hack into configure.in that adds in code
3416 to set GCC=yes if CC=icc. This seems to end up at a point in the
3417 generated configure script that is early enough to affect the setting of
3418 compiler options, which is what is needed, but I have no means of testing
3419 whether it really works. (The user who reported this had patched the
3420 generated configure script, which of course I cannot do.)
3421
3422 LATER: After change 22 below (new libtool files), the configure script
3423 seems to know about icc (and also ecc). Therefore, I have commented out
3424 this hack in configure.in.
3425
3426 7. Added support for pkg-config (2 patches were sent in).
3427
3428 8. Negated POSIX character classes that used a combination of internal tables
3429 were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
3430 [[:^ascii]]. Typically, they would match almost any characters. The other
3431 POSIX classes were not broken in this way.
3432
3433 9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
3434 to find the match, as PCRE was deluded into thinking that the match had to
3435 start at the start point or following a newline. The same bug applied to
3436 patterns with negative forward assertions or any backward assertions
3437 preceding ".*" at the start, unless the pattern required a fixed first
3438 character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
3439
3440 10. In UTF-8 mode, when moving forwards in the subject after a failed match
3441 starting at the last subject character, bytes beyond the end of the subject
3442 string were read.
3443
3444 11. Renamed the variable "class" as "classbits" to make life easier for C++
3445 users. (Previously there was a macro definition, but it apparently wasn't
3446 enough.)
3447
3448 12. Added the new field "tables" to the extra data so that tables can be passed
3449 in at exec time, or the internal tables can be re-selected. This allows
3450 a compiled regex to be saved and re-used at a later time by a different
3451 program that might have everything at different addresses.
3452
3453 13. Modified the pcre-config script so that, when run on Solaris, it shows a
3454 -R library as well as a -L library.
3455
3456 14. The debugging options of pcretest (-d on the command line or D on a
3457 pattern) showed incorrect output for anything following an extended class
3458 that contained multibyte characters and which was followed by a quantifier.
3459
3460 15. Added optional support for general category Unicode character properties
3461 via the \p, \P, and \X escapes. Unicode property support implies UTF-8
3462 support. It adds about 90K to the size of the library. The meanings of the
3463 inbuilt class escapes such as \d and \s have NOT been changed.
3464
3465 16. Updated pcredemo.c to include calls to free() to release the memory for the
3466 compiled pattern.
3467
3468 17. The generated file chartables.c was being created in the source directory
3469 instead of in the building directory. This caused the build to fail if the
3470 source directory was different from the building directory, and was
3471 read-only.
3472
3473 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
3474 file. No doubt somebody will tell me if they don't make sense... Also added
3475 Dan Mooney's comments about building on OpenVMS.
3476
3477 19. Added support for partial matching via the PCRE_PARTIAL option for
3478 pcre_exec() and the \P data escape in pcretest.
3479
3480 20. Extended pcretest with 3 new pattern features:
3481
3482 (i) A pattern option of the form ">rest-of-line" causes pcretest to
3483 write the compiled pattern to the file whose name is "rest-of-line".
3484 This is a straight binary dump of the data, with the saved pointer to
3485 the character tables forced to be NULL. The study data, if any, is
3486 written too. After writing, pcretest reads a new pattern.
3487
3488 (ii) If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
3489 compiled pattern from the given file. There must not be any
3490 occurrences of "<" in the file name (pretty unlikely); if there are,
3491 pcretest will instead treat the initial "<" as a pattern delimiter.
3492 After reading in the pattern, pcretest goes on to read data lines as
3493 usual.
3494
3495 (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
3496 and 16-bit fields in a compiled pattern, to simulate a pattern that
3497 was compiled on a host of opposite endianness.
3498
3499 21. The pcre-exec() function can now cope with patterns that were compiled on
3500 hosts of opposite endianness, with this restriction:
3501
3502 As for any compiled expression that is saved and used later, the tables
3503 pointer field cannot be preserved; the extra_data field in the arguments
3504 to pcre_exec() should be used to pass in a tables address if a value
3505 other than the default internal tables were used at compile time.
3506
3507 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
3508 now diagnosed as an error. Previously, most of the time, a negative number
3509 would have been treated as zero, but if in addition "ovector" was passed as
3510 NULL, a crash could occur.
3511
3512 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
3513 new versions from the libtool 1.5 distribution (the last one is a copy of
3514 a file called libtool.m4). This seems to have fixed the need to patch
3515 "configure" to support Darwin 1.3 (which I used to do). However, I still
3516 had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
3517 workstation).
3518
3519 24. Changed the PCRE licence to be the more standard "BSD" licence.
3520
3521
3522 Version 4.5 01-Dec-03
3523 ---------------------
3524
3525 1. There has been some re-arrangement of the code for the match() function so
3526 that it can be compiled in a version that does not call itself recursively.
3527 Instead, it keeps those local variables that need separate instances for
3528 each "recursion" in a frame on the heap, and gets/frees frames whenever it
3529 needs to "recurse". Keeping track of where control must go is done by means
3530 of setjmp/longjmp. The whole thing is implemented by a set of macros that
3531 hide most of the details from the main code, and operates only if
3532 NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
3533 "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
3534 operating.
3535
3536 To make it easier for callers to provide specially tailored get/free
3537 functions for this usage, two new functions, pcre_stack_malloc, and
3538 pcre_stack_free, are used. They are always called in strict stacking order,
3539 and the size of block requested is always the same.
3540
3541 The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
3542 PCRE has been compiled to use the stack or the heap for recursion. The
3543 -C option of pcretest uses this to show which version is compiled.
3544
3545 A new data escape \S, is added to pcretest; it causes the amounts of store
3546 obtained and freed by both kinds of malloc/free at match time to be added
3547 to the output.
3548
3549 2. Changed the locale test to use "fr_FR" instead of "fr" because that's
3550 what's available on my current Linux desktop machine.
3551
3552 3. When matching a UTF-8 string, the test for a valid string at the start has
3553 been extended. If start_offset is not zero, PCRE now checks that it points
3554 to a byte that is the start of a UTF-8 character. If not, it returns
3555 PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
3556 this is necessary because there may be backward assertions in the pattern.
3557 When matching the same subject several times, it may save resources to use
3558 PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
3559
3560 4. The code for checking the validity of UTF-8 strings has been tightened so
3561 that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
3562 containing "overlong sequences".
3563
3564 5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
3565 I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
3566 should have been "&", but it just so happened that all the cases this let
3567 through by mistake were picked up later in the function.
3568
3569 6. I had used a variable called "isblank" - this is a C99 function, causing
3570 some compilers to warn. To avoid this, I renamed it (as "blankclass").
3571
3572 7. Cosmetic: (a) only output another newline at the end of pcretest if it is
3573 prompting; (b) run "./pcretest /dev/null" at the start of the test script
3574 so the version is shown; (c) stop "make test" echoing "./RunTest".
3575
3576 8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
3577
3578 9. The prototype for memmove() for systems that don't have it was using
3579 size_t, but the inclusion of the header that defines size_t was later. I've
3580 moved the #includes for the C headers earlier to avoid this.
3581
3582 10. Added some adjustments to the code to make it easier to compiler on certain
3583 special systems:
3584
3585 (a) Some "const" qualifiers were missing.
3586 (b) Added the macro EXPORT before all exported functions; by default this
3587 is defined to be empty.
3588 (c) Changed the dftables auxiliary program (that builds chartables.c) so
3589 that it reads its output file name as an argument instead of writing
3590 to the standard output and assuming this can be redirected.
3591
3592 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
3593 class containing characters with values greater than 255, PCRE compilation
3594 went into a loop.
3595
3596 12. A recursive reference to a subpattern that was within another subpattern
3597 that had a minimum quantifier of zero caused PCRE to crash. For example,
3598 (x(y(?2))z)? provoked this bug with a subject that got as far as the
3599 recursion. If the recursively-called subpattern itself had a zero repeat,
3600 that was OK.
3601
3602 13. In pcretest, the buffer for reading a data line was set at 30K, but the
3603 buffer into which it was copied (for escape processing) was still set at
3604 1024, so long lines caused crashes.
3605
3606 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
3607 "internal error: code overflow...". This applied to any character class
3608 that was followed by a possessive quantifier.
3609
3610 15. Modified the Makefile to add libpcre.la as a prerequisite for
3611 libpcreposix.la because I was told this is needed for a parallel build to
3612 work.
3613
3614 16. If a pattern that contained .* following optional items at the start was
3615 studied, the wrong optimizing data was generated, leading to matching
3616 errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
3617 matching string must start with a or b or c. The correct conclusion for
3618 this pattern is that a match can start with any character.
3619
3620
3621 Version 4.4 13-Aug-03
3622 ---------------------
3623
3624 1. In UTF-8 mode, a character class containing characters with values between
3625 127 and 255 was not handled correctly if the compiled pattern was studied.
3626 In fixing this, I have also improved the studying algorithm for such
3627 classes (slightly).
3628
3629 2. Three internal functions had redundant arguments passed to them. Removal
3630 might give a very teeny performance improvement.
3631
3632 3. Documentation bug: the value of the capture_top field in a callout is *one
3633 more than* the number of the hightest numbered captured substring.
3634
3635 4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
3636 in incorrectly linking with a previously installed version. They now link
3637 explicitly with libpcre.la.
3638
3639 5. configure.in no longer needs to recognize Cygwin specially.
3640
3641 6. A problem in pcre.in for Windows platforms is fixed.
3642
3643 7. If a pattern was successfully studied, and the -d (or /D) flag was given to
3644 pcretest, it used to include the size of the study block as part of its
3645 output. Unfortunately, the structure contains a field that has a different
3646 size on different hardware architectures. This meant that the tests that
3647 showed this size failed. As the block is currently always of a fixed size,
3648 this information isn't actually particularly useful in pcretest output, so
3649 I have just removed it.
3650
3651 8. Three pre-processor statements accidentally did not start in column 1.
3652 Sadly, there are *still* compilers around that complain, even though
3653 standard C has not required this for well over a decade. Sigh.
3654
3655 9. In pcretest, the code for checking callouts passed small integers in the
3656 callout_data field, which is a void * field. However, some picky compilers
3657 complained about the casts involved for this on 64-bit systems. Now
3658 pcretest passes the address of the small integer instead, which should get
3659 rid of the warnings.
3660
3661 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
3662 both compile and run time, and gives an error if an invalid UTF-8 sequence
3663 is found. There is a option for disabling this check in cases where the
3664 string is known to be correct and/or the maximum performance is wanted.
3665
3666 11. In response to a bug report, I changed one line in Makefile.in from
3667
3668 -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
3669 to
3670 -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
3671
3672 to look similar to other lines, but I have no way of telling whether this
3673 is the right thing to do, as I do not use Windows. No doubt I'll get told
3674 if it's wrong...
3675
3676
3677 Version 4.3 21-May-03
3678 ---------------------
3679
3680 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
3681 Makefile.
3682
3683 2. Some refactoring to improve the quality of the code:
3684
3685 (i) The utf8_table... variables are now declared "const".
3686
3687 (ii) The code for \cx, which used the "case flipping" table to upper case
3688 lower case letters, now just substracts 32. This is ASCII-specific,
3689 but the whole concept of \cx is ASCII-specific, so it seems
3690 reasonable.
3691
3692 (iii) PCRE was using its character types table to recognize decimal and
3693 hexadecimal digits in the pattern. This is silly, because it handles
3694 only 0-9, a-f, and A-F, but the character types table is locale-
3695 specific, which means strange things might happen. A private
3696 table is now used for this - though it costs 256 bytes, a table is
3697 much faster than multiple explicit tests. Of course, the standard
3698 character types table is still used for matching digits in subject
3699 strings against \d.
3700
3701 (iv) Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
3702 ending in _t are). So I've renamed it as ESC_tee.
3703
3704 3. The first argument for regexec() in the POSIX wrapper should have been
3705 defined as "const".
3706
3707 4. Changed pcretest to use malloc() for its buffers so that they can be
3708 Electric Fenced for debugging.
3709
3710 5. There were several places in the code where, in UTF-8 mode, PCRE would try
3711 to read one or more bytes before the start of the subject string. Often this
3712 had no effect on PCRE's behaviour, but in some circumstances it could
3713 provoke a segmentation fault.
3714
3715 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
3716 to try to read one or more bytes before the start of the subject string.
3717
3718 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
3719 UTF-8 support could misbehave in various ways if the subject string
3720 contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
3721 area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
3722 back over UTF-8 characters.)
3723
3724
3725 Version 4.2 14-Apr-03
3726 ---------------------
3727
3728 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
3729
3730 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
3731 [ON_WINDOWS]: new variable, "#" on non-Windows platforms
3732 [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
3733 [WIN_PREFIX]: new variable, "cyg" for Cygwin
3734 * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
3735 and BUILD_EXEEXT
3736 Note: automatic setting of the BUILD variables is not yet working
3737 set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
3738 compile-time but not at link-time
3739 [LINK]: use for linking executables only
3740 make different versions for Windows and non-Windows
3741 [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
3742 libraries
3743 [LINK_FOR_BUILD]: new variable
3744 [OBJEXT]: use throughout
3745 [EXEEXT]: use throughout
3746 <winshared>: new target
3747 <wininstall>: new target
3748 <dftables.o>: use native compiler
3749 <dftables>: use native linker
3750 <install>: handle Windows platform correctly
3751 <clean>: ditto
3752 <check>: ditto
3753 copy DLL to top builddir before testing
3754
3755 As part of these changes, -no-undefined was removed again. This was reported
3756 to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
3757 in any case.
3758
3759 3. Some tidies to get rid of compiler warnings:
3760
3761 . In the match_data structure, match_limit was an unsigned long int, whereas
3762 match_call_count was an int. I've made them both unsigned long ints.
3763
3764 . In pcretest the fact that a const uschar * doesn't automatically cast to
3765 a void * provoked a warning.
3766
3767 . Turning on some more compiler warnings threw up some "shadow" variables
3768 and a few more missing casts.
3769
3770 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
3771 option, a class that contained a single character with a value between 128
3772 and 255 (e.g. /[\xFF]/) caused PCRE to crash.
3773
3774 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
3775 option, a class that contained several characters, but with at least one
3776 whose value was between 128 and 255 caused PCRE to crash.
3777
3778
3779 Version 4.1 12-Mar-03
3780 ---------------------
3781
3782 1. Compiling with gcc -pedantic found a couple of places where casts were
3783 needed, and a string in dftables.c that was longer than standard compilers are
3784 required to support.
3785
3786 2. Compiling with Sun's compiler found a few more places where the code could
3787 be tidied up in order to avoid warnings.
3788
3789 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
3790 first of these names is deprecated in the latest Autoconf in favour of the name
3791 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
3792 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
3793 analogy I have changed it to CFLAGS_FOR_BUILD.
3794
3795 4. Added -no-undefined to the linking command in the Makefile, because this is
3796 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
3797 linking step for the pcreposix library.
3798
3799 5. PCRE was failing to diagnose the case of two named groups with the same
3800 name.
3801
3802 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
3803 literal character that is needed in the subject for a match, and scans along to
3804 ensure that it is present before embarking on the full matching process. This
3805 saves time in cases of nested unlimited repeats that are never going to match.
3806 Problem: the scan can take a lot of time if the subject is very long (e.g.
3807 megabytes), thus penalizing straightforward matches. It is now done only if the
3808 amount of subject to be scanned is less than 1000 bytes.
3809
3810 7. A lesser problem with the same optimization is that it was recording the
3811 first character of an anchored pattern as "needed", thus provoking a search
3812 right along the subject, even when the first match of the pattern was going to
3813 fail. The "needed" character is now not set for anchored patterns, unless it
3814 follows something in the pattern that is of non-fixed length. Thus, it still
3815 fulfils its original purpose of finding quick non-matches in cases of nested
3816 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
3817
3818
3819 Version 4.0 17-Feb-03
3820 ---------------------
3821
3822 1. If a comment in an extended regex that started immediately after a meta-item
3823 extended to the end of string, PCRE compiled incorrect data. This could lead to
3824 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
3825
3826 2. Moved to autoconf 2.53 and libtool 1.4.2.
3827
3828 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
3829 the special perltest8 script is no longer needed - all the tests can be run
3830 from a single perltest script.
3831
3832 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
3833 by \s. It has now been removed in PCRE. This means it isn't recognized as
3834 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
3835 class [:space:] *does* include VT, thereby creating a mess.
3836
3837 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
3838 space and tab.
3839
3840 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
3841 its new features into the main test script, reducing the number of scripts.
3842
3843 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
3844 were backward compatible, and made the (?i) apply to the whole pattern, as if
3845 /i were given. Now it behaves more logically, and applies the option setting
3846 only to what follows. PCRE has been changed to follow suit. However, if it
3847 finds options settings right at the start of the pattern, it extracts them into
3848 the global options, as before. Thus, they show up in the info data.
3849
3850 8. Added support for the \Q...\E escape sequence. Characters in between are
3851 treated as literals. This is slightly different from Perl in that $ and @ are
3852 also handled as literals inside the quotes. In Perl, they will cause variable
3853 interpolation. Note the following examples:
3854
3855 Pattern PCRE matches Perl matches
3856
3857 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
3858 \Qabc\$xyz\E abc\$xyz abc\$xyz
3859 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
3860
3861 For compatibility with Perl, \Q...\E sequences are recognized inside character
3862 classes as well as outside them.
3863
3864 9. Re-organized 3 code statements in pcretest to avoid "overflow in
3865 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
3866 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
3867 signed/unsigned warnings.
3868
3869 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
3870 option for pcretest, so I've replaced it by a simple function that does just
3871 that job.
3872
3873 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
3874 "pcregrep -".
3875
3876 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
3877 Java package. This provides some syntactic sugar for simple cases of what my
3878 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
3879 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
3880 item, you can use this simplified notation. Note that only makes sense with
3881 greedy quantifiers. Consequently, the use of the possessive quantifier forces
3882 greediness, whatever the setting of the PCRE_UNGREEDY option.
3883
3884 13. A change of greediness default within a pattern was not taking effect at
3885 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
3886 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
3887 was abstracted outside.
3888
3889 14. PCRE now supports the \G assertion. It is true when the current matching
3890 position is at the start point of the match. This differs from \A when the
3891 starting offset is non-zero. Used with the /g option of pcretest (or similar
3892 code), it works in the same way as it does for Perl's /g option. If all
3893 alternatives of a regex begin with \G, the expression is anchored to the start
3894 match position, and the "anchored" flag is set in the compiled expression.
3895
3896 15. Some bugs concerning the handling of certain option changes within patterns
3897 have been fixed. These applied to options other than (?ims). For example,
3898 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
3899 been the other way round. Some of this was related to change 7 above.
3900
3901 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
3902 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
3903 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
3904 POSIX classes only within a class (e.g. /[[:alpha:]]/).
3905
3906 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
3907 mode. Unlike ".", it always matches newline, whatever the setting of
3908 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
3909 assertions. Perl allows it, but it doesn't (in general) work because it can't
3910 calculate the length of the lookbehind. At least, that's the case for Perl
3911 5.8.0 - I've been told they are going to document that it doesn't work in
3912 future.
3913
3914 18. Added an error diagnosis for escapes that PCRE does not support: these are
3915 \L, \l, \N, \P, \p, \U, \u, and \X.
3916
3917 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
3918 reading past the end of the pattern in cases such as /[abcd/.
3919
3920 20. PCRE was getting more memory than necessary for patterns with classes that
3921 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
3922
3923 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
3924 compiling PCRE for use with Virtual Pascal.
3925
3926 22. Small fix to the Makefile to make it work properly if the build is done
3927 outside the source tree.
3928
3929 23. Added a new extension: a condition to go with recursion. If a conditional
3930 subpattern starts with (?(R) the "true" branch is used if recursion has
3931 happened, whereas the "false" branch is used only at the top level.
3932
3933 24. When there was a very long string of literal characters (over 255 bytes
3934 without UTF support, over 250 bytes with UTF support), the computation of how
3935 much memory was required could be incorrect, leading to segfaults or other
3936 strange effects.
3937
3938 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
3939 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
3940 there was a subsequent back reference to those brackets. This meant that, for
3941 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
3942 possible to check for precisely this case. All we can do is abandon the
3943 optimization if .* occurs inside capturing brackets when there are any back
3944 references whatsoever. (See below for a better fix that came later.)
3945
3946 26. The handling of the optimization for finding the first character of a
3947 non-anchored pattern, and for finding a character that is required later in the
3948 match were failing in some cases. This didn't break the matching; it just
3949 failed to optimize when it could. The way this is done has been re-implemented.
3950
3951 27. Fixed typo in error message for invalid (?R item (it said "(?p").
3952
3953 28. Added a new feature that provides some of the functionality that Perl
3954 provides with (?{...}). The facility is termed a "callout". The way it is done
3955 in PCRE is for the caller to provide an optional function, by setting
3956 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
3957 global variable. By default it is unset, which disables all calling out. To get
3958 the function called, the regex must include (?C) at appropriate points. This
3959 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
3960 This provides a means of identifying different callout points. When PCRE
3961 reaches such a point in the regex, if pcre_callout has been set, the external
3962 function is called. It is provided with data in a structure called
3963 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
3964 matching continues; if it returns a non-zero value, the match at the current
3965 point fails. However, backtracking will occur if possible. [This was changed
3966 later and other features added - see item 49 below.]
3967
3968 29. pcretest is upgraded to test the callout functionality. It provides a
3969 callout function that displays information. By default, it shows the start of
3970 the match and the current position in the text. There are some new data escapes
3971 to vary what happens:
3972
3973 \C+ in addition, show current contents of captured substrings
3974 \C- do not supply a callout function
3975 \C!n return 1 when callout number n is reached
3976 \C!n!m return 1 when callout number n is reached for the mth time
3977
3978 30. If pcregrep was called with the -l option and just a single file name, it
3979 output "<stdin>" if a match was found, instead of the file name.
3980
3981 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
3982 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
3983 pcre_exec(). This saves a malloc/free per call. The default value of
3984 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
3985 when configuring.
3986
3987 32. The default maximum size of a compiled pattern is 64K. There have been a
3988 few cases of people hitting this limit. The code now uses macros to handle the
3989 storing of links as offsets within the compiled pattern. It defaults to 2-byte
3990 links, but this can be changed to 3 or 4 bytes by --with-link-size when
3991 configuring. Tests 2 and 5 work only with 2-byte links because they output
3992 debugging information about compiled patterns.
3993
3994 33. Internal code re-arrangements:
3995
3996 (a) Moved the debugging function for printing out a compiled regex into
3997 its own source file (printint.c) and used #include to pull it into
3998 pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
3999 separate copies.
4000
4001 (b) Defined the list of op-code names for debugging as a macro in
4002 internal.h so that it is next to the definition of the opcodes.
4003
4004 (c) Defined a table of op-code lengths for simpler skipping along compiled
4005 code. This is again a macro in internal.h so that it is next to the
4006 definition of the opcodes.
4007
4008 34. Added support for recursive calls to individual subpatterns, along the
4009 lines of Robin Houston's patch (but implemented somewhat differently).
4010
4011 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
4012 allow it to read and process whole directories in Win32. This code was
4013 contributed by Lionel Fourquaux; it has not been tested by me.
4014
4015 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
4016 used to name a group. Names consist of alphanumerics and underscores, and must
4017 be unique. Back references use the syntax (?P=name) and recursive calls use
4018 (?P>name) which is a PCRE extension to the Python extension. Groups still have
4019 numbers. The function pcre_fullinfo() can be used after compilation to extract
4020 a name/number map. There are three relevant calls:
4021
4022 PCRE_INFO_NAMEENTRYSIZE yields the size of each entry in the map
4023 PCRE_INFO_NAMECOUNT yields the number of entries
4024 PCRE_INFO_NAMETABLE yields a pointer to the map.
4025
4026 The map is a vector of fixed-size entries. The size of each entry depends on
4027 the length of the longest name used. The first two bytes of each entry are the
4028 group number, most significant byte first. There follows the corresponding
4029 name, zero terminated. The names are in alphabetical order.
4030
4031 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
4032 case instead of 255. Making it the same both with and without UTF-8 support
4033 means that the same test output works with both.
4034
4035 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
4036 calling malloc() with a zero argument.
4037
4038 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
4039 optimization. I've improved things by keeping a bitmap of backreferences with
4040 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
4041 fact referenced, the optimization can be applied. It is unlikely that a
4042 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
4043 the match to follow \n) will appear inside brackets with a number greater than
4044 31, but if it does, any back reference > 31 suppresses the optimization.
4045
4046 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
4047 of disabling numbered capturing parentheses. Any opening parenthesis that is
4048 not followed by ? behaves as if it were followed by ?: but named parentheses
4049 can still be used for capturing (and they will acquire numbers in the usual
4050 way).
4051
4052 41. Redesigned the return codes from the match() function into yes/no/error so
4053 that errors can be passed back from deep inside the nested calls. A malloc
4054 failure while inside a recursive subpattern call now causes the
4055 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
4056
4057 42. It is now possible to set a limit on the number of times the match()
4058 function is called in a call to pcre_exec(). This facility makes it possible to
4059 limit the amount of recursion and backtracking, though not in a directly
4060 obvious way, because the match() function is used in a number of different
4061 circumstances. The count starts from zero for each position in the subject
4062 string (for non-anchored patterns). The default limit is, for compatibility, a
4063 large number, namely 10 000 000. You can change this in two ways:
4064
4065 (a) When configuring PCRE before making, you can use --with-match-limit=n
4066 to set a default value for the compiled library.
4067
4068 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
4069 a different value is set. See 45 below.
4070
4071 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
4072
4073 43. Added a new function pcre_config(int, void *) to enable run-time extraction
4074 of things that can be changed at compile time. The first argument specifies
4075 what is wanted and the second points to where the information is to be placed.
4076 The current list of available information is:
4077
4078 PCRE_CONFIG_UTF8
4079
4080 The output is an integer that is set to one if UTF-8 support is available;
4081 otherwise it is set to zero.
4082
4083 PCRE_CONFIG_NEWLINE
4084
4085 The output is an integer that it set to the value of the code that is used for
4086 newline. It is either LF (10) or CR (13).
4087
4088 PCRE_CONFIG_LINK_SIZE
4089
4090 The output is an integer that contains the number of bytes used for internal
4091 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
4092
4093 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
4094
4095 The output is an integer that contains the threshold above which the POSIX
4096 interface uses malloc() for output vectors. See item 31 above.
4097
4098 PCRE_CONFIG_MATCH_LIMIT
4099
4100 The output is an unsigned integer that contains the default limit of the number
4101 of match() calls in a pcre_exec() execution. See 42 above.
4102
4103 44. pcretest has been upgraded by the addition of the -C option. This causes it
4104 to extract all the available output from the new pcre_config() function, and to
4105 output it. The program then exits immediately.
4106
4107 45. A need has arisen to pass over additional data with calls to pcre_exec() in
4108 order to support additional features. One way would have been to define
4109 pcre_exec2() (for example) with extra arguments, but this would not have been
4110 extensible, and would also have required all calls to the original function to
4111 be mapped to the new one. Instead, I have chosen to extend the mechanism that
4112 is used for passing in "extra" data from pcre_study().
4113
4114 The pcre_extra structure is now exposed and defined in pcre.h. It currently
4115 contains the following fields:
4116
4117 flags a bitmap indicating which of the following fields are set
4118 study_data opaque data from pcre_study()
4119 match_limit a way of specifying a limit on match() calls for a specific
4120 call to pcre_exec()
4121 callout_data data for callouts (see 49 below)
4122
4123 The flag bits are also defined in pcre.h, and are
4124
4125 PCRE_EXTRA_STUDY_DATA
4126 PCRE_EXTRA_MATCH_LIMIT
4127 PCRE_EXTRA_CALLOUT_DATA
4128
4129 The pcre_study() function now returns one of these new pcre_extra blocks, with
4130 the actual study data pointed to by the study_data field, and the
4131 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
4132 before. That is, this change is entirely upwards-compatible and requires no
4133 change to existing code.
4134
4135 If you want to pass in additional data to pcre_exec(), you can either place it
4136 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
4137 block.
4138
4139 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
4140 data string contains the escape sequence \M, pcretest calls pcre_exec() several
4141 times with different match limits, until it finds the minimum value needed for
4142 pcre_exec() to complete. The value is then output. This can be instructive; for
4143 most simple matches the number is quite small, but for pathological cases it
4144 gets very large very quickly.
4145
4146 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
4147 returns the size of the data block pointed to by the study_data field in a
4148 pcre_extra block, that is, the value that was passed as the argument to
4149 pcre_malloc() when PCRE was getting memory in which to place the information
4150 created by pcre_study(). The fourth argument should point to a size_t variable.
4151 pcretest has been extended so that this information is shown after a successful
4152 pcre_study() call when information about the compiled regex is being displayed.
4153
4154 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
4155 because what follows is always an absolute path. (Later: it turns out that this
4156 is more than cosmetic for MinGW, because it doesn't like empty path
4157 components.)
4158
4159 49. Some changes have been made to the callout feature (see 28 above):
4160
4161 (i) A callout function now has three choices for what it returns:
4162
4163 0 => success, carry on matching
4164 > 0 => failure at this point, but backtrack if possible
4165 < 0 => serious error, return this value from pcre_exec()
4166
4167 Negative values should normally be chosen from the set of PCRE_ERROR_xxx
4168 values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
4169 "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
4170 use by callout functions. It will never be used by PCRE itself.
4171
4172 (ii) The pcre_extra structure (see 45 above) has a void * field called
4173 callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
4174 pcre_callout_block structure has a field of the same name. The contents of
4175 the field passed in the pcre_extra structure are passed to the callout
4176 function in the corresponding field in the callout block. This makes it
4177 easier to use the same callout-containing regex from multiple threads. For
4178 testing, the pcretest program has a new data escape
4179
4180 \C*n pass the number n (may be negative) as callout_data
4181
4182 If the callout function in pcretest receives a non-zero value as
4183 callout_data, it returns that value.
4184
4185 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
4186 there were some redundant $(CFLAGS) in commands that are now specified as
4187 $(LINK), which already includes $(CFLAGS).
4188
4189 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
4190 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
4191 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
4192 one-byte characters throughout. Note that case-insensitive matching applies
4193 only to characters whose values are less than 256. PCRE doesn't support the
4194 notion of cases for higher-valued characters.
4195
4196 (i) A character class whose characters are all within 0-255 is handled as
4197 a bit map, and the map is inverted for negative classes. Previously, a
4198 character > 255 always failed to match such a class; however it should
4199 match if the class was a negative one (e.g. [^ab]). This has been fixed.
4200
4201 (ii) A negated character class with a single character < 255 is coded as
4202 "not this character" (OP_NOT). This wasn't working properly when the test
4203 character was multibyte, either singly or repeated.
4204
4205 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
4206 mode, for example: \x{100}{2,3}.
4207
4208 (iv) The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
4209 singly or repeated) now correctly test multibyte characters. However,
4210 PCRE doesn't recognize any characters with values greater than 255 as
4211 digits, spaces, or word characters. Such characters always match \D, \S,
4212 and \W, and never match \d, \s, or \w.
4213
4214 (v) Classes may now contain characters and character ranges with values
4215 greater than 255. For example: [ab\x{100}-\x{400}].
4216
4217 (vi) pcregrep now has a --utf-8 option (synonym -u) which makes it call
4218 PCRE in UTF-8 mode.
4219
4220 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
4221 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
4222 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
4223 value.)
4224
4225 53. The single man page has become too large. I have therefore split it up into
4226 a number of separate man pages. These also give rise to individual HTML pages;
4227 these are now put in a separate directory, and there is an index.html page that
4228 lists them all. Some hyperlinking between the pages has been installed.
4229
4230 54. Added convenience functions for handling named capturing parentheses.
4231
4232 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
4233 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
4234 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
4235 are faulted.
4236
4237 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
4238 calling configure. These values are used when compiling the dftables.c program
4239 which is run to generate the source of the default character tables. They
4240 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
4241 you will need to set these values.
4242
4243 57. Updated the building process for Windows DLL, as provided by Fred Cox.
4244
4245
4246 Version 3.9 02-Jan-02
4247 ---------------------
4248
4249 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
4250
4251 2. If --disable-static was given, the building process failed when trying to
4252 build pcretest and pcregrep. (For some reason it was using libtool to compile
4253 them, which is not right, as they aren't part of the library.)
4254
4255
4256 Version 3.8 18-Dec-01
4257 ---------------------
4258
4259 1. The experimental UTF-8 code was completely screwed up. It was packing the
4260 bytes in the wrong order. How dumb can you get?
4261
4262
4263 Version 3.7 29-Oct-01
4264 ---------------------
4265
4266 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
4267 This caused pcretest, when used on the test data, to segfault. Unfortunately,
4268 this didn't happen under Solaris 8, where I normally test things.
4269
4270 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
4271 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
4272 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
4273 this fix an hour or so after the initial 3.7 release.)
4274
4275
4276 Version 3.6 23-Oct-01
4277 ---------------------
4278
4279 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
4280 offsets passed as NULL with zero offset count.
4281
4282 2. The config.guess and config.sub files had not been updated when I moved to
4283 the latest autoconf.
4284
4285
4286 Version 3.5 15-Aug-01
4287 ---------------------
4288
4289 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
4290 had been forgotten.
4291
4292 2. By using declared but undefined structures, we can avoid using "void"
4293 definitions in pcre.h while keeping the internal definitions of the structures
4294 private.
4295
4296 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
4297 user point of view, this means that both static and shared libraries are built
4298 by default, but this can be individually controlled. More of the work of
4299 handling this static/shared cases is now inside libtool instead of PCRE's make
4300 file.
4301
4302 4. The pcretest utility is now installed along with pcregrep because it is
4303 useful for users (to test regexs) and by doing this, it automatically gets
4304 relinked by libtool. The documentation has been turned into a man page, so
4305 there are now .1, .txt, and .html versions in /doc.
4306
4307 5. Upgrades to pcregrep:
4308 (i) Added long-form option names like gnu grep.
4309 (ii) Added --help to list all options with an explanatory phrase.
4310 (iii) Added -r, --recursive to recurse into sub-directories.
4311 (iv) Added -f, --file to read patterns from a file.
4312
4313 6. pcre_exec() was referring to its "code" argument before testing that
4314 argument for NULL (and giving an error if it was NULL).
4315
4316 7. Upgraded Makefile.in to allow for compiling in a different directory from
4317 the source directory.
4318
4319 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
4320 options bits, the pointer it was passed was to an int instead of to an unsigned
4321 long int. This mattered only on 64-bit systems.
4322
4323 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
4324 generated) instead of pcre.in, which it its source. Also made the same change
4325 in several of the .c files.
4326
4327 10. A new release of gcc defines printf() as a macro, which broke pcretest
4328 because it had an ifdef in the middle of a string argument for printf(). Fixed
4329 by using separate calls to printf().
4330
4331 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
4332 script, to force use of CR or LF instead of \n in the source. On non-Unix
4333 systems, the value can be set in config.h.
4334
4335 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
4336 absolute limit. Changed the text of the error message to make this clear, and
4337 likewise updated the man page.
4338
4339 13. The limit of 99 on the number of capturing subpatterns has been removed.
4340 The new limit is 65535, which I hope will not be a "real" limit.
4341
4342
4343 Version 3.4 22-Aug-00
4344 ---------------------
4345
4346 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
4347
4348 2. Diagnose condition (?(0) as an error instead of crashing on matching.
4349
4350
4351 Version 3.3 01-Aug-00
4352 ---------------------
4353
4354 1. If an octal character was given, but the value was greater than \377, it
4355 was not getting masked to the least significant bits, as documented. This could
4356 lead to crashes in some systems.
4357
4358 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
4359 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
4360
4361 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
4362 These just pass their arguments on to (pcre_free)(), but they are provided
4363 because some uses of PCRE bind it to non-C systems that can call its functions,
4364 but cannot call free() or pcre_free() directly.
4365
4366 4. Add "make test" as a synonym for "make check". Corrected some comments in
4367 the Makefile.
4368
4369 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
4370 Makefile.
4371
4372 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
4373 command called pgrep for grepping around the active processes.
4374
4375 7. Added the beginnings of support for UTF-8 character strings.
4376
4377 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
4378 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
4379 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
4380 out for the ar command.)
4381
4382
4383 Version 3.2 12-May-00
4384 ---------------------
4385
4386 This is purely a bug fixing release.
4387
4388 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
4389 of ZA. This was just one example of several cases that could provoke this bug,
4390 which was introduced by change 9 of version 2.00. The code for breaking
4391 infinite loops after an iteration that matches an empty string was't working
4392 correctly.
4393
4394 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
4395 when matched against abbab (for example). After matching an empty string, it
4396 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
4397 caused it to match further down the string than it should.
4398
4399 3. The code contained an inclusion of sys/types.h. It isn't clear why this
4400 was there because it doesn't seem to be needed, and it causes trouble on some
4401 systems, as it is not a Standard C header. It has been removed.
4402
4403 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
4404 were reported on the Macintosh. The changes were from
4405
4406 while ((c = *(++ptr)) != 0 && c != '\n');
4407 to
4408 while ((c = *(++ptr)) != 0 && c != '\n') ;
4409
4410 Totally extraordinary, but if that's what it takes...
4411
4412 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
4413 available. Added HAVE_BCOPY and an autoconf test for it; if neither
4414 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
4415 assumes the way PCRE uses memmove() (always moving upwards).
4416
4417 6. PCRE is being used in one environment where strchr() is not available. There
4418 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
4419 faster code anyway.
4420
4421
4422 Version 3.1 09-Feb-00
4423 ---------------------
4424
4425 The only change in this release is the fixing of some bugs in Makefile.in for
4426 the "install" target:
4427
4428 (1) It was failing to install pcreposix.h.
4429
4430 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
4431
4432
4433 Version 3.0 01-Feb-00
4434 ---------------------
4435
4436 1. Add support for the /+ modifier to perltest (to output $` like it does in
4437 pcretest).
4438
4439 2. Add support for the /g modifier to perltest.
4440
4441 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
4442 matches null strings.
4443
4444 4. Fix perltest so that it doesn't do unwanted things when fed an empty
4445 pattern. Perl treats empty patterns specially - it reuses the most recent
4446 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
4447 effect.
4448
4449 5. The POSIX interface was broken in that it was just handing over the POSIX
4450 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
4451 required a bigger vector, with some working space on the end. This means that
4452 the POSIX wrapper now has to get and free some memory, and copy the results.
4453
4454 6. Added some simple autoconf support, placing the test data and the
4455 documentation in separate directories, re-organizing some of the
4456 information files, and making it build pcre-config (a GNU standard). Also added
4457 libtool support for building PCRE as a shared library, which is now the
4458 default.
4459
4460 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
4461 09 are not valid octal constants. Single digits will be used for minor values
4462 less than 10.
4463
4464 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
4465 existing programs that set these in the POSIX interface can use PCRE without
4466 modification.
4467
4468 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
4469 return all that pcre_info() returns, plus additional data. The pcre_info()
4470 function is retained for compatibility, but is considered to be obsolete.
4471
4472 10. Added experimental recursion feature (?R) to handle one common case that
4473 Perl 5.6 will be able to do with (?p{...}).
4474
4475 11. Added support for POSIX character classes like [:alpha:], which Perl is
4476 adopting.
4477
4478
4479 Version 2.08 31-Aug-99
4480 ----------------------
4481
4482 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
4483 trying to match at the startoffset position, but instead was moving forward to
4484 the next newline as if a previous match had failed.
4485
4486 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
4487 and could get into a loop if a null string was matched other than at the start
4488 of the subject.
4489
4490 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
4491 be distinguished at compile time, and for completeness also added PCRE_DATE.
4492
4493 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
4494 in GnuWin32 environments.
4495
4496
4497 Version 2.07 29-Jul-99
4498 ----------------------
4499
4500 1. The documentation is now supplied in plain text form and HTML as well as in
4501 the form of man page sources.
4502
4503 2. C++ compilers don't like assigning (void *) values to other pointer types.
4504 In particular this affects malloc(). Although there is no problem in Standard
4505 C, I've put in casts to keep C++ compilers happy.
4506
4507 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
4508 should be (const char *).
4509
4510 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
4511 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
4512 However, I haven't made this a standard facility. The documentation doesn't
4513 mention it, and the Makefile doesn't support it.
4514
4515 5. The Makefile now contains an "install" target, with editable destinations at
4516 the top of the file. The pcretest program is not installed.
4517
4518 6. pgrep -V now gives the PCRE version number and date.
4519
4520 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
4521 causing the entire string to be ignored, instead of just the last character.
4522
4523 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
4524 non-matching string, it can take a very, very long time, even for strings of
4525 quite modest length, because of the nested recursion. PCRE now does better in
4526 some of these cases. It does this by remembering the last required literal
4527 character in the pattern, and pre-searching the subject to ensure it is present
4528 before running the real match. In other words, it applies a heuristic to detect
4529 some types of certain failure quickly, and in the above example, if presented
4530 with a string that has no trailing " it gives "no match" very quickly.
4531
4532 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
4533 other alternatives are tried instead.
4534
4535
4536 Version 2.06 09-Jun-99
4537 ----------------------
4538
4539 1. Change pcretest's output for amount of store used to show just the code
4540 space, because the remainder (the data block) varies in size between 32-bit and
4541 64-bit systems.
4542
4543 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
4544 start matching at. This allows lookbehinds to work when searching for multiple
4545 occurrences in a string.
4546
4547 3. Added additional options to pcretest for testing multiple occurrences:
4548
4549 /+ outputs the rest of the string that follows a match
4550 /g loops for multiple occurrences, using the new startoffset argument
4551 /G loops for multiple occurrences by passing an incremented pointer
4552
4553 4. PCRE wasn't doing the "first character" optimization for patterns starting
4554 with \b or \B, though it was doing it for other lookbehind assertions. That is,
4555 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
4556 the letter 'x'. On long subject strings, this gives a significant speed-up.
4557
4558
4559 Version 2.05 21-Apr-99
4560 ----------------------
4561
4562 1. Changed the type of magic_number from int to long int so that it works
4563 properly on 16-bit systems.
4564
4565 2. Fixed a bug which caused patterns starting with .* not to work correctly
4566 when the subject string contained newline characters. PCRE was assuming
4567 anchoring for such patterns in all cases, which is not correct because .* will
4568 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
4569 DOTALL is set at top level; otherwise it knows that patterns starting with .*
4570 must be retried after every newline in the subject.
4571
4572
4573 Version 2.04 18-Feb-99
4574 ----------------------
4575
4576 1. For parenthesized subpatterns with repeats whose minimum was zero, the
4577 computation of the store needed to hold the pattern was incorrect (too large).
4578 If such patterns were nested a few deep, this could multiply and become a real
4579 problem.
4580
4581 2. Added /M option to pcretest to show the memory requirement of a specific
4582 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
4583
4584 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
4585 compiled in such a way that the backtracking after subsequent failure was
4586 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
4587 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
4588
4589
4590 Version 2.03 02-Feb-99
4591 ----------------------
4592
4593 1. Fixed typo and small mistake in man page.
4594
4595 2. Added 4th condition (GPL supersedes if conflict) and created separate
4596 LICENCE file containing the conditions.
4597
4598 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
4599 Perl, that is the internal \ allows the delimiter to be included in the
4600 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
4601 the final delimiter, add \ to the end of the pattern (to test the error).
4602
4603 4. Added the convenience functions for extracting substrings after a successful
4604 match. Updated pcretest to make it able to test these functions.
4605
4606
4607 Version 2.02 14-Jan-99
4608 ----------------------
4609
4610 1. Initialized the working variables associated with each extraction so that
4611 their saving and restoring doesn't refer to uninitialized store.
4612
4613 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
4614 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
4615 fix the problem.
4616
4617 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
4618 calls, and wasn't printing the correct value for compiling calls. Increased the
4619 default value of LOOPREPEAT, and the number of significant figures in the
4620 times.
4621
4622 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
4623
4624 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
4625 a building problem on Windows NT with a FAT file system.
4626
4627
4628 Version 2.01 21-Oct-98
4629 ----------------------
4630
4631 1. Changed the API for pcre_compile() to allow for the provision of a pointer
4632 to character tables built by pcre_maketables() in the current locale. If NULL
4633 is passed, the default tables are used.
4634
4635
4636 Version 2.00 24-Sep-98
4637 ----------------------
4638
4639 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
4640 it any more.
4641
4642 2. Allow quantification of (?>) groups, and make it work correctly.
4643
4644 3. The first character computation wasn't working for (?>) groups.
4645
4646 4. Correct the implementation of \Z (it is permitted to match on the \n at the
4647 end of the subject) and add 5.005's \z, which really does match only at the
4648 very end of the subject.
4649
4650 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
4651
4652 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
4653 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
4654 localized options. All options to pcre_study() were also removed.
4655
4656 7. Add other new features from 5.005:
4657
4658 $(?<= positive lookbehind
4659 $(?<! negative lookbehind
4660 (?imsx-imsx) added the unsetting capability
4661 such a setting is global if at outer level; local otherwise
4662 (?imsx-imsx:) non-capturing groups with option setting
4663 (?(cond)re|re) conditional pattern matching
4664
4665 A backreference to itself in a repeated group matches the previous
4666 captured string.
4667
4668 8. General tidying up of studying (both automatic and via "study")
4669 consequential on the addition of new assertions.
4670
4671 9. As in 5.005, unlimited repeated groups that could match an empty substring
4672 are no longer faulted at compile time. Instead, the loop is forcibly broken at
4673 runtime if any iteration does actually match an empty substring.
4674
4675 10. Include the RunTest script in the distribution.
4676
4677 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
4678 discrepancies, some of which were old and were also with respect to 5.004. They
4679 have now been fixed.
4680
4681
4682 Version 1.09 28-Apr-98
4683 ----------------------
4684
4685 1. A negated single character class followed by a quantifier with a minimum
4686 value of one (e.g. [^x]{1,6} ) was not compiled correctly. This could lead to
4687 program crashes, or just wrong answers. This did not apply to negated classes
4688 containing more than one character, or to minima other than one.
4689
4690
4691 Version 1.08 27-Mar-98
4692 ----------------------
4693
4694 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
4695
4696 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
4697 latter must appear before anything that relies on it in the pattern.
4698
4699
4700 Version 1.07 16-Feb-98
4701 ----------------------
4702
4703 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
4704 repeat of a potentially empty string).
4705
4706
4707 Version 1.06 23-Jan-98
4708 ----------------------
4709
4710 1. Added Markus Oberhumer's little patches for C++.
4711
4712 2. Literal strings longer than 255 characters were broken.
4713
4714
4715 Version 1.05 23-Dec-97
4716 ----------------------
4717
4718 1. Negated character classes containing more than one character were failing if
4719 PCRE_CASELESS was set at run time.
4720
4721
4722 Version 1.04 19-Dec-97
4723 ----------------------
4724
4725 1. Corrected the man page, where some "const" qualifiers had been omitted.
4726
4727 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
4728 input syntax.
4729
4730 3. Fixed memory leak which occurred when a regex with back references was
4731 matched with an offsets vector that wasn't big enough. The temporary memory
4732 that is used in this case wasn't being freed if the match failed.
4733
4734 4. Tidied pcretest to ensure it frees memory that it gets.
4735
4736 5. Temporary memory was being obtained in the case where the passed offsets
4737 vector was exactly big enough.
4738
4739 6. Corrected definition of offsetof() from change 5 below.
4740
4741 7. I had screwed up change 6 below and broken the rules for the use of
4742 setjmp(). Now fixed.
4743
4744
4745 Version 1.03 18-Dec-97
4746 ----------------------
4747
4748 1. A erroneous regex with a missing opening parenthesis was correctly
4749 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
4750 on some systems.
4751
4752 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
4753 it was reported that one broken compiler failed on the former because "code" is
4754 also an independent variable.
4755
4756 3. The erroneous regex a[]b caused an array overrun reference.
4757
4758 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
4759 fail on data ending with that character. (It was going on too far, and checking
4760 the next character, typically a binary zero.) This was specific to the
4761 optimized code for single-character negative classes.
4762
4763 5. Added a contributed patch from the TIN world which does the following:
4764
4765 + Add an undef for memmove, in case the the system defines a macro for it.
4766
4767 + Add a definition of offsetof(), in case there isn't one. (I don't know
4768 the reason behind this - offsetof() is part of the ANSI standard - but
4769 it does no harm).
4770
4771 + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
4772 most of the places where whitespace preceded '#'. I have given up and
4773 allowed the remaining 2 cases to be at the margin.
4774
4775 + Rename some variables in pcre to eliminate shadowing. This seems very
4776 pedantic, but does no harm, of course.
4777
4778 6. Moved the call to setjmp() into its own function, to get rid of warnings
4779 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
4780
4781 7. Constructs such as \d{8,} were compiling into the equivalent of
4782 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
4783 outcome, but in this particular case used more store than had been allocated,
4784 which caused the bug to be discovered because it threw up an internal error.
4785
4786 8. The debugging code in both pcre and pcretest for outputting the compiled
4787 form of a regex was going wrong in the case of back references followed by
4788 curly-bracketed repeats.
4789
4790
4791 Version 1.02 12-Dec-97
4792 ----------------------
4793
4794 1. Typos in pcre.3 and comments in the source fixed.
4795
4796 2. Applied a contributed patch to get rid of places where it used to remove
4797 'const' from variables, and fixed some signed/unsigned and uninitialized
4798 variable warnings.
4799
4800 3. Added the "runtest" target to Makefile.
4801
4802 4. Set default compiler flag to -O2 rather than just -O.
4803
4804
4805 Version 1.01 19-Nov-97
4806 ----------------------
4807
4808 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
4809 like /([ab]*)*/, that is, for classes with more than one character in them.
4810
4811 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
4812 as /((?>a*))*/ (a PCRE_EXTRA facility).
4813
4814
4815 Version 1.00 18-Nov-97
4816 ----------------------
4817
4818 1. Added compile-time macros to support systems such as SunOS4 which don't have
4819 memmove() or strerror() but have other things that can be used instead.
4820
4821 2. Arranged that "make clean" removes the executables.
4822
4823
4824 Version 0.99 27-Oct-97
4825 ----------------------
4826
4827 1. Fixed bug in code for optimizing classes with only one character. It was
4828 initializing a 32-byte map regardless, which could cause it to run off the end
4829 of the memory it had got.
4830
4831 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
4832
4833
4834 Version 0.98 22-Oct-97
4835 ----------------------
4836
4837 1. Fixed bug in code for handling temporary memory usage when there are more
4838 back references than supplied space in the ovector. This could cause segfaults.
4839
4840
4841 Version 0.97 21-Oct-97
4842 ----------------------
4843
4844 1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
4845
4846 2. Optimized negated single characters not to use a bit map.
4847
4848 3. Brought error texts together as macro definitions; clarified some of them;
4849 fixed one that was wrong - it said "range out of order" when it meant "invalid
4850 escape sequence".
4851
4852 4. Changed some char * arguments to const char *.
4853
4854 5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
4855
4856 6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
4857 pcretest.
4858
4859
4860 Version 0.96 16-Oct-97
4861 ----------------------
4862
4863 1. Added a simple "pgrep" utility to the distribution.
4864
4865 2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
4866 unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
4867 where "ddd" means "one or more decimal digits".
4868
4869 3. Fixed serious bug. If a pattern had a back reference, but the call to
4870 pcre_exec() didn't supply a large enough ovector to record the related
4871 identifying subpattern, the match always failed. PCRE now remembers the number
4872 of the largest back reference, and gets some temporary memory in which to save
4873 the offsets during matching if necessary, in order to ensure that
4874 backreferences always work.
4875
4876 4. Increased the compatibility with Perl in a number of ways:
4877
4878 (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
4879 to request this handling. The option can be set at compile or exec time.
4880
4881 (b) $ matches before a terminating newline by default; an option
4882 PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
4883 mode). The option can be set at compile or exec time.
4884
4885 (c) The handling of \ followed by a digit other than 0 is now supposed to be
4886 the same as Perl's. If the decimal number it represents is less than 10
4887 or there aren't that many previous left capturing parentheses, an octal
4888 escape is read. Inside a character class, it's always an octal escape,
4889 even if it is a single digit.
4890
4891 (d) An escaped but undefined alphabetic character is taken as a literal,
4892 unless PCRE_EXTRA is set. Currently this just reserves the remaining
4893 escapes.
4894
4895 (e) {0} is now permitted. (The previous item is removed from the compiled
4896 pattern).
4897
4898 5. Changed all the names of code files so that the basic parts are no longer
4899 than 10 characters, and abolished the teeny "globals.c" file.
4900
4901 6. Changed the handling of character classes; they are now done with a 32-byte
4902 bit map always.
4903
4904 7. Added the -d and /D options to pcretest to make it possible to look at the
4905 internals of compilation without having to recompile pcre.
4906
4907
4908 Version 0.95 23-Sep-97
4909 ----------------------
4910
4911 1. Fixed bug in pre-pass concerning escaped "normal" characters such as \x5c or
4912 \x20 at the start of a run of normal characters. These were being treated as
4913 real characters, instead of the source characters being re-checked.
4914
4915
4916 Version 0.94 18-Sep-97
4917 ----------------------
4918
4919 1. The functions are now thread-safe, with the caveat that the global variables
4920 containing pointers to malloc() and free() or alternative functions are the
4921 same for all threads.
4922
4923 2. Get pcre_study() to generate a bitmap of initial characters for non-
4924 anchored patterns when this is possible, and use it if passed to pcre_exec().
4925
4926
4927 Version 0.93 15-Sep-97
4928 ----------------------
4929
4930 1. /(b)|(:+)/ was computing an incorrect first character.
4931
4932 2. Add pcre_study() to the API and the passing of pcre_extra to pcre_exec(),
4933 but not actually doing anything yet.
4934
4935 3. Treat "-" characters in classes that cannot be part of ranges as literals,
4936 as Perl does (e.g. [-az] or [az-]).
4937
4938 4. Set the anchored flag if a branch starts with .* or .*? because that tests
4939 all possible positions.
4940
4941 5. Split up into different modules to avoid including unneeded functions in a
4942 compiled binary. However, compile and exec are still in one module. The "study"
4943 function is split off.
4944
4945 6. The character tables are now in a separate module whose source is generated
4946 by an auxiliary program - but can then be edited by hand if required. There are
4947 now no calls to isalnum(), isspace(), isdigit(), isxdigit(), tolower() or
4948 toupper() in the code.
4949
4950 7. Turn the malloc/free funtions variables into pcre_malloc and pcre_free and
4951 make them global. Abolish the function for setting them, as the caller can now
4952 set them directly.
4953
4954
4955 Version 0.92 11-Sep-97
4956 ----------------------
4957
4958 1. A repeat with a fixed maximum and a minimum of 1 for an ordinary character
4959 (e.g. /a{1,3}/) was broken (I mis-optimized it).
4960
4961 2. Caseless matching was not working in character classes if the characters in
4962 the pattern were in upper case.
4963
4964 3. Make ranges like [W-c] work in the same way as Perl for caseless matching.
4965
4966 4. Make PCRE_ANCHORED public and accept as a compile option.
4967
4968 5. Add an options word to pcre_exec() and accept PCRE_ANCHORED and
4969 PCRE_CASELESS at run time. Add escapes \A and \I to pcretest to cause it to
4970 pass them.
4971
4972 6. Give an error if bad option bits passed at compile or run time.
4973
4974 7. Add PCRE_MULTILINE at compile and exec time, and (?m) as well. Add \M to
4975 pcretest to cause it to pass that flag.
4976
4977 8. Add pcre_info(), to get the number of identifying subpatterns, the stored
4978 options, and the first character, if set.
4979
4980 9. Recognize C+ or C{n,m} where n >= 1 as providing a fixed starting character.
4981
4982
4983 Version 0.91 10-Sep-97
4984 ----------------------
4985
4986 1. PCRE was failing to diagnose unlimited repeats of subpatterns that could
4987 match the empty string as in /(a*)*/. It was looping and ultimately crashing.
4988
4989 2. PCRE was looping on encountering an indefinitely repeated back reference to
4990 a subpattern that had matched an empty string, e.g. /(a|)\1*/. It now does what
4991 Perl does - treats the match as successful.
4992
4993 ****

Properties

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

  ViewVC Help
Powered by ViewVC 1.1.5