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

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1279 - (show annotations)
Tue Mar 12 17:27:34 2013 UTC (6 years, 5 months ago) by zherczeg
File size: 244742 byte(s)
Experimental support of (*THEN) backtracking verb in the JIT compiler.
1 ChangeLog for PCRE
2 ------------------
3
4 Version 8.33 xx-xxxx-201x
5 -------------------------
6
7 1. Added 'U' to some constants that are compared to unsigned integers, to
8 avoid compiler signed/unsigned warnings. Added (int) casts to unsigned
9 variables that are added to signed variables, to ensure the result is
10 signed and can be negated.
11
12 2. Applied patch by Daniel Richard G for quashing MSVC warnings to the
13 CMake config files.
14
15 3. Revise the creation of config.h.generic so that all boolean macros are
16 #undefined, whereas non-boolean macros are #ifndef/#endif-ed. This makes
17 overriding via -D on the command line possible.
18
19 4. Changing the definition of the variable "op" in pcre_exec.c from pcre_uchar
20 to unsigned int is reported to make a quite noticeable speed difference in
21 a specific Windows environment. Testing on Linux did also appear to show
22 some benefit (and it is clearly not harmful). Also fixed the definition of
23 Xop which should be unsigned.
24
25 5. Related to (4), changing the definition of the intermediate variable cc
26 in repeated character loops from pcre_uchar to pcre_uint32 also gave speed
27 improvements.
28
29 6. Fix forward search in JIT when link size is 3 or greater. Also removed some
30 unnecessary spaces.
31
32 7. Adjust autogen.sh and configure.ac to lose warnings given by automake 1.12
33 and later.
34
35 8. Fix two buffer over read issues in 16 and 32 bit modes. Affects JIT only.
36
37 9. Optimizing fast_forward_start_bits in JIT.
38
39 10. Adding experimental support for callouts in JIT, and fixing some
40 issues revealed during this work. Namely:
41
42 (a) Unoptimized capturing brackets incorrectly reset on backtrack.
43
44 (b) Minimum length was not checked before the matching is started.
45
46 11. The value of capture_last that is passed to callouts was incorrect in some
47 cases when there was a capture on one path that was subsequently abandoned
48 after a backtrack. Also, the capture_last value is now reset after a
49 recursion, since all captures are also reset in this case.
50
51 12. The interpreter no longer returns the "too many substrings" error in the
52 case when an overflowing capture is in a branch that is subsequently
53 abandoned after a backtrack.
54
55 13. In the pathological case when an offset vector of size 2 is used, pcretest
56 now prints out the matched string after a yield of 0 or 1.
57
58 14. Inlining subpatterns in recursions, when certain conditions are fulfilled.
59 Only supported by the JIT compiler at the moment.
60
61 15. JIT compiler now supports 32 bit Macs thanks to Lawrence Velazquez.
62
63 16. Partial matches now set offsets[2] to the "bumpalong" value, that is, the
64 offset of the starting point of the matching process, provided the offsets
65 vector is large enough.
66
67 17. The \A escape now records a lookbehind value of 1, though its execution
68 does not actually inspect the previous character. This is to ensure that,
69 in partial multi-segment matching, at least one character from the old
70 segment is retained when a new segment is processed. Otherwise, if there
71 are no lookbehinds in the pattern, \A might match incorrectly at the start
72 of a new segment.
73
74 18. Added some #ifdef __VMS code into pcretest.c to help VMS implementations.
75
76 19. Redefined some pcre_uchar variables in pcre_exec.c as pcre_uint32; this
77 gives some modest performance improvement in 8-bit mode.
78
79 20. Added the PCRE-specific property \p{Xuc} for matching characters that can
80 be expressed in certain programming languages using Universal Character
81 Names.
82
83 21. Unicode validation has been updated in the light of Unicode Corrigendum #9,
84 which points out that "non characters" are not "characters that may not
85 appear in Unicode strings" but rather "characters that are reserved for
86 internal use and have only local meaning".
87
88 22. When a pattern was compiled with automatic callouts (PCRE_AUTO_CALLOUT) and
89 there was a conditional group that depended on an assertion, if the
90 assertion was false, the callout that immediately followed the alternation
91 in the condition was skipped when pcre_exec() was used for matching.
92
93 23. Allow an explicit callout to be inserted before an assertion that is the
94 condition for a conditional group, for compatibility with automatic
95 callouts, which always insert a callout at this point.
96
97 24. In 8.31, (*COMMIT) was confined to within a recursive subpattern. Perl also
98 confines (*SKIP) and (*PRUNE) in the same way, and this has now been done.
99
100 25. (*PRUNE) is now supported by the JIT compiler.
101
102 26. Fix infinite loop when /(?<=(*SKIP)ac)a/ is matched against aa.
103
104 27. Fix the case where there are two or more SKIPs with arguments that may be
105 ignored.
106
107 28. Experimental support of (*SKIP) backtracking verb in the JIT compiler.
108
109 29. Experimental support of (*THEN) backtracking verb in the JIT compiler.
110
111
112 Version 8.32 30-November-2012
113 -----------------------------
114
115 1. Improved JIT compiler optimizations for first character search and single
116 character iterators.
117
118 2. Supporting IBM XL C compilers for PPC architectures in the JIT compiler.
119 Patch by Daniel Richard G.
120
121 3. Single character iterator optimizations in the JIT compiler.
122
123 4. Improved JIT compiler optimizations for character ranges.
124
125 5. Rename the "leave" variable names to "quit" to improve WinCE compatibility.
126 Reported by Giuseppe D'Angelo.
127
128 6. The PCRE_STARTLINE bit, indicating that a match can occur only at the start
129 of a line, was being set incorrectly in cases where .* appeared inside
130 atomic brackets at the start of a pattern, or where there was a subsequent
131 *PRUNE or *SKIP.
132
133 7. Improved instruction cache flush for POWER/PowerPC.
134 Patch by Daniel Richard G.
135
136 8. Fixed a number of issues in pcregrep, making it more compatible with GNU
137 grep:
138
139 (a) There is now no limit to the number of patterns to be matched.
140
141 (b) An error is given if a pattern is too long.
142
143 (c) Multiple uses of --exclude, --exclude-dir, --include, and --include-dir
144 are now supported.
145
146 (d) --exclude-from and --include-from (multiple use) have been added.
147
148 (e) Exclusions and inclusions now apply to all files and directories, not
149 just to those obtained from scanning a directory recursively.
150
151 (f) Multiple uses of -f and --file-list are now supported.
152
153 (g) In a Windows environment, the default for -d has been changed from
154 "read" (the GNU grep default) to "skip", because otherwise the presence
155 of a directory in the file list provokes an error.
156
157 (h) The documentation has been revised and clarified in places.
158
159 9. Improve the matching speed of capturing brackets.
160
161 10. Changed the meaning of \X so that it now matches a Unicode extended
162 grapheme cluster.
163
164 11. Patch by Daniel Richard G to the autoconf files to add a macro for sorting
165 out POSIX threads when JIT support is configured.
166
167 12. Added support for PCRE_STUDY_EXTRA_NEEDED.
168
169 13. In the POSIX wrapper regcomp() function, setting re_nsub field in the preg
170 structure could go wrong in environments where size_t is not the same size
171 as int.
172
173 14. Applied user-supplied patch to pcrecpp.cc to allow PCRE_NO_UTF8_CHECK to be
174 set.
175
176 15. The EBCDIC support had decayed; later updates to the code had included
177 explicit references to (e.g.) \x0a instead of CHAR_LF. There has been a
178 general tidy up of EBCDIC-related issues, and the documentation was also
179 not quite right. There is now a test that can be run on ASCII systems to
180 check some of the EBCDIC-related things (but is it not a full test).
181
182 16. The new PCRE_STUDY_EXTRA_NEEDED option is now used by pcregrep, resulting
183 in a small tidy to the code.
184
185 17. Fix JIT tests when UTF is disabled and both 8 and 16 bit mode are enabled.
186
187 18. If the --only-matching (-o) option in pcregrep is specified multiple
188 times, each one causes appropriate output. For example, -o1 -o2 outputs the
189 substrings matched by the 1st and 2nd capturing parentheses. A separating
190 string can be specified by --om-separator (default empty).
191
192 19. Improving the first n character searches.
193
194 20. Turn case lists for horizontal and vertical white space into macros so that
195 they are defined only once.
196
197 21. This set of changes together give more compatible Unicode case-folding
198 behaviour for characters that have more than one other case when UCP
199 support is available.
200
201 (a) The Unicode property table now has offsets into a new table of sets of
202 three or more characters that are case-equivalent. The MultiStage2.py
203 script that generates these tables (the pcre_ucd.c file) now scans
204 CaseFolding.txt instead of UnicodeData.txt for character case
205 information.
206
207 (b) The code for adding characters or ranges of characters to a character
208 class has been abstracted into a generalized function that also handles
209 case-independence. In UTF-mode with UCP support, this uses the new data
210 to handle characters with more than one other case.
211
212 (c) A bug that is fixed as a result of (b) is that codepoints less than 256
213 whose other case is greater than 256 are now correctly matched
214 caselessly. Previously, the high codepoint matched the low one, but not
215 vice versa.
216
217 (d) The processing of \h, \H, \v, and \ in character classes now makes use
218 of the new class addition function, using character lists defined as
219 macros alongside the case definitions of 20 above.
220
221 (e) Caseless back references now work with characters that have more than
222 one other case.
223
224 (f) General caseless matching of characters with more than one other case
225 is supported.
226
227 22. Unicode character properties were updated from Unicode 6.2.0
228
229 23. Improved CMake support under Windows. Patch by Daniel Richard G.
230
231 24. Add support for 32-bit character strings, and UTF-32
232
233 25. Major JIT compiler update (code refactoring and bugfixing).
234 Experimental Sparc 32 support is added.
235
236 26. Applied a modified version of Daniel Richard G's patch to create
237 pcre.h.generic and config.h.generic by "make" instead of in the
238 PrepareRelease script.
239
240 27. Added a definition for CHAR_NULL (helpful for the z/OS port), and use it in
241 pcre_compile.c when checking for a zero character.
242
243 28. Introducing a native interface for JIT. Through this interface, the compiled
244 machine code can be directly executed. The purpose of this interface is to
245 provide fast pattern matching, so several sanity checks are not performed.
246 However, feature tests are still performed. The new interface provides
247 1.4x speedup compared to the old one.
248
249 29. If pcre_exec() or pcre_dfa_exec() was called with a negative value for
250 the subject string length, the error given was PCRE_ERROR_BADOFFSET, which
251 was confusing. There is now a new error PCRE_ERROR_BADLENGTH for this case.
252
253 30. In 8-bit UTF-8 mode, pcretest failed to give an error for data codepoints
254 greater than 0x7fffffff (which cannot be represented in UTF-8, even under
255 the "old" RFC 2279). Instead, it ended up passing a negative length to
256 pcre_exec().
257
258 31. Add support for GCC's visibility feature to hide internal functions.
259
260 32. Running "pcretest -C pcre8" or "pcretest -C pcre16" gave a spurious error
261 "unknown -C option" after outputting 0 or 1.
262
263 33. There is now support for generating a code coverage report for the test
264 suite in environments where gcc is the compiler and lcov is installed. This
265 is mainly for the benefit of the developers.
266
267 34. If PCRE is built with --enable-valgrind, certain memory regions are marked
268 unaddressable using valgrind annotations, allowing valgrind to detect
269 invalid memory accesses. This is mainly for the benefit of the developers.
270
271 25. (*UTF) can now be used to start a pattern in any of the three libraries.
272
273 26. Give configure error if --enable-cpp but no C++ compiler found.
274
275
276 Version 8.31 06-July-2012
277 -------------------------
278
279 1. Fixing a wrong JIT test case and some compiler warnings.
280
281 2. Removed a bashism from the RunTest script.
282
283 3. Add a cast to pcre_exec.c to fix the warning "unary minus operator applied
284 to unsigned type, result still unsigned" that was given by an MS compiler
285 on encountering the code "-sizeof(xxx)".
286
287 4. Partial matching support is added to the JIT compiler.
288
289 5. Fixed several bugs concerned with partial matching of items that consist
290 of more than one character:
291
292 (a) /^(..)\1/ did not partially match "aba" because checking references was
293 done on an "all or nothing" basis. This also applied to repeated
294 references.
295
296 (b) \R did not give a hard partial match if \r was found at the end of the
297 subject.
298
299 (c) \X did not give a hard partial match after matching one or more
300 characters at the end of the subject.
301
302 (d) When newline was set to CRLF, a pattern such as /a$/ did not recognize
303 a partial match for the string "\r".
304
305 (e) When newline was set to CRLF, the metacharacter "." did not recognize
306 a partial match for a CR character at the end of the subject string.
307
308 6. If JIT is requested using /S++ or -s++ (instead of just /S+ or -s+) when
309 running pcretest, the text "(JIT)" added to the output whenever JIT is
310 actually used to run the match.
311
312 7. Individual JIT compile options can be set in pcretest by following -s+[+]
313 or /S+[+] with a digit between 1 and 7.
314
315 8. OP_NOT now supports any UTF character not just single-byte ones.
316
317 9. (*MARK) control verb is now supported by the JIT compiler.
318
319 10. The command "./RunTest list" lists the available tests without actually
320 running any of them. (Because I keep forgetting what they all are.)
321
322 11. Add PCRE_INFO_MAXLOOKBEHIND.
323
324 12. Applied a (slightly modified) user-supplied patch that improves performance
325 when the heap is used for recursion (compiled with --disable-stack-for-
326 recursion). Instead of malloc and free for each heap frame each time a
327 logical recursion happens, frames are retained on a chain and re-used where
328 possible. This sometimes gives as much as 30% improvement.
329
330 13. As documented, (*COMMIT) is now confined to within a recursive subpattern
331 call.
332
333 14. As documented, (*COMMIT) is now confined to within a positive assertion.
334
335 15. It is now possible to link pcretest with libedit as an alternative to
336 libreadline.
337
338 16. (*COMMIT) control verb is now supported by the JIT compiler.
339
340 17. The Unicode data tables have been updated to Unicode 6.1.0.
341
342 18. Added --file-list option to pcregrep.
343
344 19. Added binary file support to pcregrep, including the -a, --binary-files,
345 -I, and --text options.
346
347 20. The madvise function is renamed for posix_madvise for QNX compatibility
348 reasons. Fixed by Giuseppe D'Angelo.
349
350 21. Fixed a bug for backward assertions with REVERSE 0 in the JIT compiler.
351
352 22. Changed the option for creating symbolic links for 16-bit man pages from
353 -s to -sf so that re-installing does not cause issues.
354
355 23. Support PCRE_NO_START_OPTIMIZE in JIT as (*MARK) support requires it.
356
357 24. Fixed a very old bug in pcretest that caused errors with restarted DFA
358 matches in certain environments (the workspace was not being correctly
359 retained). Also added to pcre_dfa_exec() a simple plausibility check on
360 some of the workspace data at the beginning of a restart.
361
362 25. \s*\R was auto-possessifying the \s* when it should not, whereas \S*\R
363 was not doing so when it should - probably a typo introduced by SVN 528
364 (change 8.10/14).
365
366 26. When PCRE_UCP was not set, \w+\x{c4} was incorrectly auto-possessifying the
367 \w+ when the character tables indicated that \x{c4} was a word character.
368 There were several related cases, all because the tests for doing a table
369 lookup were testing for characters less than 127 instead of 255.
370
371 27. If a pattern contains capturing parentheses that are not used in a match,
372 their slots in the ovector are set to -1. For those that are higher than
373 any matched groups, this happens at the end of processing. In the case when
374 there were back references that the ovector was too small to contain
375 (causing temporary malloc'd memory to be used during matching), and the
376 highest capturing number was not used, memory off the end of the ovector
377 was incorrectly being set to -1. (It was using the size of the temporary
378 memory instead of the true size.)
379
380 28. To catch bugs like 27 using valgrind, when pcretest is asked to specify an
381 ovector size, it uses memory at the end of the block that it has got.
382
383 29. Check for an overlong MARK name and give an error at compile time. The
384 limit is 255 for the 8-bit library and 65535 for the 16-bit library.
385
386 30. JIT compiler update.
387
388 31. JIT is now supported on jailbroken iOS devices. Thanks for Ruiger
389 Rill for the patch.
390
391 32. Put spaces around SLJIT_PRINT_D in the JIT compiler. Required by CXX11.
392
393 33. Variable renamings in the PCRE-JIT compiler. No functionality change.
394
395 34. Fixed typos in pcregrep: in two places there was SUPPORT_LIBZ2 instead of
396 SUPPORT_LIBBZ2. This caused a build problem when bzip2 but not gzip (zlib)
397 was enabled.
398
399 35. Improve JIT code generation for greedy plus quantifier.
400
401 36. When /((?:a?)*)*c/ or /((?>a?)*)*c/ was matched against "aac", it set group
402 1 to "aa" instead of to an empty string. The bug affected repeated groups
403 that could potentially match an empty string.
404
405 37. Optimizing single character iterators in JIT.
406
407 38. Wide characters specified with \uxxxx in JavaScript mode are now subject to
408 the same checks as \x{...} characters in non-JavaScript mode. Specifically,
409 codepoints that are too big for the mode are faulted, and in a UTF mode,
410 disallowed codepoints are also faulted.
411
412 39. If PCRE was compiled with UTF support, in three places in the DFA
413 matcher there was code that should only have been obeyed in UTF mode, but
414 was being obeyed unconditionally. In 8-bit mode this could cause incorrect
415 processing when bytes with values greater than 127 were present. In 16-bit
416 mode the bug would be provoked by values in the range 0xfc00 to 0xdc00. In
417 both cases the values are those that cannot be the first data item in a UTF
418 character. The three items that might have provoked this were recursions,
419 possessively repeated groups, and atomic groups.
420
421 40. Ensure that libpcre is explicitly listed in the link commands for pcretest
422 and pcregrep, because some OS require shared objects to be explicitly
423 passed to ld, causing the link step to fail if they are not.
424
425 41. There were two incorrect #ifdefs in pcre_study.c, meaning that, in 16-bit
426 mode, patterns that started with \h* or \R* might be incorrectly matched.
427
428
429 Version 8.30 04-February-2012
430 -----------------------------
431
432 1. Renamed "isnumber" as "is_a_number" because in some Mac environments this
433 name is defined in ctype.h.
434
435 2. Fixed a bug in fixed-length calculation for lookbehinds that would show up
436 only in quite long subpatterns.
437
438 3. Removed the function pcre_info(), which has been obsolete and deprecated
439 since it was replaced by pcre_fullinfo() in February 2000.
440
441 4. For a non-anchored pattern, if (*SKIP) was given with a name that did not
442 match a (*MARK), and the match failed at the start of the subject, a
443 reference to memory before the start of the subject could occur. This bug
444 was introduced by fix 17 of release 8.21.
445
446 5. A reference to an unset group with zero minimum repetition was giving
447 totally wrong answers (in non-JavaScript-compatibility mode). For example,
448 /(another)?(\1?)test/ matched against "hello world test". This bug was
449 introduced in release 8.13.
450
451 6. Add support for 16-bit character strings (a large amount of work involving
452 many changes and refactorings).
453
454 7. RunGrepTest failed on msys because \r\n was replaced by whitespace when the
455 command "pattern=`printf 'xxx\r\njkl'`" was run. The pattern is now taken
456 from a file.
457
458 8. Ovector size of 2 is also supported by JIT based pcre_exec (the ovector size
459 rounding is not applied in this particular case).
460
461 9. The invalid Unicode surrogate codepoints U+D800 to U+DFFF are now rejected
462 if they appear, or are escaped, in patterns.
463
464 10. Get rid of a number of -Wunused-but-set-variable warnings.
465
466 11. The pattern /(?=(*:x))(q|)/ matches an empty string, and returns the mark
467 "x". The similar pattern /(?=(*:x))((*:y)q|)/ did not return a mark at all.
468 Oddly, Perl behaves the same way. PCRE has been fixed so that this pattern
469 also returns the mark "x". This bug applied to capturing parentheses,
470 non-capturing parentheses, and atomic parentheses. It also applied to some
471 assertions.
472
473 12. Stephen Kelly's patch to CMakeLists.txt allows it to parse the version
474 information out of configure.ac instead of relying on pcre.h.generic, which
475 is not stored in the repository.
476
477 13. Applied Dmitry V. Levin's patch for a more portable method for linking with
478 -lreadline.
479
480 14. ZH added PCRE_CONFIG_JITTARGET; added its output to pcretest -C.
481
482 15. Applied Graycode's patch to put the top-level frame on the stack rather
483 than the heap when not using the stack for recursion. This gives a
484 performance improvement in many cases when recursion is not deep.
485
486 16. Experimental code added to "pcretest -C" to output the stack frame size.
487
488
489 Version 8.21 12-Dec-2011
490 ------------------------
491
492 1. Updating the JIT compiler.
493
494 2. JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
495 are added as well.
496
497 3. Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
498 PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
499 calling _pcre_jit_exec. Some extra comments are added.
500
501 4. (*MARK) settings inside atomic groups that do not contain any capturing
502 parentheses, for example, (?>a(*:m)), were not being passed out. This bug
503 was introduced by change 18 for 8.20.
504
505 5. Supporting of \x, \U and \u in JavaScript compatibility mode based on the
506 ECMA-262 standard.
507
508 6. Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
509 erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
510 This bug was probably introduced by change 9 of 8.13.
511
512 7. While fixing 6 above, I noticed that a number of other items were being
513 incorrectly rejected as "not fixed length". This arose partly because newer
514 opcodes had not been added to the fixed-length checking code. I have (a)
515 corrected the bug and added tests for these items, and (b) arranged for an
516 error to occur if an unknown opcode is encountered while checking for fixed
517 length instead of just assuming "not fixed length". The items that were
518 rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
519 (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
520 repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
521
522 8. A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
523 being incorrectly compiled and would have given unpredicatble results.
524
525 9. A possessively repeated subpattern with minimum repeat count greater than
526 one behaved incorrectly. For example, (A){2,}+ behaved as if it was
527 (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
528 the first (A) could occur when it should not.
529
530 10. Add a cast and remove a redundant test from the code.
531
532 11. JIT should use pcre_malloc/pcre_free for allocation.
533
534 12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
535 best practice nowadays, and helps with cross-compiling. (If the exec_prefix
536 is anything other than /usr, -L is still shown).
537
538 13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
539
540 14. Perl does not support \N without a following name in a [] class; PCRE now
541 also gives an error.
542
543 15. If a forward reference was repeated with an upper limit of around 2000,
544 it caused the error "internal error: overran compiling workspace". The
545 maximum number of forward references (including repeats) was limited by the
546 internal workspace, and dependent on the LINK_SIZE. The code has been
547 rewritten so that the workspace expands (via pcre_malloc) if necessary, and
548 the default depends on LINK_SIZE. There is a new upper limit (for safety)
549 of around 200,000 forward references. While doing this, I also speeded up
550 the filling in of repeated forward references.
551
552 16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
553 incorrectly expecting the subject to contain another "a" after the start.
554
555 17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
556 in the match, the SKIP should be ignored. This was not happening; instead
557 the SKIP was being treated as NOMATCH. For patterns such as
558 /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
559 tested.
560
561 18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
562 now much more compatible with Perl, in particular in cases where the result
563 is a non-match for a non-anchored pattern. For example, if
564 /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
565 "m", where previously it did not return a name. A side effect of this
566 change is that for partial matches, the last encountered mark name is
567 returned, as for non matches. A number of tests that were previously not
568 Perl-compatible have been moved into the Perl-compatible test files. The
569 refactoring has had the pleasing side effect of removing one argument from
570 the match() function, thus reducing its stack requirements.
571
572 19. If the /S+ option was used in pcretest to study a pattern using JIT,
573 subsequent uses of /S (without +) incorrectly behaved like /S+.
574
575 21. Retrieve executable code size support for the JIT compiler and fixing
576 some warnings.
577
578 22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
579 not work when the shorter character appeared right at the end of the
580 subject string.
581
582 23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
583 systems.
584
585 24. Added PCRE_INFO_JITSIZE to pass on the value from (21) above, and also
586 output it when the /M option is used in pcretest.
587
588 25. The CheckMan script was not being included in the distribution. Also, added
589 an explicit "perl" to run Perl scripts from the PrepareRelease script
590 because this is reportedly needed in Windows.
591
592 26. If study data was being save in a file and studying had not found a set of
593 "starts with" bytes for the pattern, the data written to the file (though
594 never used) was taken from uninitialized memory and so caused valgrind to
595 complain.
596
597 27. Updated RunTest.bat as provided by Sheri Pierce.
598
599 28. Fixed a possible uninitialized memory bug in pcre_jit_compile.c.
600
601 29. Computation of memory usage for the table of capturing group names was
602 giving an unnecessarily large value.
603
604
605 Version 8.20 21-Oct-2011
606 ------------------------
607
608 1. Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
609 a POSIX class. After further experiments with Perl, which convinced me that
610 Perl has bugs and confusions, a closing square bracket is no longer allowed
611 in a POSIX name. This bug also affected patterns with classes that started
612 with full stops.
613
614 2. If a pattern such as /(a)b|ac/ is matched against "ac", there is no
615 captured substring, but while checking the failing first alternative,
616 substring 1 is temporarily captured. If the output vector supplied to
617 pcre_exec() was not big enough for this capture, the yield of the function
618 was still zero ("insufficient space for captured substrings"). This cannot
619 be totally fixed without adding another stack variable, which seems a lot
620 of expense for a edge case. However, I have improved the situation in cases
621 such as /(a)(b)x|abc/ matched against "abc", where the return code
622 indicates that fewer than the maximum number of slots in the ovector have
623 been set.
624
625 3. Related to (2) above: when there are more back references in a pattern than
626 slots in the output vector, pcre_exec() uses temporary memory during
627 matching, and copies in the captures as far as possible afterwards. It was
628 using the entire output vector, but this conflicts with the specification
629 that only 2/3 is used for passing back captured substrings. Now it uses
630 only the first 2/3, for compatibility. This is, of course, another edge
631 case.
632
633 4. Zoltan Herczeg's just-in-time compiler support has been integrated into the
634 main code base, and can be used by building with --enable-jit. When this is
635 done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
636 runtime --no-jit option is given.
637
638 5. When the number of matches in a pcre_dfa_exec() run exactly filled the
639 ovector, the return from the function was zero, implying that there were
640 other matches that did not fit. The correct "exactly full" value is now
641 returned.
642
643 6. If a subpattern that was called recursively or as a subroutine contained
644 (*PRUNE) or any other control that caused it to give a non-standard return,
645 invalid errors such as "Error -26 (nested recursion at the same subject
646 position)" or even infinite loops could occur.
647
648 7. If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
649 computing the minimum length on reaching *ACCEPT, and so ended up with the
650 wrong value of 1 rather than 0. Further investigation indicates that
651 computing a minimum subject length in the presence of *ACCEPT is difficult
652 (think back references, subroutine calls), and so I have changed the code
653 so that no minimum is registered for a pattern that contains *ACCEPT.
654
655 8. If (*THEN) was present in the first (true) branch of a conditional group,
656 it was not handled as intended. [But see 16 below.]
657
658 9. Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
659 Sheri Pierce.
660
661 10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
662 the first byte in a match must be "a".
663
664 11. Change 17 for 8.13 increased the recursion depth for patterns like
665 /a(?:.)*?a/ drastically. I've improved things by remembering whether a
666 pattern contains any instances of (*THEN). If it does not, the old
667 optimizations are restored. It would be nice to do this on a per-group
668 basis, but at the moment that is not feasible.
669
670 12. In some environments, the output of pcretest -C is CRLF terminated. This
671 broke RunTest's code that checks for the link size. A single white space
672 character after the value is now allowed for.
673
674 13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
675 For "fr", it uses the Windows-specific input and output files.
676
677 14. If (*THEN) appeared in a group that was called recursively or as a
678 subroutine, it did not work as intended. [But see next item.]
679
680 15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
681 pattern fragments (but not containing any | characters). If A and B are
682 matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
683 was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
684 D. In other words, Perl considers parentheses that do not contain any |
685 characters to be part of a surrounding alternative, whereas PCRE was
686 treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
687 differently. PCRE now behaves in the same way as Perl, except in the case
688 of subroutine/recursion calls such as (?1) which have in any case always
689 been different (but PCRE had them first :-).
690
691 16. Related to 15 above: Perl does not treat the | in a conditional group as
692 creating alternatives. Such a group is treated in the same way as an
693 ordinary group without any | characters when processing (*THEN). PCRE has
694 been changed to match Perl's behaviour.
695
696 17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
697 RunGrepTest script failed.
698
699 18. Change 22 for version 13 caused atomic groups to use more stack. This is
700 inevitable for groups that contain captures, but it can lead to a lot of
701 stack use in large patterns. The old behaviour has been restored for atomic
702 groups that do not contain any capturing parentheses.
703
704 19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
705 suppress the check for a minimum subject length at run time. (If it was
706 given to pcre_exec() or pcre_dfa_exec() it did work.)
707
708 20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
709 fail to work when decoding hex characters in data strings in EBCDIC
710 environments.
711
712 21. It appears that in at least one Mac OS environment, the isxdigit() function
713 is implemented as a macro that evaluates to its argument more than once,
714 contravening the C 90 Standard (I haven't checked a later standard). There
715 was an instance in pcretest which caused it to go wrong when processing
716 \x{...} escapes in subject strings. The has been rewritten to avoid using
717 things like p++ in the argument of isxdigit().
718
719
720 Version 8.13 16-Aug-2011
721 ------------------------
722
723 1. The Unicode data tables have been updated to Unicode 6.0.0.
724
725 2. Two minor typos in pcre_internal.h have been fixed.
726
727 3. Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
728 pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
729 in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
730
731 4. There were a number of related bugs in the code for matching backrefences
732 caselessly in UTF-8 mode when codes for the characters concerned were
733 different numbers of bytes. For example, U+023A and U+2C65 are an upper
734 and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
735 (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
736 code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
737 2-byte code at the end of the subject (it thought there wasn't enough data
738 left).
739
740 5. Comprehensive information about what went wrong is now returned by
741 pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
742 as the output vector has at least 2 elements. The offset of the start of
743 the failing character and a reason code are placed in the vector.
744
745 6. When the UTF-8 string check fails for pcre_compile(), the offset that is
746 now returned is for the first byte of the failing character, instead of the
747 last byte inspected. This is an incompatible change, but I hope it is small
748 enough not to be a problem. It makes the returned offset consistent with
749 pcre_exec() and pcre_dfa_exec().
750
751 7. pcretest now gives a text phrase as well as the error number when
752 pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
753 failure, the offset and reason code are output.
754
755 8. When \R was used with a maximizing quantifier it failed to skip backwards
756 over a \r\n pair if the subsequent match failed. Instead, it just skipped
757 back over a single character (\n). This seems wrong (because it treated the
758 two characters as a single entity when going forwards), conflicts with the
759 documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
760 behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
761 has been changed.
762
763 9. Some internal refactoring has changed the processing so that the handling
764 of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
765 time (the PCRE_DOTALL option was changed this way some time ago: version
766 7.7 change 16). This has made it possible to abolish the OP_OPT op code,
767 which was always a bit of a fudge. It also means that there is one less
768 argument for the match() function, which reduces its stack requirements
769 slightly. This change also fixes an incompatibility with Perl: the pattern
770 (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
771
772 10. More internal refactoring has drastically reduced the number of recursive
773 calls to match() for possessively repeated groups such as (abc)++ when
774 using pcre_exec().
775
776 11. While implementing 10, a number of bugs in the handling of groups were
777 discovered and fixed:
778
779 (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
780 (a|)*(?1) gave a compile-time internal error.
781 ((a|)+)+ did not notice that the outer group could match an empty string.
782 (^a|^)+ was not marked as anchored.
783 (.*a|.*)+ was not marked as matching at start or after a newline.
784
785 12. Yet more internal refactoring has removed another argument from the match()
786 function. Special calls to this function are now indicated by setting a
787 value in a variable in the "match data" data block.
788
789 13. Be more explicit in pcre_study() instead of relying on "default" for
790 opcodes that mean there is no starting character; this means that when new
791 ones are added and accidentally left out of pcre_study(), testing should
792 pick them up.
793
794 14. The -s option of pcretest has been documented for ages as being an old
795 synonym of -m (show memory usage). I have changed it to mean "force study
796 for every regex", that is, assume /S for every regex. This is similar to -i
797 and -d etc. It's slightly incompatible, but I'm hoping nobody is still
798 using it. It makes it easier to run collections of tests with and without
799 study enabled, and thereby test pcre_study() more easily. All the standard
800 tests are now run with and without -s (but some patterns can be marked as
801 "never study" - see 20 below).
802
803 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
804 restoration of the capturing data to the outer values was not happening
805 correctly.
806
807 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
808 empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
809 pattern had matched an empty string, and so incorrectly returned a no
810 match.
811
812 17. There was optimizing code for the last branch of non-capturing parentheses,
813 and also for the obeyed branch of a conditional subexpression, which used
814 tail recursion to cut down on stack usage. Unfortunately, now that there is
815 the possibility of (*THEN) occurring in these branches, tail recursion is
816 no longer possible because the return has to be checked for (*THEN). These
817 two optimizations have therefore been removed. [But see 8.20/11 above.]
818
819 18. If a pattern containing \R was studied, it was assumed that \R always
820 matched two bytes, thus causing the minimum subject length to be
821 incorrectly computed because \R can also match just one byte.
822
823 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
824 was incorrectly computed.
825
826 20. If /S is present twice on a test pattern in pcretest input, it now
827 *disables* studying, thereby overriding the use of -s on the command line
828 (see 14 above). This is necessary for one or two tests to keep the output
829 identical in both cases.
830
831 21. When (*ACCEPT) was used in an assertion that matched an empty string and
832 PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
833
834 22. When an atomic group that contained a capturing parenthesis was
835 successfully matched, but the branch in which it appeared failed, the
836 capturing was not being forgotten if a higher numbered group was later
837 captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
838 group 1 to "a", when in fact it should be unset. This applied to multi-
839 branched capturing and non-capturing groups, repeated or not, and also to
840 positive assertions (capturing in negative assertions does not happen
841 in PCRE) and also to nested atomic groups.
842
843 23. Add the ++ qualifier feature to pcretest, to show the remainder of the
844 subject after a captured substring, to make it easier to tell which of a
845 number of identical substrings has been captured.
846
847 24. The way atomic groups are processed by pcre_exec() has been changed so that
848 if they are repeated, backtracking one repetition now resets captured
849 values correctly. For example, if ((?>(a+)b)+aabab) is matched against
850 "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
851 "aaa". Previously, it would have been "a". As part of this code
852 refactoring, the way recursive calls are handled has also been changed.
853
854 25. If an assertion condition captured any substrings, they were not passed
855 back unless some other capturing happened later. For example, if
856 (?(?=(a))a) was matched against "a", no capturing was returned.
857
858 26. When studying a pattern that contained subroutine calls or assertions,
859 the code for finding the minimum length of a possible match was handling
860 direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
861 group 1 called group 2 while simultaneously a separate group 2 called group
862 1). A stack overflow occurred in this case. I have fixed this by limiting
863 the recursion depth to 10.
864
865 27. Updated RunTest.bat in the distribution to the version supplied by Tom
866 Fortmann. This supports explicit test numbers on the command line, and has
867 argument validation and error reporting.
868
869 28. An instance of \X with an unlimited repeat could fail if at any point the
870 first character it looked at was a mark character.
871
872 29. Some minor code refactoring concerning Unicode properties and scripts
873 should reduce the stack requirement of match() slightly.
874
875 30. Added the '=' option to pcretest to check the setting of unused capturing
876 slots at the end of the pattern, which are documented as being -1, but are
877 not included in the return count.
878
879 31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
880 compiled something random. Now it gives a compile-time error (as does
881 Perl).
882
883 32. A *MARK encountered during the processing of a positive assertion is now
884 recorded and passed back (compatible with Perl).
885
886 33. If --only-matching or --colour was set on a pcregrep call whose pattern
887 had alternative anchored branches, the search for a second match in a line
888 was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
889 matched the line "0102" twice. The same bug affected patterns that started
890 with a backwards assertion. For example /\b01|\b02/ also matched "0102"
891 twice.
892
893 34. Previously, PCRE did not allow quantification of assertions. However, Perl
894 does, and because of capturing effects, quantifying parenthesized
895 assertions may at times be useful. Quantifiers are now allowed for
896 parenthesized assertions.
897
898 35. A minor code tidy in pcre_compile() when checking options for \R usage.
899
900 36. \g was being checked for fancy things in a character class, when it should
901 just be a literal "g".
902
903 37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
904 appearance of a nested POSIX class supersedes an apparent external class.
905 For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
906 unescaped square brackets may also appear as part of class names. For
907 example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
908 more like Perl. (But see 8.20/1 above.)
909
910 38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
911 was because it thought it was \N{name}, which is not supported).
912
913 39. Add minix to OS list not supporting the -S option in pcretest.
914
915 40. PCRE tries to detect cases of infinite recursion at compile time, but it
916 cannot analyze patterns in sufficient detail to catch mutual recursions
917 such as ((?1))((?2)). There is now a runtime test that gives an error if a
918 subgroup is called recursively as a subpattern for a second time at the
919 same position in the subject string. In previous releases this might have
920 been caught by the recursion limit, or it might have run out of stack.
921
922 41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
923 happen only once. PCRE was, however incorrectly giving a compile time error
924 "recursive call could loop indefinitely" because it cannot analyze the
925 pattern in sufficient detail. The compile time test no longer happens when
926 PCRE is compiling a conditional subpattern, but actual runaway loops are
927 now caught at runtime (see 40 above).
928
929 42. It seems that Perl allows any characters other than a closing parenthesis
930 to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
931 has been changed to be the same.
932
933 43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
934 as not to get warnings when autogen.sh is called. Also changed
935 AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
936
937 44. To help people who use pcregrep to scan files containing exceedingly long
938 lines, the following changes have been made:
939
940 (a) The default value of the buffer size parameter has been increased from
941 8K to 20K. (The actual buffer used is three times this size.)
942
943 (b) The default can be changed by ./configure --with-pcregrep-bufsize when
944 PCRE is built.
945
946 (c) A --buffer-size=n option has been added to pcregrep, to allow the size
947 to be set at run time.
948
949 (d) Numerical values in pcregrep options can be followed by K or M, for
950 example --buffer-size=50K.
951
952 (e) If a line being scanned overflows pcregrep's buffer, an error is now
953 given and the return code is set to 2.
954
955 45. Add a pointer to the latest mark to the callout data block.
956
957 46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
958 partial match of an empty string instead of no match. This was specific to
959 the use of ".".
960
961 47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
962 complete match instead of a partial match. This bug was dependent on both
963 the PCRE_UTF8 and PCRE_DOTALL options being set.
964
965 48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
966 starting byte set, because \b was not being ignored.
967
968
969 Version 8.12 15-Jan-2011
970 ------------------------
971
972 1. Fixed some typos in the markup of the man pages, and wrote a script that
973 checks for such things as part of the documentation building process.
974
975 2. On a big-endian 64-bit system, pcregrep did not correctly process the
976 --match-limit and --recursion-limit options (added for 8.11). In
977 particular, this made one of the standard tests fail. (The integer value
978 went into the wrong half of a long int.)
979
980 3. If the --colour option was given to pcregrep with -v (invert match), it
981 did strange things, either producing crazy output, or crashing. It should,
982 of course, ignore a request for colour when reporting lines that do not
983 match.
984
985 4. Another pcregrep bug caused similar problems if --colour was specified with
986 -M (multiline) and the pattern match finished with a line ending.
987
988 5. In pcregrep, when a pattern that ended with a literal newline sequence was
989 matched in multiline mode, the following line was shown as part of the
990 match. This seems wrong, so I have changed it.
991
992 6. Another pcregrep bug in multiline mode, when --colour was specified, caused
993 the check for further matches in the same line (so they could be coloured)
994 to overrun the end of the current line. If another match was found, it was
995 incorrectly shown (and then shown again when found in the next line).
996
997 7. If pcregrep was compiled under Windows, there was a reference to the
998 function pcregrep_exit() before it was defined. I am assuming this was
999 the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
1000 reported by a user. I've moved the definition above the reference.
1001
1002
1003 Version 8.11 10-Dec-2010
1004 ------------------------
1005
1006 1. (*THEN) was not working properly if there were untried alternatives prior
1007 to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
1008 backtracked to try for "b" instead of moving to the next alternative branch
1009 at the same level (in this case, to look for "c"). The Perl documentation
1010 is clear that when (*THEN) is backtracked onto, it goes to the "next
1011 alternative in the innermost enclosing group".
1012
1013 2. (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
1014 such as (A(*COMMIT)B(*THEN)C|D) any failure after matching A should
1015 result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
1016 (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
1017 (*THEN).
1018
1019 3. If \s appeared in a character class, it removed the VT character from
1020 the class, even if it had been included by some previous item, for example
1021 in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
1022 of \s, but is part of the POSIX "space" class.)
1023
1024 4. A partial match never returns an empty string (because you can always
1025 match an empty string at the end of the subject); however the checking for
1026 an empty string was starting at the "start of match" point. This has been
1027 changed to the "earliest inspected character" point, because the returned
1028 data for a partial match starts at this character. This means that, for
1029 example, /(?<=abc)def/ gives a partial match for the subject "abc"
1030 (previously it gave "no match").
1031
1032 5. Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
1033 of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
1034 previously a full match would be given. However, setting PCRE_PARTIAL_HARD
1035 has an implication that the given string is incomplete (because a partial
1036 match is preferred over a full match). For this reason, these items now
1037 give a partial match in this situation. [Aside: previously, the one case
1038 /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
1039 match rather than a full match, which was wrong by the old rules, but is
1040 now correct.]
1041
1042 6. There was a bug in the handling of #-introduced comments, recognized when
1043 PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
1044 If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
1045 UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
1046 scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
1047 but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
1048 places in pcre_compile().
1049
1050 7. Related to (6) above, when pcre_compile() was skipping #-introduced
1051 comments when looking ahead for named forward references to subpatterns,
1052 the only newline sequence it recognized was NL. It now handles newlines
1053 according to the set newline convention.
1054
1055 8. SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
1056 former, but used strtoul(), whereas pcretest avoided strtoul() but did not
1057 cater for a lack of strerror(). These oversights have been fixed.
1058
1059 9. Added --match-limit and --recursion-limit to pcregrep.
1060
1061 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
1062
1063 11. When the -o option was used, pcregrep was setting a return code of 1, even
1064 when matches were found, and --line-buffered was not being honoured.
1065
1066 12. Added an optional parentheses number to the -o and --only-matching options
1067 of pcregrep.
1068
1069 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
1070 can match an empty string. The code to do it in pcretest and pcredemo
1071 needed fixing:
1072
1073 (a) When the newline convention was "crlf", pcretest got it wrong, skipping
1074 only one byte after an empty string match just before CRLF (this case
1075 just got forgotten; "any" and "anycrlf" were OK).
1076
1077 (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
1078 mode when an empty string match preceded an ASCII character followed by
1079 a non-ASCII character. (The code for advancing by one character rather
1080 than one byte was nonsense.)
1081
1082 (c) The pcredemo.c sample program did not have any code at all to handle
1083 the cases when CRLF is a valid newline sequence.
1084
1085 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
1086 as a starting offset was within the subject string. There is now a new
1087 error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
1088 negative or greater than the length of the string. In order to test this,
1089 pcretest is extended to allow the setting of negative starting offsets.
1090
1091 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
1092 starting offset points to the beginning of a UTF-8 character was
1093 unnecessarily clumsy. I tidied it up.
1094
1095 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
1096 bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
1097
1098 17. Nobody had reported that the --include_dir option, which was added in
1099 release 7.7 should have been called --include-dir (hyphen, not underscore)
1100 for compatibility with GNU grep. I have changed it to --include-dir, but
1101 left --include_dir as an undocumented synonym, and the same for
1102 --exclude-dir, though that is not available in GNU grep, at least as of
1103 release 2.5.4.
1104
1105 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
1106 characters from a string of bytes) have been redefined so as not to use
1107 loops, in order to improve performance in some environments. At the same
1108 time, I abstracted some of the common code into auxiliary macros to save
1109 repetition (this should not affect the compiled code).
1110
1111 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
1112 compile-time error is now given if \c is not followed by an ASCII
1113 character, that is, a byte less than 128. (In EBCDIC mode, the code is
1114 different, and any byte value is allowed.)
1115
1116 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
1117 START_OPTIMIZE option, which is now allowed at compile time - but just
1118 passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
1119 to pcregrep and other applications that have no direct access to PCRE
1120 options. The new /Y option in pcretest sets this option when calling
1121 pcre_compile().
1122
1123 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
1124 back references. Groups containing recursive back references were forced to
1125 be atomic by that change, but in the case of named groups, the amount of
1126 memory required was incorrectly computed, leading to "Failed: internal
1127 error: code overflow". This has been fixed.
1128
1129 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
1130 pcretest.c, to avoid build problems in some Borland environments.
1131
1132
1133 Version 8.10 25-Jun-2010
1134 ------------------------
1135
1136 1. Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
1137 THEN.
1138
1139 2. (*ACCEPT) was not working when inside an atomic group.
1140
1141 3. Inside a character class, \B is treated as a literal by default, but
1142 faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
1143 causes the error). The code is unchanged, but I tidied the documentation.
1144
1145 4. Inside a character class, PCRE always treated \R and \X as literals,
1146 whereas Perl faults them if its -w option is set. I have changed PCRE so
1147 that it faults them when PCRE_EXTRA is set.
1148
1149 5. Added support for \N, which always matches any character other than
1150 newline. (It is the same as "." when PCRE_DOTALL is not set.)
1151
1152 6. When compiling pcregrep with newer versions of gcc which may have
1153 FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
1154 declared with attribute warn_unused_result" were given. Just casting the
1155 result to (void) does not stop the warnings; a more elaborate fudge is
1156 needed. I've used a macro to implement this.
1157
1158 7. Minor change to pcretest.c to avoid a compiler warning.
1159
1160 8. Added four artifical Unicode properties to help with an option to make
1161 \s etc use properties (see next item). The new properties are: Xan
1162 (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
1163
1164 9. Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
1165 use Unicode properties. (*UCP) at the start of a pattern can be used to set
1166 this option. Modified pcretest to add /W to test this facility. Added
1167 REG_UCP to make it available via the POSIX interface.
1168
1169 10. Added --line-buffered to pcregrep.
1170
1171 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
1172 studied, and the match started with a letter with a code point greater than
1173 127 whose first byte was different to the first byte of the other case of
1174 the letter, the other case of this starting letter was not recognized
1175 (#976).
1176
1177 12. If a pattern that was studied started with a repeated Unicode property
1178 test, for example, \p{Nd}+, there was the theoretical possibility of
1179 setting up an incorrect bitmap of starting bytes, but fortunately it could
1180 not have actually happened in practice until change 8 above was made (it
1181 added property types that matched character-matching opcodes).
1182
1183 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
1184 possible starting bytes for non-anchored patterns.
1185
1186 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
1187 \R, and also a number of cases that involve Unicode properties, both
1188 explicit and implicit when PCRE_UCP is set.
1189
1190 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
1191 input, it could crash or give wrong results if characters with values
1192 greater than 0xc0 were present in the subject string. (Detail: it assumed
1193 UTF-8 input when processing these items.)
1194
1195 16. Added a lot of (int) casts to avoid compiler warnings in systems where
1196 size_t is 64-bit (#991).
1197
1198 17. Added a check for running out of memory when PCRE is compiled with
1199 --disable-stack-for-recursion (#990).
1200
1201 18. If the last data line in a file for pcretest does not have a newline on
1202 the end, a newline was missing in the output.
1203
1204 19. The default pcre_chartables.c file recognizes only ASCII characters (values
1205 less than 128) in its various bitmaps. However, there is a facility for
1206 generating tables according to the current locale when PCRE is compiled. It
1207 turns out that in some environments, 0x85 and 0xa0, which are Unicode space
1208 characters, are recognized by isspace() and therefore were getting set in
1209 these tables, and indeed these tables seem to approximate to ISO 8859. This
1210 caused a problem in UTF-8 mode when pcre_study() was used to create a list
1211 of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
1212 which of course cannot start UTF-8 characters. I have changed the code so
1213 that only real ASCII characters (less than 128) and the correct starting
1214 bytes for UTF-8 encodings are set for characters greater than 127 when in
1215 UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
1216 altogether.)
1217
1218 20. Added the /T option to pcretest so as to be able to run tests with non-
1219 standard character tables, thus making it possible to include the tests
1220 used for 19 above in the standard set of tests.
1221
1222 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
1223 reference to a subpattern the other side of a comment that contains an
1224 opening parenthesis caused either an internal compiling error, or a
1225 reference to the wrong subpattern.
1226
1227
1228 Version 8.02 19-Mar-2010
1229 ------------------------
1230
1231 1. The Unicode data tables have been updated to Unicode 5.2.0.
1232
1233 2. Added the option --libs-cpp to pcre-config, but only when C++ support is
1234 configured.
1235
1236 3. Updated the licensing terms in the pcregexp.pas file, as agreed with the
1237 original author of that file, following a query about its status.
1238
1239 4. On systems that do not have stdint.h (e.g. Solaris), check for and include
1240 inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
1241
1242 5. A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
1243 quantifier applied to a forward-referencing subroutine call, could compile
1244 incorrect code or give the error "internal error: previously-checked
1245 referenced subpattern not found".
1246
1247 6. Both MS Visual Studio and Symbian OS have problems with initializing
1248 variables to point to external functions. For these systems, therefore,
1249 pcre_malloc etc. are now initialized to local functions that call the
1250 relevant global functions.
1251
1252 7. There were two entries missing in the vectors called coptable and poptable
1253 in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
1254 I've fixed the data, and added a kludgy way of testing at compile time that
1255 the lengths are correct (equal to the number of opcodes).
1256
1257 8. Following on from 7, I added a similar kludge to check the length of the
1258 eint vector in pcreposix.c.
1259
1260 9. Error texts for pcre_compile() are held as one long string to avoid too
1261 much relocation at load time. To find a text, the string is searched,
1262 counting zeros. There was no check for running off the end of the string,
1263 which could happen if a new error number was added without updating the
1264 string.
1265
1266 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
1267
1268 11. \K was not working if it appeared in an atomic group or in a group that
1269 was called as a "subroutine", or in an assertion. Perl 5.11 documents that
1270 \K is "not well defined" if used in an assertion. PCRE now accepts it if
1271 the assertion is positive, but not if it is negative.
1272
1273 12. Change 11 fortuitously reduced the size of the stack frame used in the
1274 "match()" function of pcre_exec.c by one pointer. Forthcoming
1275 implementation of support for (*MARK) will need an extra pointer on the
1276 stack; I have reserved it now, so that the stack frame size does not
1277 decrease.
1278
1279 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
1280 item in branch that calls a recursion is a subroutine call - as in the
1281 second branch in the above example - was incorrectly given the compile-
1282 time error "recursive call could loop indefinitely" because pcre_compile()
1283 was not correctly checking the subroutine for matching a non-empty string.
1284
1285 14. The checks for overrunning compiling workspace could trigger after an
1286 overrun had occurred. This is a "should never occur" error, but it can be
1287 triggered by pathological patterns such as hundreds of nested parentheses.
1288 The checks now trigger 100 bytes before the end of the workspace.
1289
1290 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
1291
1292
1293 Version 8.01 19-Jan-2010
1294 ------------------------
1295
1296 1. If a pattern contained a conditional subpattern with only one branch (in
1297 particular, this includes all (*DEFINE) patterns), a call to pcre_study()
1298 computed the wrong minimum data length (which is of course zero for such
1299 subpatterns). This could cause incorrect "no match" results.
1300
1301 2. For patterns such as (?i)a(?-i)b|c where an option setting at the start of
1302 the pattern is reset in the first branch, pcre_compile() failed with
1303 "internal error: code overflow at offset...". This happened only when
1304 the reset was to the original external option setting. (An optimization
1305 abstracts leading options settings into an external setting, which was the
1306 cause of this.)
1307
1308 3. A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
1309 of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
1310 assertion pattern did not match (meaning that the assertion was true), it
1311 was incorrectly treated as false if the SKIP had been reached during the
1312 matching. This also applied to assertions used as conditions.
1313
1314 4. If an item that is not supported by pcre_dfa_exec() was encountered in an
1315 assertion subpattern, including such a pattern used as a condition,
1316 unpredictable results occurred, instead of the error return
1317 PCRE_ERROR_DFA_UITEM.
1318
1319 5. The C++ GlobalReplace function was not working like Perl for the special
1320 situation when an empty string is matched. It now does the fancy magic
1321 stuff that is necessary.
1322
1323 6. In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
1324 removed. (These were left over from very, very early versions of PCRE.)
1325
1326 7. Some cosmetic changes to the code to make life easier when compiling it
1327 as part of something else:
1328
1329 (a) Change DEBUG to PCRE_DEBUG.
1330
1331 (b) In pcre_compile(), rename the member of the "branch_chain" structure
1332 called "current" as "current_branch", to prevent a collision with the
1333 Linux macro when compiled as a kernel module.
1334
1335 (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
1336 prevent a collision with the Linux macro when compiled as a kernel
1337 module.
1338
1339 8. In pcre_compile() there are some checks for integer overflows that used to
1340 cast potentially large values to (double). This has been changed to that
1341 when building, a check for int64_t is made, and if it is found, it is used
1342 instead, thus avoiding the use of floating point arithmetic. (There is no
1343 other use of FP in PCRE.) If int64_t is not found, the fallback is to
1344 double.
1345
1346 9. Added two casts to avoid signed/unsigned warnings from VS Studio Express
1347 2005 (difference between two addresses compared to an unsigned value).
1348
1349 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
1350 custom one, because of the following reported problem in Windows:
1351
1352 - libbz2 uses the Pascal calling convention (WINAPI) for the functions
1353 under Win32.
1354 - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
1355 therefore missing the function definition.
1356 - The compiler thus generates a "C" signature for the test function.
1357 - The linker fails to find the "C" function.
1358 - PCRE fails to configure if asked to do so against libbz2.
1359
1360 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
1361 messages were output:
1362
1363 Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
1364 rerunning libtoolize, to keep the correct libtool macros in-tree.
1365 Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
1366
1367 I have done both of these things.
1368
1369 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
1370 most of the time, it *can* run out if it is given a pattern that contains a
1371 runaway infinite recursion. I updated the discussion in the pcrestack man
1372 page.
1373
1374 13. Now that we have gone to the x.xx style of version numbers, the minor
1375 version may start with zero. Using 08 or 09 is a bad idea because users
1376 might check the value of PCRE_MINOR in their code, and 08 or 09 may be
1377 interpreted as invalid octal numbers. I've updated the previous comment in
1378 configure.ac, and also added a check that gives an error if 08 or 09 are
1379 used.
1380
1381 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
1382 causing partial matching to fail when the end of the subject matched \W
1383 in a UTF-8 pattern where \W was quantified with a minimum of 3.
1384
1385 15. There were some discrepancies between the declarations in pcre_internal.h
1386 of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
1387 their definitions. The declarations used "const uschar *" and the
1388 definitions used USPTR. Even though USPTR is normally defined as "const
1389 unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
1390 reported that: "This difference in casting confuses some C++ compilers, for
1391 example, SunCC recognizes above declarations as different functions and
1392 generates broken code for hbpcre." I have changed the declarations to use
1393 USPTR.
1394
1395 16. GNU libtool is named differently on some systems. The autogen.sh script now
1396 tries several variants such as glibtoolize (MacOSX) and libtoolize1x
1397 (FreeBSD).
1398
1399 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
1400 (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
1401 comment: "Figure out how to create a longlong from a string: strtoll and
1402 equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
1403 instance, but it only takes 2 args instead of 3!"
1404
1405 18. A subtle bug concerned with back references has been fixed by a change of
1406 specification, with a corresponding code fix. A pattern such as
1407 ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
1408 refers, was giving matches when it shouldn't. For example, xa=xaaa would
1409 match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
1410 same bug. Such groups have to be quantified to be useful, or contained
1411 inside another quantified group. (If there's no repetition, the reference
1412 can never match.) The problem arises because, having left the group and
1413 moved on to the rest of the pattern, a later failure that backtracks into
1414 the group uses the captured value from the final iteration of the group
1415 rather than the correct earlier one. I have fixed this in PCRE by forcing
1416 any group that contains a reference to itself to be an atomic group; that
1417 is, there cannot be any backtracking into it once it has completed. This is
1418 similar to recursive and subroutine calls.
1419
1420
1421 Version 8.00 19-Oct-09
1422 ----------------------
1423
1424 1. The table for translating pcre_compile() error codes into POSIX error codes
1425 was out-of-date, and there was no check on the pcre_compile() error code
1426 being within the table. This could lead to an OK return being given in
1427 error.
1428
1429 2. Changed the call to open a subject file in pcregrep from fopen(pathname,
1430 "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
1431 in a Windows environment.
1432
1433 3. The pcregrep --count option prints the count for each file even when it is
1434 zero, as does GNU grep. However, pcregrep was also printing all files when
1435 --files-with-matches was added. Now, when both options are given, it prints
1436 counts only for those files that have at least one match. (GNU grep just
1437 prints the file name in this circumstance, but including the count seems
1438 more useful - otherwise, why use --count?) Also ensured that the
1439 combination -clh just lists non-zero counts, with no names.
1440
1441 4. The long form of the pcregrep -F option was incorrectly implemented as
1442 --fixed_strings instead of --fixed-strings. This is an incompatible change,
1443 but it seems right to fix it, and I didn't think it was worth preserving
1444 the old behaviour.
1445
1446 5. The command line items --regex=pattern and --regexp=pattern were not
1447 recognized by pcregrep, which required --regex pattern or --regexp pattern
1448 (with a space rather than an '='). The man page documented the '=' forms,
1449 which are compatible with GNU grep; these now work.
1450
1451 6. No libpcreposix.pc file was created for pkg-config; there was just
1452 libpcre.pc and libpcrecpp.pc. The omission has been rectified.
1453
1454 7. Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
1455 when UCP support is not needed, by modifying the Python script that
1456 generates it from Unicode data files. This should not matter if the module
1457 is correctly used as a library, but I received one complaint about 50K of
1458 unwanted data. My guess is that the person linked everything into his
1459 program rather than using a library. Anyway, it does no harm.
1460
1461 8. A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
1462 was a minimum greater than 1 for a wide character in a possessive
1463 repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
1464 which had an unlimited repeat of a nested, fixed maximum repeat of a wide
1465 character. Chaos in the form of incorrect output or a compiling loop could
1466 result.
1467
1468 9. The restrictions on what a pattern can contain when partial matching is
1469 requested for pcre_exec() have been removed. All patterns can now be
1470 partially matched by this function. In addition, if there are at least two
1471 slots in the offset vector, the offset of the earliest inspected character
1472 for the match and the offset of the end of the subject are set in them when
1473 PCRE_ERROR_PARTIAL is returned.
1474
1475 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
1476 synonymous with PCRE_PARTIAL, for backwards compatibility, and
1477 PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
1478 and may be more useful for multi-segment matching.
1479
1480 11. Partial matching with pcre_exec() is now more intuitive. A partial match
1481 used to be given if ever the end of the subject was reached; now it is
1482 given only if matching could not proceed because another character was
1483 needed. This makes a difference in some odd cases such as Z(*FAIL) with the
1484 string "Z", which now yields "no match" instead of "partial match". In the
1485 case of pcre_dfa_exec(), "no match" is given if every matching path for the
1486 final character ended with (*FAIL).
1487
1488 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
1489 if the pattern had a "must contain" character that was already found in the
1490 earlier partial match, unless partial matching was again requested. For
1491 example, with the pattern /dog.(body)?/, the "must contain" character is
1492 "g". If the first part-match was for the string "dog", restarting with
1493 "sbody" failed. This bug has been fixed.
1494
1495 13. The string returned by pcre_dfa_exec() after a partial match has been
1496 changed so that it starts at the first inspected character rather than the
1497 first character of the match. This makes a difference only if the pattern
1498 starts with a lookbehind assertion or \b or \B (\K is not supported by
1499 pcre_dfa_exec()). It's an incompatible change, but it makes the two
1500 matching functions compatible, and I think it's the right thing to do.
1501
1502 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
1503 so that the demonstration program is easily available in environments where
1504 PCRE has not been installed from source.
1505
1506 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
1507 libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
1508 library.
1509
1510 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
1511 It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
1512 is not the first non-POSIX option to be added. Clearly some people find
1513 these options useful.
1514
1515 17. If a caller to the POSIX matching function regexec() passes a non-zero
1516 value for nmatch with a NULL value for pmatch, the value of
1517 nmatch is forced to zero.
1518
1519 18. RunGrepTest did not have a test for the availability of the -u option of
1520 the diff command, as RunTest does. It now checks in the same way as
1521 RunTest, and also checks for the -b option.
1522
1523 19. If an odd number of negated classes containing just a single character
1524 interposed, within parentheses, between a forward reference to a named
1525 subpattern and the definition of the subpattern, compilation crashed with
1526 an internal error, complaining that it could not find the referenced
1527 subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
1528 [The bug was that it was starting one character too far in when skipping
1529 over the character class, thus treating the ] as data rather than
1530 terminating the class. This meant it could skip too much.]
1531
1532 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
1533 /g option in pcretest when the pattern contains \K, which makes it possible
1534 to have an empty string match not at the start, even when the pattern is
1535 anchored. Updated pcretest and pcredemo to use this option.
1536
1537 21. If the maximum number of capturing subpatterns in a recursion was greater
1538 than the maximum at the outer level, the higher number was returned, but
1539 with unset values at the outer level. The correct (outer level) value is
1540 now given.
1541
1542 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
1543 PCRE did not set those parentheses (unlike Perl). I have now found a way to
1544 make it do so. The string so far is captured, making this feature
1545 compatible with Perl.
1546
1547 23. The tests have been re-organized, adding tests 11 and 12, to make it
1548 possible to check the Perl 5.10 features against Perl 5.10.
1549
1550 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
1551 pattern matches a fixed length string. PCRE did not allow this; now it
1552 does. Neither allows recursion.
1553
1554 25. I finally figured out how to implement a request to provide the minimum
1555 length of subject string that was needed in order to match a given pattern.
1556 (It was back references and recursion that I had previously got hung up
1557 on.) This code has now been added to pcre_study(); it finds a lower bound
1558 to the length of subject needed. It is not necessarily the greatest lower
1559 bound, but using it to avoid searching strings that are too short does give
1560 some useful speed-ups. The value is available to calling programs via
1561 pcre_fullinfo().
1562
1563 26. While implementing 25, I discovered to my embarrassment that pcretest had
1564 not been passing the result of pcre_study() to pcre_dfa_exec(), so the
1565 study optimizations had never been tested with that matching function.
1566 Oops. What is worse, even when it was passed study data, there was a bug in
1567 pcre_dfa_exec() that meant it never actually used it. Double oops. There
1568 were also very few tests of studied patterns with pcre_dfa_exec().
1569
1570 27. If (?| is used to create subpatterns with duplicate numbers, they are now
1571 allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
1572 on the other side of the coin, they are no longer allowed to have different
1573 names, because these cannot be distinguished in PCRE, and this has caused
1574 confusion. (This is a difference from Perl.)
1575
1576 28. When duplicate subpattern names are present (necessarily with different
1577 numbers, as required by 27 above), and a test is made by name in a
1578 conditional pattern, either for a subpattern having been matched, or for
1579 recursion in such a pattern, all the associated numbered subpatterns are
1580 tested, and the overall condition is true if the condition is true for any
1581 one of them. This is the way Perl works, and is also more like the way
1582 testing by number works.
1583
1584
1585 Version 7.9 11-Apr-09
1586 ---------------------
1587
1588 1. When building with support for bzlib/zlib (pcregrep) and/or readline
1589 (pcretest), all targets were linked against these libraries. This included
1590 libpcre, libpcreposix, and libpcrecpp, even though they do not use these
1591 libraries. This caused unwanted dependencies to be created. This problem
1592 has been fixed, and now only pcregrep is linked with bzlib/zlib and only
1593 pcretest is linked with readline.
1594
1595 2. The "typedef int BOOL" in pcre_internal.h that was included inside the
1596 "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
1597 moved outside it again, because FALSE and TRUE are already defined in AIX,
1598 but BOOL is not.
1599
1600 3. The pcre_config() function was treating the PCRE_MATCH_LIMIT and
1601 PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
1602
1603 4. The pcregrep documentation said spaces were inserted as well as colons (or
1604 hyphens) following file names and line numbers when outputting matching
1605 lines. This is not true; no spaces are inserted. I have also clarified the
1606 wording for the --colour (or --color) option.
1607
1608 5. In pcregrep, when --colour was used with -o, the list of matching strings
1609 was not coloured; this is different to GNU grep, so I have changed it to be
1610 the same.
1611
1612 6. When --colo(u)r was used in pcregrep, only the first matching substring in
1613 each matching line was coloured. Now it goes on to look for further matches
1614 of any of the test patterns, which is the same behaviour as GNU grep.
1615
1616 7. A pattern that could match an empty string could cause pcregrep to loop; it
1617 doesn't make sense to accept an empty string match in pcregrep, so I have
1618 locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
1619 seems to be how GNU grep behaves.
1620
1621 8. The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
1622 start or after a newline", because the conditional assertion was not being
1623 correctly handled. The rule now is that both the assertion and what follows
1624 in the first alternative must satisfy the test.
1625
1626 9. If auto-callout was enabled in a pattern with a conditional group whose
1627 condition was an assertion, PCRE could crash during matching, both with
1628 pcre_exec() and pcre_dfa_exec().
1629
1630 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
1631 used for matching.
1632
1633 11. Unicode property support in character classes was not working for
1634 characters (bytes) greater than 127 when not in UTF-8 mode.
1635
1636 12. Added the -M command line option to pcretest.
1637
1638 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
1639
1640 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
1641
1642 16. Added comments and documentation about mis-use of no_arg in the C++
1643 wrapper.
1644
1645 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
1646 from Martin Jerabek that uses macro names for all relevant character and
1647 string constants.
1648
1649 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
1650 SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
1651 SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
1652 these, but not everybody uses configure.
1653
1654 19. A conditional group that had only one branch was not being correctly
1655 recognized as an item that could match an empty string. This meant that an
1656 enclosing group might also not be so recognized, causing infinite looping
1657 (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
1658 with the subject "ab", where knowledge that the repeated group can match
1659 nothing is needed in order to break the loop.
1660
1661 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
1662 exec(), but without supplying a callout function, matching went wrong.
1663
1664 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
1665 leak if the size of the offset vector was greater than 30. When the vector
1666 is smaller, the saved offsets during recursion go onto a local stack
1667 vector, but for larger vectors malloc() is used. It was failing to free
1668 when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
1669 error, in fact).
1670
1671 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
1672 heapframe that is used only when UTF-8 support is enabled. This caused no
1673 problem, but was untidy.
1674
1675 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
1676 CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
1677 included within another project.
1678
1679 24. Steven Van Ingelgem's patches to add more options to the CMake support,
1680 slightly modified by me:
1681
1682 (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
1683 not building pcregrep.
1684
1685 (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
1686 if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
1687
1688 25. Forward references, both numeric and by name, in patterns that made use of
1689 duplicate group numbers, could behave incorrectly or give incorrect errors,
1690 because when scanning forward to find the reference group, PCRE was not
1691 taking into account the duplicate group numbers. A pattern such as
1692 ^X(?3)(a)(?|(b)|(q))(Y) is an example.
1693
1694 26. Changed a few more instances of "const unsigned char *" to USPTR, making
1695 the feature of a custom pointer more persuasive (as requested by a user).
1696
1697 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
1698 pcretest.c, inside #ifndefs, because it seems they are sometimes already
1699 pre-defined.
1700
1701 28. Added support for (*UTF8) at the start of a pattern.
1702
1703 29. Arrange for flags added by the "release type" setting in CMake to be shown
1704 in the configuration summary.
1705
1706
1707 Version 7.8 05-Sep-08
1708 ---------------------
1709
1710 1. Replaced UCP searching code with optimized version as implemented for Ad
1711 Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
1712 stage table and inline lookup instead of a function, giving speed ups of 2
1713 to 5 times on some simple patterns that I tested. Permission was given to
1714 distribute the MultiStage2.py script that generates the tables (it's not in
1715 the tarball, but is in the Subversion repository).
1716
1717 2. Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
1718 scripts.
1719
1720 3. Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
1721 a group with a zero qualifier. The result of the study could be incorrect,
1722 or the function might crash, depending on the pattern.
1723
1724 4. Caseless matching was not working for non-ASCII characters in back
1725 references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
1726 It now works when Unicode Property Support is available.
1727
1728 5. In pcretest, an escape such as \x{de} in the data was always generating
1729 a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
1730 non-UTF-8 mode. If the value is greater than 255, it gives a warning about
1731 truncation.
1732
1733 6. Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
1734
1735 7. Added two (int) casts to pcregrep when printing the difference of two
1736 pointers, in case they are 64-bit values.
1737
1738 8. Added comments about Mac OS X stack usage to the pcrestack man page and to
1739 test 2 if it fails.
1740
1741 9. Added PCRE_CALL_CONVENTION just before the names of all exported functions,
1742 and a #define of that name to empty if it is not externally set. This is to
1743 allow users of MSVC to set it if necessary.
1744
1745 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
1746 the convenience functions in the pcre_get.c source file.
1747
1748 11. An option change at the start of a pattern that had top-level alternatives
1749 could cause overwriting and/or a crash. This command provoked a crash in
1750 some environments:
1751
1752 printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
1753
1754 This potential security problem was recorded as CVE-2008-2371.
1755
1756 12. For a pattern where the match had to start at the beginning or immediately
1757 after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
1758 pcre_dfa_exec() could read past the end of the passed subject if there was
1759 no match. To help with detecting such bugs (e.g. with valgrind), I modified
1760 pcretest so that it places the subject at the end of its malloc-ed buffer.
1761
1762 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
1763 exec() might read past the end of the data buffer in UTF-8 mode.
1764
1765 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
1766 the data contained the byte 0x85 as part of a UTF-8 character within its
1767 first line. This applied both to normal and DFA matching.
1768
1769 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
1770 /^[^d]*?$/8 failed to match "abc".
1771
1772 16. Added a missing copyright notice to pcrecpp_internal.h.
1773
1774 17. Make it more clear in the documentation that values returned from
1775 pcre_exec() in ovector are byte offsets, not character counts.
1776
1777 18. Tidied a few places to stop certain compilers from issuing warnings.
1778
1779 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
1780 supplied by Stefan Weber. I made a further small update for 7.8 because
1781 there is a change of source arrangements: the pcre_searchfuncs.c module is
1782 replaced by pcre_ucd.c.
1783
1784
1785 Version 7.7 07-May-08
1786 ---------------------
1787
1788 1. Applied Craig's patch to sort out a long long problem: "If we can't convert
1789 a string to a long long, pretend we don't even have a long long." This is
1790 done by checking for the strtoq, strtoll, and _strtoi64 functions.
1791
1792 2. Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
1793 pre-7.6 versions, which defined a global no_arg variable instead of putting
1794 it in the RE class. (See also #8 below.)
1795
1796 3. Remove a line of dead code, identified by coverity and reported by Nuno
1797 Lopes.
1798
1799 4. Fixed two related pcregrep bugs involving -r with --include or --exclude:
1800
1801 (1) The include/exclude patterns were being applied to the whole pathnames
1802 of files, instead of just to the final components.
1803
1804 (2) If there was more than one level of directory, the subdirectories were
1805 skipped unless they satisfied the include/exclude conditions. This is
1806 inconsistent with GNU grep (and could even be seen as contrary to the
1807 pcregrep specification - which I improved to make it absolutely clear).
1808 The action now is always to scan all levels of directory, and just
1809 apply the include/exclude patterns to regular files.
1810
1811 5. Added the --include_dir and --exclude_dir patterns to pcregrep, and used
1812 --exclude_dir in the tests to avoid scanning .svn directories.
1813
1814 6. Applied Craig's patch to the QuoteMeta function so that it escapes the
1815 NUL character as backslash + 0 rather than backslash + NUL, because PCRE
1816 doesn't support NULs in patterns.
1817
1818 7. Added some missing "const"s to declarations of static tables in
1819 pcre_compile.c and pcre_dfa_exec.c.
1820
1821 8. Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
1822 caused by fix #2 above. (Subsequently also a second patch to fix the
1823 first patch. And a third patch - this was a messy problem.)
1824
1825 9. Applied Craig's patch to remove the use of push_back().
1826
1827 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
1828 matching function regexec().
1829
1830 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
1831 which, however, unlike Perl's \g{...}, are subroutine calls, not back
1832 references. PCRE supports relative numbers with this syntax (I don't think
1833 Oniguruma does).
1834
1835 12. Previously, a group with a zero repeat such as (...){0} was completely
1836 omitted from the compiled regex. However, this means that if the group
1837 was called as a subroutine from elsewhere in the pattern, things went wrong
1838 (an internal error was given). Such groups are now left in the compiled
1839 pattern, with a new opcode that causes them to be skipped at execution
1840 time.
1841
1842 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
1843 to the way PCRE behaves:
1844
1845 (a) A lone ] character is dis-allowed (Perl treats it as data).
1846
1847 (b) A back reference to an unmatched subpattern matches an empty string
1848 (Perl fails the current match path).
1849
1850 (c) A data ] in a character class must be notated as \] because if the
1851 first data character in a class is ], it defines an empty class. (In
1852 Perl it is not possible to have an empty class.) The empty class []
1853 never matches; it forces failure and is equivalent to (*FAIL) or (?!).
1854 The negative empty class [^] matches any one character, independently
1855 of the DOTALL setting.
1856
1857 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
1858 non-existent subpattern following a character class starting with ']' and
1859 containing () gave an internal compiling error instead of "reference to
1860 non-existent subpattern". Fortunately, when the pattern did exist, the
1861 compiled code was correct. (When scanning forwards to check for the
1862 existencd of the subpattern, it was treating the data ']' as terminating
1863 the class, so got the count wrong. When actually compiling, the reference
1864 was subsequently set up correctly.)
1865
1866 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
1867 it was being rejected as not supported by pcre_dfa_exec(), even though
1868 other assertions are supported. I have made pcre_dfa_exec() support
1869 (*FAIL).
1870
1871 16. The implementation of 13c above involved the invention of a new opcode,
1872 OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
1873 cannot be changed at match time, I realized I could make a small
1874 improvement to matching performance by compiling OP_ALLANY instead of
1875 OP_ANY for "." when DOTALL was set, and then removing the runtime tests
1876 on the OP_ANY path.
1877
1878 17. Compiling pcretest on Windows with readline support failed without the
1879 following two fixes: (1) Make the unistd.h include conditional on
1880 HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
1881
1882 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
1883 ncurses library to be included for pcretest when ReadLine support is
1884 requested, but also to allow for it to be overridden. This patch came from
1885 Daniel Bergström.
1886
1887 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
1888 as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
1889 any errors with the current Unicode tables. Thanks to Peter Kankowski for
1890 spotting this.
1891
1892
1893 Version 7.6 28-Jan-08
1894 ---------------------
1895
1896 1. A character class containing a very large number of characters with
1897 codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
1898 overflow.
1899
1900 2. Patch to cut out the "long long" test in pcrecpp_unittest when
1901 HAVE_LONG_LONG is not defined.
1902
1903 3. Applied Christian Ehrlicher's patch to update the CMake build files to
1904 bring them up to date and include new features. This patch includes:
1905
1906 - Fixed PH's badly added libz and libbz2 support.
1907 - Fixed a problem with static linking.
1908 - Added pcredemo. [But later removed - see 7 below.]
1909 - Fixed dftables problem and added an option.
1910 - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
1911 HAVE_LONG_LONG.
1912 - Added readline support for pcretest.
1913 - Added an listing of the option settings after cmake has run.
1914
1915 4. A user submitted a patch to Makefile that makes it easy to create
1916 "pcre.dll" under mingw when using Configure/Make. I added stuff to
1917 Makefile.am that cause it to include this special target, without
1918 affecting anything else. Note that the same mingw target plus all
1919 the other distribution libraries and programs are now supported
1920 when configuring with CMake (see 6 below) instead of with
1921 Configure/Make.
1922
1923 5. Applied Craig's patch that moves no_arg into the RE class in the C++ code.
1924 This is an attempt to solve the reported problem "pcrecpp::no_arg is not
1925 exported in the Windows port". It has not yet been confirmed that the patch
1926 solves the problem, but it does no harm.
1927
1928 6. Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
1929 NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
1930 with CMake, and also correct the comment about stack recursion.
1931
1932 7. Remove the automatic building of pcredemo from the ./configure system and
1933 from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
1934 of a program that users should build themselves after PCRE is installed, so
1935 building it automatically is not really right. What is more, it gave
1936 trouble in some build environments.
1937
1938 8. Further tidies to CMakeLists.txt from Sheri and Christian.
1939
1940
1941 Version 7.5 10-Jan-08
1942 ---------------------
1943
1944 1. Applied a patch from Craig: "This patch makes it possible to 'ignore'
1945 values in parens when parsing an RE using the C++ wrapper."
1946
1947 2. Negative specials like \S did not work in character classes in UTF-8 mode.
1948 Characters greater than 255 were excluded from the class instead of being
1949 included.
1950
1951 3. The same bug as (2) above applied to negated POSIX classes such as
1952 [:^space:].
1953
1954 4. PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
1955 defined or documented. It seems to have been a typo for PCRE_STATIC, so
1956 I have changed it.
1957
1958 5. The construct (?&) was not diagnosed as a syntax error (it referenced the
1959 first named subpattern) and a construct such as (?&a) would reference the
1960 first named subpattern whose name started with "a" (in other words, the
1961 length check was missing). Both these problems are fixed. "Subpattern name
1962 expected" is now given for (?&) (a zero-length name), and this patch also
1963 makes it give the same error for \k'' (previously it complained that that
1964 was a reference to a non-existent subpattern).
1965
1966 6. The erroneous patterns (?+-a) and (?-+a) give different error messages;
1967 this is right because (?- can be followed by option settings as well as by
1968 digits. I have, however, made the messages clearer.
1969
1970 7. Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
1971 than the number used in the conditional) now cause a compile-time error.
1972 This is actually not compatible with Perl, which accepts such patterns, but
1973 treats the conditional as always being FALSE (as PCRE used to), but it
1974 seems to me that giving a diagnostic is better.
1975
1976 8. Change "alphameric" to the more common word "alphanumeric" in comments
1977 and messages.
1978
1979 9. Fix two occurrences of "backslash" in comments that should have been
1980 "backspace".
1981
1982 10. Remove two redundant lines of code that can never be obeyed (their function
1983 was moved elsewhere).
1984
1985 11. The program that makes PCRE's Unicode character property table had a bug
1986 which caused it to generate incorrect table entries for sequences of
1987 characters that have the same character type, but are in different scripts.
1988 It amalgamated them into a single range, with the script of the first of
1989 them. In other words, some characters were in the wrong script. There were
1990 thirteen such cases, affecting characters in the following ranges:
1991
1992 U+002b0 - U+002c1
1993 U+0060c - U+0060d
1994 U+0061e - U+00612
1995 U+0064b - U+0065e
1996 U+0074d - U+0076d
1997 U+01800 - U+01805
1998 U+01d00 - U+01d77
1999 U+01d9b - U+01dbf
2000 U+0200b - U+0200f
2001 U+030fc - U+030fe
2002 U+03260 - U+0327f
2003 U+0fb46 - U+0fbb1
2004 U+10450 - U+1049d
2005
2006 12. The -o option (show only the matching part of a line) for pcregrep was not
2007 compatible with GNU grep in that, if there was more than one match in a
2008 line, it showed only the first of them. It now behaves in the same way as
2009 GNU grep.
2010
2011 13. If the -o and -v options were combined for pcregrep, it printed a blank
2012 line for every non-matching line. GNU grep prints nothing, and pcregrep now
2013 does the same. The return code can be used to tell if there were any
2014 non-matching lines.
2015
2016 14. Added --file-offsets and --line-offsets to pcregrep.
2017
2018 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
2019 infinitely looping recursion. The bug was that positive lookaheads were not
2020 being skipped when checking for a possible empty match (negative lookaheads
2021 and both kinds of lookbehind were skipped).
2022
2023 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
2024 inclusion of <windows.h> to before rather than after the definition of
2025 INVALID_FILE_ATTRIBUTES (patch from David Byron).
2026
2027 17. Specifying a possessive quantifier with a specific limit for a Unicode
2028 character property caused pcre_compile() to compile bad code, which led at
2029 runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
2030 are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
2031 caused the error; without that there was no problem.
2032
2033 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
2034
2035 19. Added --enable-pcretest-libreadline.
2036
2037 20. In pcrecpp.cc, the variable 'count' was incremented twice in
2038 RE::GlobalReplace(). As a result, the number of replacements returned was
2039 double what it should be. I removed one of the increments, but Craig sent a
2040 later patch that removed the other one (the right fix) and added unit tests
2041 that check the return values (which was not done before).
2042
2043 21. Several CMake things:
2044
2045 (1) Arranged that, when cmake is used on Unix, the libraries end up with
2046 the names libpcre and libpcreposix, not just pcre and pcreposix.
2047
2048 (2) The above change means that pcretest and pcregrep are now correctly
2049 linked with the newly-built libraries, not previously installed ones.
2050
2051 (3) Added PCRE_SUPPORT_LIBREADLINE, PCRE_SUPPORT_LIBZ, PCRE_SUPPORT_LIBBZ2.
2052
2053 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
2054 crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
2055 UTF-8 newline character). The key issue is that the pattern starts .*;
2056 this means that the match must be either at the beginning, or after a
2057 newline. The bug was in the code for advancing after a failed match and
2058 checking that the new position followed a newline. It was not taking
2059 account of UTF-8 characters correctly.
2060
2061 23. PCRE was behaving differently from Perl in the way it recognized POSIX
2062 character classes. PCRE was not treating the sequence [:...:] as a
2063 character class unless the ... were all letters. Perl, however, seems to
2064 allow any characters between [: and :], though of course it rejects as
2065 unknown any "names" that contain non-letters, because all the known class
2066 names consist only of letters. Thus, Perl gives an error for [[:1234:]],
2067 for example, whereas PCRE did not - it did not recognize a POSIX character
2068 class. This seemed a bit dangerous, so the code has been changed to be
2069 closer to Perl. The behaviour is not identical to Perl, because PCRE will
2070 diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
2071 treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
2072 Perl does, and where it didn't before.
2073
2074 24. Rewrite so as to remove the single use of %n from pcregrep because in some
2075 Windows environments %n is disabled by default.
2076
2077
2078 Version 7.4 21-Sep-07
2079 ---------------------
2080
2081 1. Change 7.3/28 was implemented for classes by looking at the bitmap. This
2082 means that a class such as [\s] counted as "explicit reference to CR or
2083 LF". That isn't really right - the whole point of the change was to try to
2084 help when there was an actual mention of one of the two characters. So now
2085 the change happens only if \r or \n (or a literal CR or LF) character is
2086 encountered.
2087
2088 2. The 32-bit options word was also used for 6 internal flags, but the numbers
2089 of both had grown to the point where there were only 3 bits left.
2090 Fortunately, there was spare space in the data structure, and so I have
2091 moved the internal flags into a new 16-bit field to free up more option
2092 bits.
2093
2094 3. The appearance of (?J) at the start of a pattern set the DUPNAMES option,
2095 but did not set the internal JCHANGED flag - either of these is enough to
2096 control the way the "get" function works - but the PCRE_INFO_JCHANGED
2097 facility is supposed to tell if (?J) was ever used, so now (?J) at the
2098 start sets both bits.
2099
2100 4. Added options (at build time, compile time, exec time) to change \R from
2101 matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
2102
2103 5. doc/pcresyntax.html was missing from the distribution.
2104
2105 6. Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
2106 compatibility, even though it is no longer used.
2107
2108 7. Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
2109 strtoull to pcrecpp.cc to select the available functions in WIN32 when the
2110 windows.h file is present (where different names are used). [This was
2111 reversed later after testing - see 16 below.]
2112
2113 8. Changed all #include <config.h> to #include "config.h". There were also
2114 some further <pcre.h> cases that I changed to "pcre.h".
2115
2116 9. When pcregrep was used with the --colour option, it missed the line ending
2117 sequence off the lines that it output.
2118
2119 10. It was pointed out to me that arrays of string pointers cause lots of
2120 relocations when a shared library is dynamically loaded. A technique of
2121 using a single long string with a table of offsets can drastically reduce
2122 these. I have refactored PCRE in four places to do this. The result is
2123 dramatic:
2124
2125 Originally: 290
2126 After changing UCP table: 187
2127 After changing error message table: 43
2128 After changing table of "verbs" 36
2129 After changing table of Posix names 22
2130
2131 Thanks to the folks working on Gregex for glib for this insight.
2132
2133 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
2134 unicode-properties was also set.
2135
2136 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
2137
2138 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
2139 checked only for CRLF.
2140
2141 14. Added casts to pcretest.c to avoid compiler warnings.
2142
2143 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
2144
2145 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
2146 and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
2147 entirely. This removes changes made in 7 above.
2148
2149 17. The CMake files have been updated, and there is now more information about
2150 building with CMake in the NON-UNIX-USE document.
2151
2152
2153 Version 7.3 28-Aug-07
2154 ---------------------
2155
2156 1. In the rejigging of the build system that eventually resulted in 7.1, the
2157 line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
2158 brackets there is not right, since it causes compilers to look for an
2159 installed pcre.h, not the version that is in the source that is being
2160 compiled (which of course may be different). I have changed it back to:
2161
2162 #include "pcre.h"
2163
2164 I have a vague recollection that the change was concerned with compiling in
2165 different directories, but in the new build system, that is taken care of
2166 by the VPATH setting the Makefile.
2167
2168 2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
2169 when the subject happened to end in the byte 0x85 (e.g. if the last
2170 character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
2171 characters but of course it shouldn't be taken as a newline when it is part
2172 of another character. The bug was that, for an unlimited repeat of . in
2173 not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
2174 characters when looking for a newline.
2175
2176 3. A small performance improvement in the DOTALL UTF-8 mode .* case.
2177
2178 4. Debugging: adjusted the names of opcodes for different kinds of parentheses
2179 in debug output.
2180
2181 5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
2182 long printing in the pcrecpp unittest when running under MinGW.
2183
2184 6. ESC_K was left out of the EBCDIC table.
2185
2186 7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
2187 parentheses; I made it 1000, which seemed large enough. Unfortunately, the
2188 limit also applies to "virtual nesting" when a pattern is recursive, and in
2189 this case 1000 isn't so big. I have been able to remove this limit at the
2190 expense of backing off one optimization in certain circumstances. Normally,
2191 when pcre_exec() would call its internal match() function recursively and
2192 immediately return the result unconditionally, it uses a "tail recursion"
2193 feature to save stack. However, when a subpattern that can match an empty
2194 string has an unlimited repetition quantifier, it no longer makes this
2195 optimization. That gives it a stack frame in which to save the data for
2196 checking that an empty string has been matched. Previously this was taken
2197 from the 1000-entry workspace that had been reserved. So now there is no
2198 explicit limit, but more stack is used.
2199
2200 8. Applied Daniel's patches to solve problems with the import/export magic
2201 syntax that is required for Windows, and which was going wrong for the
2202 pcreposix and pcrecpp parts of the library. These were overlooked when this
2203 problem was solved for the main library.
2204
2205 9. There were some crude static tests to avoid integer overflow when computing
2206 the size of patterns that contain repeated groups with explicit upper
2207 limits. As the maximum quantifier is 65535, the maximum group length was
2208 set at 30,000 so that the product of these two numbers did not overflow a
2209 32-bit integer. However, it turns out that people want to use groups that
2210 are longer than 30,000 bytes (though not repeat them that many times).
2211 Change 7.0/17 (the refactoring of the way the pattern size is computed) has
2212 made it possible to implement the integer overflow checks in a much more
2213 dynamic way, which I have now done. The artificial limitation on group
2214 length has been removed - we now have only the limit on the total length of
2215 the compiled pattern, which depends on the LINK_SIZE setting.
2216
2217 10. Fixed a bug in the documentation for get/copy named substring when
2218 duplicate names are permitted. If none of the named substrings are set, the
2219 functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
2220 empty string.
2221
2222 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
2223 instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
2224 because the ] is interpreted as the first data character and the
2225 terminating ] is not found. PCRE has been made compatible with Perl in this
2226 regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
2227 cause memory overwriting.
2228
2229 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
2230 string has been matched (to stop an infinite loop). It was not recognizing
2231 a conditional subpattern that could match an empty string if that
2232 subpattern was within another subpattern. For example, it looped when
2233 trying to match (((?(1)X|))*) but it was OK with ((?(1)X|)*) where the
2234 condition was not nested. This bug has been fixed.
2235
2236 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
2237 past the start of the subject in the presence of bytes with the top bit
2238 set, for example "\x8aBCD".
2239
2240 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
2241 (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
2242
2243 14. Optimized (?!) to (*FAIL).
2244
2245 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
2246 This restricts code points to be within the range 0 to 0x10FFFF, excluding
2247 the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
2248 full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
2249 does: it's just the validity check that is more restrictive.
2250
2251 16. Inserted checks for integer overflows during escape sequence (backslash)
2252 processing, and also fixed erroneous offset values for syntax errors during
2253 backslash processing.
2254
2255 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
2256 for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
2257
2258 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
2259 caused an overrun.
2260
2261 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
2262 something other than just ASCII characters) inside a group that had an
2263 unlimited repeat caused a loop at compile time (while checking to see
2264 whether the group could match an empty string).
2265
2266 20. Debugging a pattern containing \p or \P could cause a crash. For example,
2267 [\P{Any}] did so. (Error in the code for printing property names.)
2268
2269 21. An orphan \E inside a character class could cause a crash.
2270
2271 22. A repeated capturing bracket such as (A)? could cause a wild memory
2272 reference during compilation.
2273
2274 23. There are several functions in pcre_compile() that scan along a compiled
2275 expression for various reasons (e.g. to see if it's fixed length for look
2276 behind). There were bugs in these functions when a repeated \p or \P was
2277 present in the pattern. These operators have additional parameters compared
2278 with \d, etc, and these were not being taken into account when moving along
2279 the compiled data. Specifically:
2280
2281 (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
2282 length.
2283
2284 (b) An item such as \pL+ within a repeated group could cause crashes or
2285 loops.
2286
2287 (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
2288 "reference to non-existent subpattern" error.
2289
2290 (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
2291
2292 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
2293 characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
2294
2295 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
2296
2297 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
2298 character were causing crashes (broken optimization).
2299
2300 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
2301 \p or \P) caused a compile-time loop.
2302
2303 28. More problems have arisen in unanchored patterns when CRLF is a valid line
2304 break. For example, the unstudied pattern [\r\n]A does not match the string
2305 "\r\nA" because change 7.0/46 below moves the current point on by two
2306 characters after failing to match at the start. However, the pattern \nA
2307 *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
2308 the same is true. There doesn't seem any very clean way out of this, but
2309 what I have chosen to do makes the common cases work: PCRE now takes note
2310 of whether there can be an explicit match for \r or \n anywhere in the
2311 pattern, and if so, 7.0/46 no longer applies. As part of this change,
2312 there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
2313 pattern has explicit CR or LF references.
2314
2315 29. Added (*CR) etc for changing newline setting at start of pattern.
2316
2317
2318 Version 7.2 19-Jun-07
2319 ---------------------
2320
2321 1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
2322 which is apparently normally available under Windows.
2323
2324 2. Re-jig the pcregrep tests with different newline settings in an attempt
2325 to make them independent of the local environment's newline setting.
2326
2327 3. Add code to configure.ac to remove -g from the CFLAGS default settings.
2328
2329 4. Some of the "internals" tests were previously cut out when the link size
2330 was not 2, because the output contained actual offsets. The recent new
2331 "Z" feature of pcretest means that these can be cut out, making the tests
2332 usable with all link sizes.
2333
2334 5. Implemented Stan Switzer's goto replacement for longjmp() when not using
2335 stack recursion. This gives a massive performance boost under BSD, but just
2336 a small improvement under Linux. However, it saves one field in the frame
2337 in all cases.
2338
2339 6. Added more features from the forthcoming Perl 5.10:
2340
2341 (a) (?-n) (where n is a string of digits) is a relative subroutine or
2342 recursion call. It refers to the nth most recently opened parentheses.
2343
2344 (b) (?+n) is also a relative subroutine call; it refers to the nth next
2345 to be opened parentheses.
2346
2347 (c) Conditions that refer to capturing parentheses can be specified
2348 relatively, for example, (?(-2)... or (?(+3)...
2349
2350 (d) \K resets the start of the current match so that everything before
2351 is not part of it.
2352
2353 (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
2354
2355 (f) \g{name} is another synonym - part of Perl 5.10's unification of
2356 reference syntax.
2357
2358 (g) (?| introduces a group in which the numbering of parentheses in each
2359 alternative starts with the same number.
2360
2361 (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
2362
2363 7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
2364 PCRE_INFO_JCHANGED.
2365
2366 8. A pattern such as (.*(.)?)* caused pcre_exec() to fail by either not
2367 terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
2368 for detecting groups that can match an empty string.
2369
2370 9. A pattern with a very large number of alternatives (more than several
2371 hundred) was running out of internal workspace during the pre-compile
2372 phase, where pcre_compile() figures out how much memory will be needed. A
2373 bit of new cunning has reduced the workspace needed for groups with
2374 alternatives. The 1000-alternative test pattern now uses 12 bytes of
2375 workspace instead of running out of the 4096 that are available.
2376
2377 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
2378
2379 11. Applied patch from Google to remove an optimization that didn't quite work.
2380 The report of the bug said:
2381
2382 pcrecpp::RE("a*").FullMatch("aaa") matches, while
2383 pcrecpp::RE("a*?").FullMatch("aaa") does not, and
2384 pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
2385
2386 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
2387 it matched the wrong number of bytes.
2388
2389
2390 Version 7.1 24-Apr-07
2391 ---------------------
2392
2393 1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
2394 that is more "standard", making use of automake and other Autotools. There
2395 is some re-arrangement of the files and adjustment of comments consequent
2396 on this.
2397
2398 2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
2399 for recursive directory scanning broke on some systems because the files
2400 are not scanned in any specific order and on different systems the order
2401 was different. A call to "sort" has been inserted into RunGrepTest for the
2402 approprate test as a short-term fix. In the longer term there may be an
2403 alternative.
2404
2405 3. I had an email from Eric Raymond about problems translating some of PCRE's
2406 man pages to HTML (despite the fact that I distribute HTML pages, some
2407 people do their own conversions for various reasons). The problems
2408 concerned the use of low-level troff macros .br and .in. I have therefore
2409 removed all such uses from the man pages (some were redundant, some could
2410 be replaced by .nf/.fi pairs). The 132html script that I use to generate
2411 HTML has been updated to handle .nf/.fi and to complain if it encounters
2412 .br or .in.
2413
2414 4. Updated comments in configure.ac that get placed in config.h.in and also
2415 arranged for config.h to be included in the distribution, with the name
2416 config.h.generic, for the benefit of those who have to compile without
2417 Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
2418
2419 5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
2420 Weber: (1) pcre_internal.h was missing some function renames; (2) updated
2421 makevp.bat for the current PCRE, using the additional files
2422 makevp_c.txt, makevp_l.txt, and pcregexp.pas.
2423
2424 6. A Windows user reported a minor discrepancy with test 2, which turned out
2425 to be caused by a trailing space on an input line that had got lost in his
2426 copy. The trailing space was an accident, so I've just removed it.
2427
2428 7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
2429 that is needed.
2430
2431 8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
2432 as "const" (a) because they are and (b) because it helps the PHP
2433 maintainers who have recently made a script to detect big data structures
2434 in the php code that should be moved to the .rodata section. I remembered
2435 to update Builducptable as well, so it won't revert if ucptable.h is ever
2436 re-created.
2437
2438 9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
2439 pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
2440 order to be able to cut out the UTF-8 tables in the latter when UTF-8
2441 support is not required. This saves 1.5-2K of code, which is important in
2442 some applications.
2443
2444 Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
2445 so as not to refer to the tables, even though these functions will never be
2446 called when UTF-8 support is disabled. Otherwise there are problems with a
2447 shared library.
2448
2449 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
2450
2451 (a) It was defining its arguments as char * instead of void *.
2452
2453 (b) It was assuming that all moves were upwards in memory; this was true
2454 a long time ago when I wrote it, but is no longer the case.
2455
2456 The emulated memove() is provided for those environments that have neither
2457 memmove() nor bcopy(). I didn't think anyone used it these days, but that
2458 is clearly not the case, as these two bugs were recently reported.
2459
2460 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
2461 and Detrail to create the HTML documentation, the .txt form of the man
2462 pages, and it removes trailing spaces from listed files. It also creates
2463 pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
2464 case, it wraps all the #defines with #ifndefs. This script should be run
2465 before "make dist".
2466
2467 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
2468 with Unicode property support.
2469
2470 (a) For a maximizing quantifier, if the two different cases of the
2471 character were of different lengths in their UTF-8 codings (there are
2472 some cases like this - I found 11), and the matching function had to
2473 back up over a mixture of the two cases, it incorrectly assumed they
2474 were both the same length.
2475
2476 (b) When PCRE was configured to use the heap rather than the stack for
2477 recursion during matching, it was not correctly preserving the data for
2478 the other case of a UTF-8 character when checking ahead for a match
2479 while processing a minimizing repeat. If the check also involved
2480 matching a wide character, but failed, corruption could cause an
2481 erroneous result when trying to check for a repeat of the original
2482 character.
2483
2484 13. Some tidying changes to the testing mechanism:
2485
2486 (a) The RunTest script now detects the internal link size and whether there
2487 is UTF-8 and UCP support by running ./pcretest -C instead of relying on
2488 values substituted by "configure". (The RunGrepTest script already did
2489 this for UTF-8.) The configure.ac script no longer substitutes the
2490 relevant variables.
2491
2492 (b) The debugging options /B and /D in pcretest show the compiled bytecode
2493 with length and offset values. This means that the output is different
2494 for different internal link sizes. Test 2 is skipped for link sizes
2495 other than 2 because of this, bypassing the problem. Unfortunately,
2496 there was also a test in test 3 (the locale tests) that used /B and
2497 failed for link sizes other than 2. Rather than cut the whole test out,
2498 I have added a new /Z option to pcretest that replaces the length and
2499 offset values with spaces. This is now used to make test 3 independent
2500 of link size. (Test 2 will be tidied up later.)
2501
2502 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
2503 segmentation fault instead of returning the appropriate error message.
2504
2505 15. In multiline mode when the newline sequence was set to "any", the pattern
2506 ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
2507 This doesn't seem right; it now treats the CRLF combination as the line
2508 ending, and so does not match in that case. It's only a pattern such as ^$
2509 that would hit this one: something like ^ABC$ would have failed after \r
2510 and then tried again after \r\n.
2511
2512 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
2513 in an attempt to make files that differ only in their line terminators
2514 compare equal. This works on Linux.
2515
2516 17. Under certain error circumstances pcregrep might try to free random memory
2517 as it exited. This is now fixed, thanks to valgrind.
2518
2519 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
2520 "abc\r\n\r\n", it found an unwanted second match after the second \r. This
2521 was because its rules for how to advance for /g after matching an empty
2522 string at the end of a line did not allow for this case. They now check for
2523 it specially.
2524
2525 20. pcretest is supposed to handle patterns and data of any length, by
2526 extending its buffers when necessary. It was getting this wrong when the
2527 buffer for a data line had to be extended.
2528
2529 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
2530 CRLF as a newline sequence.
2531
2532 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
2533 out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
2534 I have nevertheless tidied it up.
2535
2536 23. Added some casts to kill warnings from HP-UX ia64 compiler.
2537
2538 24. Added a man page for pcre-config.
2539
2540
2541 Version 7.0 19-Dec-06
2542 ---------------------
2543
2544 1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
2545 moving to gcc 4.1.1.
2546
2547 2. The -S option for pcretest uses setrlimit(); I had omitted to #include
2548 sys/time.h, which is documented as needed for this function. It doesn't
2549 seem to matter on Linux, but it showed up on some releases of OS X.
2550
2551 3. It seems that there are systems where bytes whose values are greater than
2552 127 match isprint() in the "C" locale. The "C" locale should be the
2553 default when a C program starts up. In most systems, only ASCII printing
2554 characters match isprint(). This difference caused the output from pcretest
2555 to vary, making some of the tests fail. I have changed pcretest so that:
2556
2557 (a) When it is outputting text in the compiled version of a pattern, bytes
2558 other than 32-126 are always shown as hex escapes.
2559
2560 (b) When it is outputting text that is a matched part of a subject string,
2561 it does the same, unless a different locale has been set for the match
2562 (using the /L modifier). In this case, it uses isprint() to decide.
2563
2564 4. Fixed a major bug that caused incorrect computation of the amount of memory
2565 required for a compiled pattern when options that changed within the
2566 pattern affected the logic of the preliminary scan that determines the
2567 length. The relevant options are -x, and -i in UTF-8 mode. The result was
2568 that the computed length was too small. The symptoms of this bug were
2569 either the PCRE error "internal error: code overflow" from pcre_compile(),
2570 or a glibc crash with a message such as "pcretest: free(): invalid next
2571 size (fast)". Examples of patterns that provoked this bug (shown in
2572 pcretest format) are:
2573
2574 /(?-x: )/x
2575 /(?x)(?-x: \s*#\s*)/
2576 /((?i)[\x{c0}])/8
2577 /(?i:[\x{c0}])/8
2578
2579 HOWEVER: Change 17 below makes this fix obsolete as the memory computation
2580 is now done differently.
2581
2582 5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
2583 wrapper classes; (b) implement a new function in the C++ scanner that is
2584 more efficient than the old way of doing things because it avoids levels of
2585 recursion in the regex matching; (c) add a paragraph to the documentation
2586 for the FullMatch() function.
2587
2588 6. The escape sequence \n was being treated as whatever was defined as
2589 "newline". Not only was this contrary to the documentation, which states
2590 that \n is character 10 (hex 0A), but it also went horribly wrong when
2591 "newline" was defined as CRLF. This has been fixed.
2592
2593 7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
2594 was being set to -1 for the "end of line" case (supposedly a value that no
2595 character can have). Though this value is never used (the check for end of
2596 line is "zero bytes in current character"), it caused compiler complaints.
2597 I've changed it to 0xffffffff.
2598
2599 8. In pcre_version.c, the version string was being built by a sequence of
2600 C macros that, in the event of PCRE_PRERELEASE being defined as an empty
2601 string (as it is for production releases) called a macro with an empty
2602 argument. The C standard says the result of this is undefined. The gcc
2603 compiler treats it as an empty string (which was what was wanted) but it is
2604 reported that Visual C gives an error. The source has been hacked around to
2605 avoid this problem.
2606
2607 9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
2608 builds of pcretest, and changed the call to _setmode() to use _O_BINARY
2609 instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
2610 of them did).
2611
2612 10. Originally, pcretest opened its input and output without "b"; then I was
2613 told that "b" was needed in some environments, so it was added for release
2614 5.0 to both the input and output. (It makes no difference on Unix-like
2615 systems.) Later I was told that it is wrong for the input on Windows. I've
2616 now abstracted the modes into two macros, to make it easier to fiddle with
2617 them, and removed "b" from the input mode under Windows.
2618
2619 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
2620
2621 12. Added -help and --help to pcretest as an official way of being reminded
2622 of the options.
2623
2624 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
2625 and pcrecpp.cc because they annoy compilers at high warning levels.
2626
2627 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
2628
2629 15. Fixed an occurrence of == in configure.ac that should have been = (shell
2630 scripts are not C programs :-) and which was not noticed because it works
2631 on Linux.
2632
2633 16. pcretest is supposed to handle any length of pattern and data line (as one
2634 line or as a continued sequence of lines) by extending its input buffer if
2635 necessary. This feature was broken for very long pattern lines, leading to
2636 a string of junk being passed to pcre_compile() if the pattern was longer
2637 than about 50K.
2638
2639 17. I have done a major re-factoring of the way pcre_compile() computes the
2640 amount of memory needed for a compiled pattern. Previously, there was code
2641 that made a preliminary scan of the pattern in order to do this. That was
2642 OK when PCRE was new, but as the facilities have expanded, it has become
2643 harder and harder to keep it in step with the real compile phase, and there
2644 have been a number of bugs (see for example, 4 above). I have now found a
2645 cunning way of running the real compile function in a "fake" mode that
2646 enables it to compute how much memory it would need, while actually only
2647 ever using a few hundred bytes of working memory and without too many
2648 tests of the mode. This should make future maintenance and development
2649 easier. A side effect of this work is that the limit of 200 on the nesting
2650 depth of parentheses has been removed (though this was never a serious
2651 limitation, I suspect). However, there is a downside: pcre_compile() now
2652 runs more slowly than before (30% or more, depending on the pattern). I
2653 hope this isn't a big issue. There is no effect on runtime performance.
2654
2655 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
2656 newline (only possible for the last line of a file) and it was a
2657 pattern that set a locale (followed by /Lsomething), pcretest crashed.
2658
2659 19. Added additional timing features to pcretest. (1) The -tm option now times
2660 matching only, not compiling. (2) Both -t and -tm can be followed, as a
2661 separate command line item, by a number that specifies the number of
2662 repeats to use when timing. The default is 50000; this gives better
2663 precision, but takes uncomfortably long for very large patterns.
2664
2665 20. Extended pcre_study() to be more clever in cases where a branch of a
2666 subpattern has no definite first character. For example, (a*|b*)[cd] would
2667 previously give no result from pcre_study(). Now it recognizes that the
2668 first character must be a, b, c, or d.
2669
2670 21. There was an incorrect error "recursive call could loop indefinitely" if
2671 a subpattern (or the entire pattern) that was being tested for matching an
2672 empty string contained only one non-empty item after a nested subpattern.
2673 For example, the pattern (?>\x{100}*)\d(?R) provoked this error
2674 incorrectly, because the \d was being skipped in the check.
2675
2676 22. The pcretest program now has a new pattern option /B and a command line
2677 option -b, which is equivalent to adding /B to every pattern. This causes
2678 it to show the compiled bytecode, without the additional information that
2679 -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
2680 is the same as /B/I).
2681
2682 23. A new optimization is now able automatically to treat some sequences such
2683 as a*b as a*+b. More specifically, if something simple (such as a character
2684 or a simple class like \d) has an unlimited quantifier, and is followed by
2685 something that cannot possibly match the quantified thing, the quantifier
2686 is automatically "possessified".
2687
2688 24. A recursive reference to a subpattern whose number was greater than 39
2689 went wrong under certain circumstances in UTF-8 mode. This bug could also
2690 have affected the operation of pcre_study().
2691
2692 25. Realized that a little bit of performance could be had by replacing
2693 (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
2694
2695 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
2696
2697 27. Possessive quantifiers such as a++ were previously implemented by turning
2698 them into atomic groups such as ($>a+). Now they have their own opcodes,
2699 which improves performance. This includes the automatically created ones
2700 from 23 above.
2701
2702 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
2703 lookahead was broken if it was not anchored. PCRE was mistakenly expecting
2704 the first matched character to be a colon. This applied both to named and
2705 numbered groups.
2706
2707 29. The ucpinternal.h header file was missing its idempotency #ifdef.
2708
2709 30. I was sent a "project" file called libpcre.a.dev which I understand makes
2710 building PCRE on Windows easier, so I have included it in the distribution.
2711
2712 31. There is now a check in pcretest against a ridiculously large number being
2713 returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
2714 loop, the loop is abandoned.
2715
2716 32. Forward references to subpatterns in conditions such as (?(2)...) where
2717 subpattern 2 is defined later cause pcre_compile() to search forwards in
2718 the pattern for the relevant set of parentheses. This search went wrong
2719 when there were unescaped parentheses in a character class, parentheses
2720 escaped with \Q...\E, or parentheses in a #-comment in /x mode.
2721
2722 33. "Subroutine" calls and backreferences were previously restricted to
2723 referencing subpatterns earlier in the regex. This restriction has now
2724 been removed.
2725
2726 34. Added a number of extra features that are going to be in Perl 5.10. On the
2727 whole, these are just syntactic alternatives for features that PCRE had
2728 previously implemented using the Python syntax or my own invention. The
2729 other formats are all retained for compatibility.
2730
2731 (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
2732 as (?P<name>...). The new forms, as well as being in Perl 5.10, are
2733 also .NET compatible.
2734
2735 (b) A recursion or subroutine call to a named group can now be defined as
2736 (?&name) as well as (?P>name).
2737
2738 (c) A backreference to a named group can now be defined as \k<name> or
2739 \k'name' as well as (?P=name). The new forms, as well as being in Perl
2740 5.10, are also .NET compatible.
2741
2742 (d) A conditional reference to a named group can now use the syntax
2743 (?(<name>) or (?('name') as well as (?(name).
2744
2745 (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
2746 groups (named and numbered) that are never evaluated inline, but can be
2747 called as "subroutines" from elsewhere. In effect, the DEFINE condition
2748 is always false. There may be only one alternative in such a group.
2749
2750 (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
2751 as the simple (?(R). The condition is true only if the most recent
2752 recursion is that of the given number or name. It does not search out
2753 through the entire recursion stack.
2754
2755 (g) The escape \gN or \g{N} has been added, where N is a positive or
2756 negative number, specifying an absolute or relative reference.
2757
2758 35. Tidied to get rid of some further signed/unsigned compiler warnings and
2759 some "unreachable code" warnings.
2760
2761 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
2762 things, this adds five new scripts.
2763
2764 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
2765 There were also incompatibilities regarding the handling of \Q..\E inside
2766 character classes, for example with patterns like [\Qa\E-\Qz\E] where the
2767 hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
2768
2769 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
2770 matches an empty string, and forcibly breaks the loop. There were bugs in
2771 this code in non-simple cases. For a pattern such as ^(a()*)* matched
2772 against aaaa the result was just "a" rather than "aaaa", for example. Two
2773 separate and independent bugs (that affected different cases) have been
2774 fixed.
2775
2776 39. Refactored the code to abolish the use of different opcodes for small
2777 capturing bracket numbers. This is a tidy that I avoided doing when I
2778 removed the limit on the number of capturing brackets for 3.5 back in 2001.
2779 The new approach is not only tidier, it makes it possible to reduce the
2780 memory needed to fix the previous bug (38).
2781
2782 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
2783 sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
2784 processing dot, circumflex, or dollar metacharacters, or #-comments in /x
2785 mode.
2786
2787 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
2788 report.
2789
2790 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
2791 copy construction and assignment in the C++ wrapper.
2792
2793 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
2794 couple of bugs that could have given wrong results in the "--newline=crlf"
2795 case.
2796
2797 44. Added a number of casts and did some reorganization of signed/unsigned int
2798 variables following suggestions from Dair Grant. Also renamed the variable
2799 "this" as "item" because it is a C++ keyword.
2800
2801 45. Arranged for dftables to add
2802
2803 #include "pcre_internal.h"
2804
2805 to pcre_chartables.c because without it, gcc 4.x may remove the array
2806 definition from the final binary if PCRE is built into a static library and
2807 dead code stripping is activated.
2808
2809 46. For an unanchored pattern, if a match attempt fails at the start of a
2810 newline sequence, and the newline setting is CRLF or ANY, and the next two
2811 characters are CRLF, advance by two characters instead of one.
2812
2813
2814 Version 6.7 04-Jul-06
2815 ---------------------
2816
2817 1. In order to handle tests when input lines are enormously long, pcretest has
2818 been re-factored so that it automatically extends its buffers when
2819 necessary. The code is crude, but this _is_ just a test program. The
2820 default size has been increased from 32K to 50K.
2821
2822 2. The code in pcre_study() was using the value of the re argument before
2823 testing it for NULL. (Of course, in any sensible call of the function, it
2824 won't be NULL.)
2825
2826 3. The memmove() emulation function in pcre_internal.h, which is used on
2827 systems that lack both memmove() and bcopy() - that is, hardly ever -
2828 was missing a "static" storage class specifier.
2829
2830 4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
2831 containing an extended class (one that cannot be represented by a bitmap
2832 because it contains high-valued characters or Unicode property items, e.g.
2833 [\pZ]). Almost always one would set UTF-8 mode when processing such a
2834 pattern, but PCRE should not loop if you do not (it no longer does).
2835 [Detail: two cases were found: (a) a repeated subpattern containing an
2836 extended class; (b) a recursive reference to a subpattern that followed a
2837 previous extended class. It wasn't skipping over the extended class
2838 correctly when UTF-8 mode was not set.]
2839
2840 5. A negated single-character class was not being recognized as fixed-length
2841 in lookbehind assertions such as (?<=[^f]), leading to an incorrect
2842 compile error "lookbehind assertion is not fixed length".
2843
2844 6. The RunPerlTest auxiliary script was showing an unexpected difference
2845 between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
2846 write a Perl script that can interpret lines of an input file either as
2847 byte characters or as UTF-8, which is what "perltest" was being required to
2848 do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
2849 can't do is switch easily at run time between having the "use utf8;" pragma
2850 or not. In the end, I fudged it by using the RunPerlTest script to insert
2851 "use utf8;" explicitly for the UTF-8 tests.
2852
2853 7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
2854 the end of the subject string, contrary to the documentation and to what
2855 Perl does. This was true of both matching functions. Now it matches only at
2856 the start of the subject and immediately after *internal* newlines.
2857
2858 8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
2859 a pointer to an int instead of a pointer to an unsigned long int. This
2860 caused problems on 64-bit systems.
2861
2862 9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
2863 instance of the 'standard' template library not being so standard".
2864
2865 10. There was no check on the number of named subpatterns nor the maximum
2866 length of a subpattern name. The product of these values is used to compute
2867 the size of the memory block for a compiled pattern. By supplying a very
2868 long subpattern name and a large number of named subpatterns, the size
2869 computation could be caused to overflow. This is now prevented by limiting
2870 the length of names to 32 characters, and the number of named subpatterns
2871 to 10,000.
2872
2873 11. Subpatterns that are repeated with specific counts have to be replicated in
2874 the compiled pattern. The size of memory for this was computed from the
2875 length of the subpattern and the repeat count. The latter is limited to
2876 65535, but there was no limit on the former, meaning that integer overflow
2877 could in principle occur. The compiled length of a repeated subpattern is
2878 now limited to 30,000 bytes in order to prevent this.
2879
2880 12. Added the optional facility to have named substrings with the same name.
2881
2882 13. Added the ability to use a named substring as a condition, using the
2883 Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
2884 are numbers (not recommended). Forward references are permitted.
2885
2886 14. Added forward references in named backreferences (if you see what I mean).
2887
2888 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
2889 pattern could run off the end of the subject. For example, the pattern
2890 "(?s)(.{1,5})"8 did this with the subject "ab".
2891
2892 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
2893 PCRE_CASELESS was set when matching characters that were quantified with ?
2894 or *.
2895
2896 17. A character class other than a single negated character that had a minimum
2897 but no maximum quantifier - for example [ab]{6,} - was not handled
2898 correctly by pce_dfa_exec(). It would match only one character.
2899
2900 18. A valid (though odd) pattern that looked like a POSIX character
2901 class but used an invalid character after [ (for example [[,abc,]]) caused
2902 pcre_compile() to give the error "Failed: internal error: code overflow" or
2903 in some cases to crash with a glibc free() error. This could even happen if
2904 the pattern terminated after [[ but there just happened to be a sequence of
2905 letters, a binary zero, and a closing ] in the memory that followed.
2906
2907 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
2908 over the years. Originally (before any Unicode support), just the bottom 8
2909 bits were taken. Thus, for example, \500 really meant \100. Nowadays the
2910 output from "man perlunicode" includes this:
2911
2912 The regular expression compiler produces polymorphic opcodes. That
2913 is, the pattern adapts to the data and automatically switches to
2914 the Unicode character scheme when presented with Unicode data--or
2915 instead uses a traditional byte scheme when presented with byte
2916 data.
2917
2918 Sadly, a wide octal escape does not cause a switch, and in a string with
2919 no other multibyte characters, these octal escapes are treated as before.
2920 Thus, in Perl, the pattern /\500/ actually matches \100 but the pattern
2921 /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
2922 Unicode string.
2923
2924 I have not perpetrated such confusion in PCRE. Up till now, it took just
2925 the bottom 8 bits, as in old Perl. I have now made octal escapes with
2926 values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
2927 translate to the appropriate multibyte character.
2928
2929 29. Applied some refactoring to reduce the number of warnings from Microsoft
2930 and Borland compilers. This has included removing the fudge introduced
2931 seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
2932 a warning about an unused variable.
2933
2934 21. PCRE has not included VT (character 0x0b) in the set of whitespace
2935 characters since release 4.0, because Perl (from release 5.004) does not.
2936 [Or at least, is documented not to: some releases seem to be in conflict
2937 with the documentation.] However, when a pattern was studied with
2938 pcre_study() and all its branches started with \s, PCRE still included VT
2939 as a possible starting character. Of course, this did no harm; it just
2940 caused an unnecessary match attempt.
2941
2942 22. Removed a now-redundant internal flag bit that recorded the fact that case
2943 dependency changed within the pattern. This was once needed for "required
2944 byte" processing, but is no longer used. This recovers a now-scarce options
2945 bit. Also moved the least significant internal flag bit to the most-
2946 significant bit of the word, which was not previously used (hangover from
2947 the days when it was an int rather than a uint) to free up another bit for
2948 the future.
2949
2950 23. Added support for CRLF line endings as well as CR and LF. As well as the
2951 default being selectable at build time, it can now be changed at runtime
2952 via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
2953 specify that it is scanning data with non-default line endings.
2954
2955 24. Changed the definition of CXXLINK to make it agree with the definition of
2956 LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
2957
2958 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
2959 recursions. This makes a big different to stack usage for some patterns.
2960
2961 26. If a subpattern containing a named recursion or subroutine reference such
2962 as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
2963 the space required for the compiled pattern went wrong and gave too small a
2964 value. Depending on the environment, this could lead to "Failed: internal
2965 error: code overflow at offset 49" or "glibc detected double free or
2966 corruption" errors.
2967
2968 27. Applied patches from Google (a) to support the new newline modes and (b) to
2969 advance over multibyte UTF-8 characters in GlobalReplace.
2970
2971 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
2972 difference for some implementation of PCRE in some Windows version.
2973
2974 29. Added some extra testing facilities to pcretest:
2975
2976 \q<number> in a data line sets the "match limit" value
2977 \Q<number> in a data line sets the "match recursion limt" value
2978 -S <number> sets the stack size, where <number> is in megabytes
2979
2980 The -S option isn't available for Windows.
2981
2982
2983 Version 6.6 06-Feb-06
2984 ---------------------
2985
2986 1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
2987 in pcreposix.h. I have copied the definition from pcre.h.
2988
2989 2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
2990 because pcre.h is no longer a built file.
2991
2992 3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
2993 not normally included in the compiled code.
2994
2995
2996 Version 6.5 01-Feb-06
2997 ---------------------
2998
2999 1. When using the partial match feature with pcre_dfa_exec(), it was not
3000 anchoring the second and subsequent partial matches at the new starting
3001 point. This could lead to incorrect results. For example, with the pattern
3002 /1234/, partially matching against "123" and then "a4" gave a match.
3003
3004 2. Changes to pcregrep:
3005
3006 (a) All non-match returns from pcre_exec() were being treated as failures
3007 to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
3008 error message is output. Some extra information is given for the
3009 PCRE_ERROR_MATCHLIMIT and PCRE_ERROR_RECURSIONLIMIT errors, which are
3010 probably the only errors that are likely to be caused by users (by
3011 specifying a regex that has nested indefinite repeats, for instance).
3012 If there are more than 20 of these errors, pcregrep is abandoned.
3013
3014 (b) A binary zero was treated as data while matching, but terminated the
3015 output line if it was written out. This has been fixed: binary zeroes
3016 are now no different to any other data bytes.
3017
3018 (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
3019 used to set a locale for matching. The --locale=xxxx long option has
3020 been added (no short equivalent) to specify a locale explicitly on the
3021 pcregrep command, overriding the environment variables.
3022
3023 (d) When -B was used with -n, some line numbers in the output were one less
3024 than they should have been.
3025
3026 (e) Added the -o (--only-matching) option.
3027
3028 (f) If -A or -C was used with -c (count only), some lines of context were
3029 accidentally printed for the final match.
3030
3031 (g) Added the -H (--with-filename) option.
3032
3033 (h) The combination of options -rh failed to suppress file names for files
3034 that were found from directory arguments.
3035
3036 (i) Added the -D (--devices) and -d (--directories) options.
3037
3038 (j) Added the -F (--fixed-strings) option.
3039
3040 (k) Allow "-" to be used as a file name for -f as well as for a data file.
3041
3042 (l) Added the --colo(u)r option.
3043
3044 (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
3045 is not present by default.
3046
3047 3. A nasty bug was discovered in the handling of recursive patterns, that is,
3048 items such as (?R) or (?1), when the recursion could match a number of
3049 alternatives. If it matched one of the alternatives, but subsequently,
3050 outside the recursion, there was a failure, the code tried to back up into
3051 the recursion. However, because of the way PCRE is implemented, this is not
3052 possible, and the result was an incorrect result from the match.
3053
3054 In order to prevent this happening, the specification of recursion has
3055 been changed so that all such subpatterns are automatically treated as
3056 atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
3057
3058 4. I had overlooked the fact that, in some locales, there are characters for
3059 which isalpha() is true but neither isupper() nor islower() are true. In
3060 the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
3061 and ordfeminine) are like this. This affected the treatment of \w and \W
3062 when they appeared in character classes, but not when they appeared outside
3063 a character class. The bit map for "word" characters is now created
3064 separately from the results of isalnum() instead of just taking it from the
3065 upper, lower, and digit maps. (Plus the underscore character, of course.)
3066
3067 5. The above bug also affected the handling of POSIX character classes such as
3068 [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
3069 permanent tables. Instead, the bit maps for such a class were previously
3070 created as the appropriate unions of the upper, lower, and digit bitmaps.
3071 Now they are created by subtraction from the [[:word:]] class, which has
3072 its own bitmap.
3073
3074 6. The [[:blank:]] character class matches horizontal, but not vertical space.
3075 It is created by subtracting the vertical space characters (\x09, \x0a,
3076 \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
3077 subtraction was done in the overall bitmap for a character class, meaning
3078 that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
3079 be recognized. This bug has been fixed.
3080
3081 7. Patches from the folks at Google:
3082
3083 (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
3084 real life, but is still worth protecting against".
3085
3086 (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
3087 regular expressions".
3088
3089 (c) pcre_scanner.cc: avoid use of std::count() because not all systems
3090 have it.
3091
3092 (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
3093 "configure" and the latter not, in order to fix a problem somebody had
3094 with compiling the Arg class on HP-UX.
3095
3096 (e) Improve the error-handling of the C++ wrapper a little bit.
3097
3098 (f) New tests for checking recursion limiting.
3099
3100 8. The pcre_memmove() function, which is used only if the environment does not
3101 have a standard memmove() function (and is therefore rarely compiled),
3102 contained two bugs: (a) use of int instead of size_t, and (b) it was not
3103 returning a result (though PCRE never actually uses the result).
3104
3105 9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
3106 large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
3107 returned instead of calling malloc() with an overflowing number that would
3108 most likely cause subsequent chaos.
3109
3110 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
3111
3112 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
3113 with this option is matched, the nmatch and pmatch options of regexec() are
3114 ignored.
3115
3116 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
3117 provided in case anyone wants to the the POSIX interface with UTF-8
3118 strings.
3119
3120 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
3121 C++ linking (needed for some HP-UX environments).
3122
3123 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
3124 (unused parameter) and in the pcre_printint() function (omitted "default"
3125 switch label when the default is to do nothing).
3126
3127 15. Added some code to make it possible, when PCRE is compiled as a C++
3128 library, to replace subject pointers for pcre_exec() with a smart pointer
3129 class, thus making it possible to process discontinuous strings.
3130
3131 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
3132 much the same function. They were added by different people who were trying
3133 to make PCRE easy to compile on non-Unix systems. It has been suggested
3134 that PCRE_EXPORT be abolished now that there is more automatic apparatus
3135 for compiling on Windows systems. I have therefore replaced it with
3136 PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
3137 defaults to "extern" for C or "extern C" for C++, which works fine on
3138 Unix-like systems. It is now possible to override the value of PCRE_DATA_
3139 SCOPE with something explicit in config.h. In addition:
3140
3141 (a) pcreposix.h still had just "extern" instead of either of these macros;
3142 I have replaced it with PCRE_DATA_SCOPE.
3143
3144 (b) Functions such as _pcre_xclass(), which are internal to the library,
3145 but external in the C sense, all had PCRE_EXPORT in their definitions.
3146 This is apparently wrong for the Windows case, so I have removed it.
3147 (It makes no difference on Unix-like systems.)
3148
3149 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
3150 of recursive calls to match(). This is different to MATCH_LIMIT because
3151 that limits the total number of calls to match(), not all of which increase
3152 the depth of recursion. Limiting the recursion depth limits the amount of
3153 stack (or heap if NO_RECURSE is set) that is used. The default can be set
3154 when PCRE is compiled, and changed at run time. A patch from Google adds
3155 this functionality to the C++ interface.
3156
3157 18. Changes to the handling of Unicode character properties:
3158
3159 (a) Updated the table to Unicode 4.1.0.
3160
3161 (b) Recognize characters that are not in the table as "Cn" (undefined).
3162
3163 (c) I revised the way the table is implemented to a much improved format
3164 which includes recognition of ranges. It now supports the ranges that
3165 are defined in UnicodeData.txt, and it also amalgamates other
3166 characters into ranges. This has reduced the number of entries in the
3167 table from around 16,000 to around 3,000, thus reducing its size
3168 considerably. I realized I did not need to use a tree structure after
3169 all - a binary chop search is just as efficient. Having reduced the
3170 number of entries, I extended their size from 6 bytes to 8 bytes to
3171 allow for more data.
3172
3173 (d) Added support for Unicode script names via properties such as \p{Han}.
3174
3175 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
3176 matching that character.
3177
3178 20. When matching a repeated Unicode property with a minimum greater than zero,
3179 (for example \pL{2,}), PCRE could look past the end of the subject if it
3180 reached it while seeking the minimum number of characters. This could
3181 happen only if some of the characters were more than one byte long, because
3182 there is a check for at least the minimum number of bytes.
3183
3184 21. Refactored the implementation of \p and \P so as to be more general, to
3185 allow for more different types of property in future. This has changed the
3186 compiled form incompatibly. Anybody with saved compiled patterns that use
3187 \p or \P will have to recompile them.
3188
3189 22. Added "Any" and "L&" to the supported property types.
3190
3191 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
3192 but give a compile time error if the value is greater than 0xff.
3193
3194 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
3195 accidentally not being installed or uninstalled.
3196
3197 25. The pcre.h file was built from pcre.h.in, but the only changes that were
3198 made were to insert the current release number. This seemed silly, because
3199 it made things harder for people building PCRE on systems that don't run
3200 "configure". I have turned pcre.h into a distributed file, no longer built
3201 by "configure", with the version identification directly included. There is
3202 no longer a pcre.h.in file.
3203
3204 However, this change necessitated a change to the pcre-config script as
3205 well. It is built from pcre-config.in, and one of the substitutions was the
3206 release number. I have updated configure.ac so that ./configure now finds
3207 the release number by grepping pcre.h.
3208
3209 26. Added the ability to run the tests under valgrind.
3210
3211
3212 Version 6.4 05-Sep-05
3213 ---------------------
3214
3215 1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
3216 "--" to be printed when multiple files were scanned, even when none of the
3217 -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
3218 consider it to be a bug, and have restored the previous behaviour.
3219
3220 2. A couple of code tidies to get rid of compiler warnings.
3221
3222 3. The pcretest program used to cheat by referring to symbols in the library
3223 whose names begin with _pcre_. These are internal symbols that are not
3224 really supposed to be visible externally, and in some environments it is
3225 possible to suppress them. The cheating is now confined to including
3226 certain files from the library's source, which is a bit cleaner.
3227
3228 4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
3229 file's purpose clearer.
3230
3231 5. Reorganized pcre_ucp_findchar().
3232
3233
3234 Version 6.3 15-Aug-05
3235 ---------------------
3236
3237 1. The file libpcre.pc.in did not have general read permission in the tarball.
3238
3239 2. There were some problems when building without C++ support:
3240
3241 (a) If C++ support was not built, "make install" and "make test" still
3242 tried to test it.
3243
3244 (b) There were problems when the value of CXX was explicitly set. Some
3245 changes have been made to try to fix these, and ...
3246
3247 (c) --disable-cpp can now be used to explicitly disable C++ support.
3248
3249 (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
3250 backslash in a target when C++ was disabled. This confuses some
3251 versions of "make", apparently. Using an intermediate variable solves
3252 this. (Same for CPP_LOBJ.)
3253
3254 3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
3255 (non-Windows) now includes $(CFLAGS) because these flags are sometimes
3256 necessary on certain architectures.
3257
3258 4. Added a setting of -export-symbols-regex to the link command to remove
3259 those symbols that are exported in the C sense, but actually are local
3260 within the library, and not documented. Their names all begin with
3261 "_pcre_". This is not a perfect job, because (a) we have to except some
3262 symbols that pcretest ("illegally") uses, and (b) the facility isn't always
3263 available (and never for static libraries). I have made a note to try to
3264 find a way round (a) in the future.
3265
3266
3267 Version 6.2 01-Aug-05
3268 ---------------------
3269
3270 1. There was no test for integer overflow of quantifier values. A construction
3271 such as {1111111111111111} would give undefined results. What is worse, if
3272 a minimum quantifier for a parenthesized subpattern overflowed and became
3273 negative, the calculation of the memory size went wrong. This could have
3274 led to memory overwriting.
3275
3276 2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
3277
3278 3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
3279 operating environments where this matters.
3280
3281 4. Applied Giuseppe Maxia's patch to add additional features for controlling
3282 PCRE options from within the C++ wrapper.
3283
3284 5. Named capturing subpatterns were not being correctly counted when a pattern
3285 was compiled. This caused two problems: (a) If there were more than 100
3286 such subpatterns, the calculation of the memory needed for the whole
3287 compiled pattern went wrong, leading to an overflow error. (b) Numerical
3288 back references of the form \12, where the number was greater than 9, were
3289 not recognized as back references, even though there were sufficient
3290 previous subpatterns.
3291
3292 6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
3293 versions of gcc, e.g. 2.95.4.
3294
3295
3296 Version 6.1 21-Jun-05
3297 ---------------------
3298
3299 1. There was one reference to the variable "posix" in pcretest.c that was not
3300 surrounded by "#if !defined NOPOSIX".
3301
3302 2. Make it possible to compile pcretest without DFA support, UTF8 support, or
3303 the cross-check on the old pcre_info() function, for the benefit of the
3304 cut-down version of PCRE that is currently imported into Exim.
3305
3306 3. A (silly) pattern starting with (?i)(?-i) caused an internal space
3307 allocation error. I've done the easy fix, which wastes 2 bytes for sensible
3308 patterns that start (?i) but I don't think that matters. The use of (?i) is
3309 just an example; this all applies to the other options as well.
3310
3311 4. Since libtool seems to echo the compile commands it is issuing, the output
3312 from "make" can be reduced a bit by putting "@" in front of each libtool
3313 compile command.
3314
3315 5. Patch from the folks at Google for configure.in to be a bit more thorough
3316 in checking for a suitable C++ installation before trying to compile the
3317 C++ stuff. This should fix a reported problem when a compiler was present,
3318 but no suitable headers.
3319
3320 6. The man pages all had just "PCRE" as their title. I have changed them to
3321 be the relevant file name. I have also arranged that these names are
3322 retained in the file doc/pcre.txt, which is a concatenation in text format
3323 of all the man pages except the little individual ones for each function.
3324
3325 7. The NON-UNIX-USE file had not been updated for the different set of source
3326 files that come with release 6. I also added a few comments about the C++
3327 wrapper.
3328
3329
3330 Version 6.0 07-Jun-05
3331 ---------------------
3332
3333 1. Some minor internal re-organization to help with my DFA experiments.
3334
3335 2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
3336 didn't matter for the library itself when fully configured, but did matter
3337 when compiling without UCP support, or within Exim, where the ucp files are
3338 not imported.
3339
3340 3. Refactoring of the library code to split up the various functions into
3341 different source modules. The addition of the new DFA matching code (see
3342 below) to a single monolithic source would have made it really too
3343 unwieldy, quite apart from causing all the code to be include in a
3344 statically linked application, when only some functions are used. This is
3345 relevant even without the DFA addition now that patterns can be compiled in
3346 one application and matched in another.
3347
3348 The downside of splitting up is that there have to be some external
3349 functions and data tables that are used internally in different modules of
3350 the library but which are not part of the API. These have all had their
3351 names changed to start with "_pcre_" so that they are unlikely to clash
3352 with other external names.
3353
3354 4. Added an alternate matching function, pcre_dfa_exec(), which matches using
3355 a different (DFA) algorithm. Although it is slower than the original
3356 function, it does have some advantages for certain types of matching
3357 problem.
3358
3359 5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
3360 including restarting after a partial match.
3361
3362 6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
3363 defined when compiling for Windows was sent to me. I have put it into the
3364 code, though I have no means of testing or verifying it.
3365
3366 7. Added the pcre_refcount() auxiliary function.
3367
3368 8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
3369 match before or at the first newline in the subject string. In pcretest,
3370 the /f option on a pattern can be used to set this.
3371
3372 9. A repeated \w when used in UTF-8 mode with characters greater than 256
3373 would behave wrongly. This has been present in PCRE since release 4.0.
3374
3375 10. A number of changes to the pcregrep command:
3376
3377 (a) Refactored how -x works; insert ^(...)$ instead of setting
3378 PCRE_ANCHORED and checking the length, in preparation for adding
3379 something similar for -w.
3380
3381 (b) Added the -w (match as a word) option.
3382
3383 (c) Refactored the way lines are read and buffered so as to have more
3384 than one at a time available.
3385
3386 (d) Implemented a pcregrep test script.
3387
3388 (e) Added the -M (multiline match) option. This allows patterns to match
3389 over several lines of the subject. The buffering ensures that at least
3390 8K, or the rest of the document (whichever is the shorter) is available
3391 for matching (and similarly the previous 8K for lookbehind assertions).
3392
3393 (f) Changed the --help output so that it now says
3394
3395 -w, --word-regex(p)
3396
3397 instead of two lines, one with "regex" and the other with "regexp"
3398 because that confused at least one person since the short forms are the
3399 same. (This required a bit of code, as the output is generated
3400 automatically from a table. It wasn't just a text change.)
3401
3402 (g) -- can be used to terminate pcregrep options if the next thing isn't an
3403 option but starts with a hyphen. Could be a pattern or a path name
3404 starting with a hyphen, for instance.
3405
3406 (h) "-" can be given as a file name to represent stdin.
3407
3408 (i) When file names are being printed, "(standard input)" is used for
3409 the standard input, for compatibility with GNU grep. Previously
3410 "<stdin>" was used.
3411
3412 (j) The option --label=xxx can be used to supply a name to be used for
3413 stdin when file names are being printed. There is no short form.
3414
3415 (k) Re-factored the options decoding logic because we are going to add
3416 two more options that take data. Such options can now be given in four
3417 different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
3418
3419 (l) Added the -A, -B, and -C options for requesting that lines of context
3420 around matches be printed.
3421
3422 (m) Added the -L option to print the names of files that do not contain
3423 any matching lines, that is, the complement of -l.
3424
3425 (n) The return code is 2 if any file cannot be opened, but pcregrep does
3426 continue to scan other files.
3427
3428 (o) The -s option was incorrectly implemented. For compatibility with other
3429 greps, it now suppresses the error message for a non-existent or non-
3430 accessible file (but not the return code). There is a new option called
3431 -q that suppresses the output of matching lines, which was what -s was
3432 previously doing.
3433
3434 (p) Added --include and --exclude options to specify files for inclusion
3435 and exclusion when recursing.
3436
3437 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
3438 Hopefully, it now does.
3439
3440 12. Missing cast in pcre_study().
3441
3442 13. Added an "uninstall" target to the makefile.
3443
3444 14. Replaced "extern" in the function prototypes in Makefile.in with
3445 "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
3446 world, but is set differently for Windows.
3447
3448 15. Added a second compiling function called pcre_compile2(). The only
3449 difference is that it has an extra argument, which is a pointer to an
3450 integer error code. When there is a compile-time failure, this is set
3451 non-zero, in addition to the error test pointer being set to point to an
3452 error message. The new argument may be NULL if no error number is required
3453 (but then you may as well call pcre_compile(), which is now just a
3454 wrapper). This facility is provided because some applications need a
3455 numeric error indication, but it has also enabled me to tidy up the way
3456 compile-time errors are handled in the POSIX wrapper.
3457
3458 16. Added VPATH=.libs to the makefile; this should help when building with one
3459 prefix path and installing with another. (Or so I'm told by someone who
3460 knows more about this stuff than I do.)
3461
3462 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
3463 passes PCRE_DOTALL to the pcre_compile() function, making the "." character
3464 match everything, including newlines. This is not POSIX-compatible, but
3465 somebody wanted the feature. From pcretest it can be activated by using
3466 both the P and the s flags.
3467
3468 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
3469
3470 19. libpcre.pc was being incorrectly installed as executable.
3471
3472 20. A couple of places in pcretest check for end-of-line by looking for '\n';
3473 it now also looks for '\r' so that it will work unmodified on Windows.
3474
3475 21. Added Google's contributed C++ wrapper to the distribution.
3476
3477 22. Added some untidy missing memory free() calls in pcretest, to keep
3478 Electric Fence happy when testing.
3479
3480
3481
3482 Version 5.0 13-Sep-04
3483 ---------------------
3484
3485 1. Internal change: literal characters are no longer packed up into items
3486 containing multiple characters in a single byte-string. Each character
3487 is now matched using a separate opcode. However, there may be more than one
3488 byte in the character in UTF-8 mode.
3489
3490 2. The pcre_callout_block structure has two new fields: pattern_position and
3491 next_item_length. These contain the offset in the pattern to the next match
3492 item, and its length, respectively.
3493
3494 3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
3495 insertion of callouts before each pattern item. Added the /C option to
3496 pcretest to make use of this.
3497
3498 4. On the advice of a Windows user, the lines
3499
3500 #if defined(_WIN32) || defined(WIN32)
3501 _setmode( _fileno( stdout ), 0x8000 );
3502 #endif /* defined(_WIN32) || defined(WIN32) */
3503
3504 have been added to the source of pcretest. This apparently does useful
3505 magic in relation to line terminators.
3506
3507 5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
3508 for the benefit of those environments where the "b" makes a difference.
3509
3510 6. The icc compiler has the same options as gcc, but "configure" doesn't seem
3511 to know about it. I have put a hack into configure.in that adds in code
3512 to set GCC=yes if CC=icc. This seems to end up at a point in the
3513 generated configure script that is early enough to affect the setting of
3514 compiler options, which is what is needed, but I have no means of testing
3515 whether it really works. (The user who reported this had patched the
3516 generated configure script, which of course I cannot do.)
3517
3518 LATER: After change 22 below (new libtool files), the configure script
3519 seems to know about icc (and also ecc). Therefore, I have commented out
3520 this hack in configure.in.
3521
3522 7. Added support for pkg-config (2 patches were sent in).
3523
3524 8. Negated POSIX character classes that used a combination of internal tables
3525 were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
3526 [[:^ascii]]. Typically, they would match almost any characters. The other
3527 POSIX classes were not broken in this way.
3528
3529 9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
3530 to find the match, as PCRE was deluded into thinking that the match had to
3531 start at the start point or following a newline. The same bug applied to
3532 patterns with negative forward assertions or any backward assertions
3533 preceding ".*" at the start, unless the pattern required a fixed first
3534 character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
3535
3536 10. In UTF-8 mode, when moving forwards in the subject after a failed match
3537 starting at the last subject character, bytes beyond the end of the subject
3538 string were read.
3539
3540 11. Renamed the variable "class" as "classbits" to make life easier for C++
3541 users. (Previously there was a macro definition, but it apparently wasn't
3542 enough.)
3543
3544 12. Added the new field "tables" to the extra data so that tables can be passed
3545 in at exec time, or the internal tables can be re-selected. This allows
3546 a compiled regex to be saved and re-used at a later time by a different
3547 program that might have everything at different addresses.
3548
3549 13. Modified the pcre-config script so that, when run on Solaris, it shows a
3550 -R library as well as a -L library.
3551
3552 14. The debugging options of pcretest (-d on the command line or D on a
3553 pattern) showed incorrect output for anything following an extended class
3554 that contained multibyte characters and which was followed by a quantifier.
3555
3556 15. Added optional support for general category Unicode character properties
3557 via the \p, \P, and \X escapes. Unicode property support implies UTF-8
3558 support. It adds about 90K to the size of the library. The meanings of the
3559 inbuilt class escapes such as \d and \s have NOT been changed.
3560
3561 16. Updated pcredemo.c to include calls to free() to release the memory for the
3562 compiled pattern.
3563
3564 17. The generated file chartables.c was being created in the source directory
3565 instead of in the building directory. This caused the build to fail if the
3566 source directory was different from the building directory, and was
3567 read-only.
3568
3569 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
3570 file. No doubt somebody will tell me if they don't make sense... Also added
3571 Dan Mooney's comments about building on OpenVMS.
3572
3573 19. Added support for partial matching via the PCRE_PARTIAL option for
3574 pcre_exec() and the \P data escape in pcretest.
3575
3576 20. Extended pcretest with 3 new pattern features:
3577
3578 (i) A pattern option of the form ">rest-of-line" causes pcretest to
3579 write the compiled pattern to the file whose name is "rest-of-line".
3580 This is a straight binary dump of the data, with the saved pointer to
3581 the character tables forced to be NULL. The study data, if any, is
3582 written too. After writing, pcretest reads a new pattern.
3583
3584 (ii) If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
3585 compiled pattern from the given file. There must not be any
3586 occurrences of "<" in the file name (pretty unlikely); if there are,
3587 pcretest will instead treat the initial "<" as a pattern delimiter.
3588 After reading in the pattern, pcretest goes on to read data lines as
3589 usual.
3590
3591 (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
3592 and 16-bit fields in a compiled pattern, to simulate a pattern that
3593 was compiled on a host of opposite endianness.
3594
3595 21. The pcre-exec() function can now cope with patterns that were compiled on
3596 hosts of opposite endianness, with this restriction:
3597
3598 As for any compiled expression that is saved and used later, the tables
3599 pointer field cannot be preserved; the extra_data field in the arguments
3600 to pcre_exec() should be used to pass in a tables address if a value
3601 other than the default internal tables were used at compile time.
3602
3603 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
3604 now diagnosed as an error. Previously, most of the time, a negative number
3605 would have been treated as zero, but if in addition "ovector" was passed as
3606 NULL, a crash could occur.
3607
3608 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
3609 new versions from the libtool 1.5 distribution (the last one is a copy of
3610 a file called libtool.m4). This seems to have fixed the need to patch
3611 "configure" to support Darwin 1.3 (which I used to do). However, I still
3612 had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
3613 workstation).
3614
3615 24. Changed the PCRE licence to be the more standard "BSD" licence.
3616
3617
3618 Version 4.5 01-Dec-03
3619 ---------------------
3620
3621 1. There has been some re-arrangement of the code for the match() function so
3622 that it can be compiled in a version that does not call itself recursively.
3623 Instead, it keeps those local variables that need separate instances for
3624 each "recursion" in a frame on the heap, and gets/frees frames whenever it
3625 needs to "recurse". Keeping track of where control must go is done by means
3626 of setjmp/longjmp. The whole thing is implemented by a set of macros that
3627 hide most of the details from the main code, and operates only if
3628 NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
3629 "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
3630 operating.
3631
3632 To make it easier for callers to provide specially tailored get/free
3633 functions for this usage, two new functions, pcre_stack_malloc, and
3634 pcre_stack_free, are used. They are always called in strict stacking order,
3635 and the size of block requested is always the same.
3636
3637 The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
3638 PCRE has been compiled to use the stack or the heap for recursion. The
3639 -C option of pcretest uses this to show which version is compiled.
3640
3641 A new data escape \S, is added to pcretest; it causes the amounts of store
3642 obtained and freed by both kinds of malloc/free at match time to be added
3643 to the output.
3644
3645 2. Changed the locale test to use "fr_FR" instead of "fr" because that's
3646 what's available on my current Linux desktop machine.
3647
3648 3. When matching a UTF-8 string, the test for a valid string at the start has
3649 been extended. If start_offset is not zero, PCRE now checks that it points
3650 to a byte that is the start of a UTF-8 character. If not, it returns
3651 PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
3652 this is necessary because there may be backward assertions in the pattern.
3653 When matching the same subject several times, it may save resources to use
3654 PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
3655
3656 4. The code for checking the validity of UTF-8 strings has been tightened so
3657 that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
3658 containing "overlong sequences".
3659
3660 5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
3661 I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
3662 should have been "&", but it just so happened that all the cases this let
3663 through by mistake were picked up later in the function.
3664
3665 6. I had used a variable called "isblank" - this is a C99 function, causing
3666 some compilers to warn. To avoid this, I renamed it (as "blankclass").
3667
3668 7. Cosmetic: (a) only output another newline at the end of pcretest if it is
3669 prompting; (b) run "./pcretest /dev/null" at the start of the test script
3670 so the version is shown; (c) stop "make test" echoing "./RunTest".
3671
3672 8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
3673
3674 9. The prototype for memmove() for systems that don't have it was using
3675 size_t, but the inclusion of the header that defines size_t was later. I've
3676 moved the #includes for the C headers earlier to avoid this.
3677
3678 10. Added some adjustments to the code to make it easier to compiler on certain
3679 special systems:
3680
3681 (a) Some "const" qualifiers were missing.
3682 (b) Added the macro EXPORT before all exported functions; by default this
3683 is defined to be empty.
3684 (c) Changed the dftables auxiliary program (that builds chartables.c) so
3685 that it reads its output file name as an argument instead of writing
3686 to the standard output and assuming this can be redirected.
3687
3688 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
3689 class containing characters with values greater than 255, PCRE compilation
3690 went into a loop.
3691
3692 12. A recursive reference to a subpattern that was within another subpattern
3693 that had a minimum quantifier of zero caused PCRE to crash. For example,
3694 (x(y(?2))z)? provoked this bug with a subject that got as far as the
3695 recursion. If the recursively-called subpattern itself had a zero repeat,
3696 that was OK.
3697
3698 13. In pcretest, the buffer for reading a data line was set at 30K, but the
3699 buffer into which it was copied (for escape processing) was still set at
3700 1024, so long lines caused crashes.
3701
3702 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
3703 "internal error: code overflow...". This applied to any character class
3704 that was followed by a possessive quantifier.
3705
3706 15. Modified the Makefile to add libpcre.la as a prerequisite for
3707 libpcreposix.la because I was told this is needed for a parallel build to
3708 work.
3709
3710 16. If a pattern that contained .* following optional items at the start was
3711 studied, the wrong optimizing data was generated, leading to matching
3712 errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
3713 matching string must start with a or b or c. The correct conclusion for
3714 this pattern is that a match can start with any character.
3715
3716
3717 Version 4.4 13-Aug-03
3718 ---------------------
3719
3720 1. In UTF-8 mode, a character class containing characters with values between
3721 127 and 255 was not handled correctly if the compiled pattern was studied.
3722 In fixing this, I have also improved the studying algorithm for such
3723 classes (slightly).
3724
3725 2. Three internal functions had redundant arguments passed to them. Removal
3726 might give a very teeny performance improvement.
3727
3728 3. Documentation bug: the value of the capture_top field in a callout is *one
3729 more than* the number of the hightest numbered captured substring.
3730
3731 4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
3732 in incorrectly linking with a previously installed version. They now link
3733 explicitly with libpcre.la.
3734
3735 5. configure.in no longer needs to recognize Cygwin specially.
3736
3737 6. A problem in pcre.in for Windows platforms is fixed.
3738
3739 7. If a pattern was successfully studied, and the -d (or /D) flag was given to
3740 pcretest, it used to include the size of the study block as part of its
3741 output. Unfortunately, the structure contains a field that has a different
3742 size on different hardware architectures. This meant that the tests that
3743 showed this size failed. As the block is currently always of a fixed size,
3744 this information isn't actually particularly useful in pcretest output, so
3745 I have just removed it.
3746
3747 8. Three pre-processor statements accidentally did not start in column 1.
3748 Sadly, there are *still* compilers around that complain, even though
3749 standard C has not required this for well over a decade. Sigh.
3750
3751 9. In pcretest, the code for checking callouts passed small integers in the
3752 callout_data field, which is a void * field. However, some picky compilers
3753 complained about the casts involved for this on 64-bit systems. Now
3754 pcretest passes the address of the small integer instead, which should get
3755 rid of the warnings.
3756
3757 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
3758 both compile and run time, and gives an error if an invalid UTF-8 sequence
3759 is found. There is a option for disabling this check in cases where the
3760 string is known to be correct and/or the maximum performance is wanted.
3761
3762 11. In response to a bug report, I changed one line in Makefile.in from
3763
3764 -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
3765 to
3766 -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
3767
3768 to look similar to other lines, but I have no way of telling whether this
3769 is the right thing to do, as I do not use Windows. No doubt I'll get told
3770 if it's wrong...
3771
3772
3773 Version 4.3 21-May-03
3774 ---------------------
3775
3776 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
3777 Makefile.
3778
3779 2. Some refactoring to improve the quality of the code:
3780
3781 (i) The utf8_table... variables are now declared "const".
3782
3783 (ii) The code for \cx, which used the "case flipping" table to upper case
3784 lower case letters, now just substracts 32. This is ASCII-specific,
3785 but the whole concept of \cx is ASCII-specific, so it seems
3786 reasonable.
3787
3788 (iii) PCRE was using its character types table to recognize decimal and
3789 hexadecimal digits in the pattern. This is silly, because it handles
3790 only 0-9, a-f, and A-F, but the character types table is locale-
3791 specific, which means strange things might happen. A private
3792 table is now used for this - though it costs 256 bytes, a table is
3793 much faster than multiple explicit tests. Of course, the standard
3794 character types table is still used for matching digits in subject
3795 strings against \d.
3796
3797 (iv) Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
3798 ending in _t are). So I've renamed it as ESC_tee.
3799
3800 3. The first argument for regexec() in the POSIX wrapper should have been
3801 defined as "const".
3802
3803 4. Changed pcretest to use malloc() for its buffers so that they can be
3804 Electric Fenced for debugging.
3805
3806 5. There were several places in the code where, in UTF-8 mode, PCRE would try
3807 to read one or more bytes before the start of the subject string. Often this
3808 had no effect on PCRE's behaviour, but in some circumstances it could
3809 provoke a segmentation fault.
3810
3811 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
3812 to try to read one or more bytes before the start of the subject string.
3813
3814 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
3815 UTF-8 support could misbehave in various ways if the subject string
3816 contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
3817 area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
3818 back over UTF-8 characters.)
3819
3820
3821 Version 4.2 14-Apr-03
3822 ---------------------
3823
3824 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
3825
3826 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
3827 [ON_WINDOWS]: new variable, "#" on non-Windows platforms
3828 [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
3829 [WIN_PREFIX]: new variable, "cyg" for Cygwin
3830 * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
3831 and BUILD_EXEEXT
3832 Note: automatic setting of the BUILD variables is not yet working
3833 set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
3834 compile-time but not at link-time
3835 [LINK]: use for linking executables only
3836 make different versions for Windows and non-Windows
3837 [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
3838 libraries
3839 [LINK_FOR_BUILD]: new variable
3840 [OBJEXT]: use throughout
3841 [EXEEXT]: use throughout
3842 <winshared>: new target
3843 <wininstall>: new target
3844 <dftables.o>: use native compiler
3845 <dftables>: use native linker
3846 <install>: handle Windows platform correctly
3847 <clean>: ditto
3848 <check>: ditto
3849 copy DLL to top builddir before testing
3850
3851 As part of these changes, -no-undefined was removed again. This was reported
3852 to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
3853 in any case.
3854
3855 3. Some tidies to get rid of compiler warnings:
3856
3857 . In the match_data structure, match_limit was an unsigned long int, whereas
3858 match_call_count was an int. I've made them both unsigned long ints.
3859
3860 . In pcretest the fact that a const uschar * doesn't automatically cast to
3861 a void * provoked a warning.
3862
3863 . Turning on some more compiler warnings threw up some "shadow" variables
3864 and a few more missing casts.
3865
3866 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
3867 option, a class that contained a single character with a value between 128
3868 and 255 (e.g. /[\xFF]/) caused PCRE to crash.
3869
3870 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
3871 option, a class that contained several characters, but with at least one
3872 whose value was between 128 and 255 caused PCRE to crash.
3873
3874
3875 Version 4.1 12-Mar-03
3876 ---------------------
3877
3878 1. Compiling with gcc -pedantic found a couple of places where casts were
3879 needed, and a string in dftables.c that was longer than standard compilers are
3880 required to support.
3881
3882 2. Compiling with Sun's compiler found a few more places where the code could
3883 be tidied up in order to avoid warnings.
3884
3885 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
3886 first of these names is deprecated in the latest Autoconf in favour of the name
3887 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
3888 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
3889 analogy I have changed it to CFLAGS_FOR_BUILD.
3890
3891 4. Added -no-undefined to the linking command in the Makefile, because this is
3892 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
3893 linking step for the pcreposix library.
3894
3895 5. PCRE was failing to diagnose the case of two named groups with the same
3896 name.
3897
3898 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
3899 literal character that is needed in the subject for a match, and scans along to
3900 ensure that it is present before embarking on the full matching process. This
3901 saves time in cases of nested unlimited repeats that are never going to match.
3902 Problem: the scan can take a lot of time if the subject is very long (e.g.
3903 megabytes), thus penalizing straightforward matches. It is now done only if the
3904 amount of subject to be scanned is less than 1000 bytes.
3905
3906 7. A lesser problem with the same optimization is that it was recording the
3907 first character of an anchored pattern as "needed", thus provoking a search
3908 right along the subject, even when the first match of the pattern was going to
3909 fail. The "needed" character is now not set for anchored patterns, unless it
3910 follows something in the pattern that is of non-fixed length. Thus, it still
3911 fulfils its original purpose of finding quick non-matches in cases of nested
3912 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
3913
3914
3915 Version 4.0 17-Feb-03
3916 ---------------------
3917
3918 1. If a comment in an extended regex that started immediately after a meta-item
3919 extended to the end of string, PCRE compiled incorrect data. This could lead to
3920 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
3921
3922 2. Moved to autoconf 2.53 and libtool 1.4.2.
3923
3924 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
3925 the special perltest8 script is no longer needed - all the tests can be run
3926 from a single perltest script.
3927
3928 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
3929 by \s. It has now been removed in PCRE. This means it isn't recognized as
3930 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
3931 class [:space:] *does* include VT, thereby creating a mess.
3932
3933 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
3934 space and tab.
3935
3936 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
3937 its new features into the main test script, reducing the number of scripts.
3938
3939 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
3940 were backward compatible, and made the (?i) apply to the whole pattern, as if
3941 /i were given. Now it behaves more logically, and applies the option setting
3942 only to what follows. PCRE has been changed to follow suit. However, if it
3943 finds options settings right at the start of the pattern, it extracts them into
3944 the global options, as before. Thus, they show up in the info data.
3945
3946 8. Added support for the \Q...\E escape sequence. Characters in between are
3947 treated as literals. This is slightly different from Perl in that $ and @ are
3948 also handled as literals inside the quotes. In Perl, they will cause variable
3949 interpolation. Note the following examples:
3950
3951 Pattern PCRE matches Perl matches
3952
3953 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
3954 \Qabc\$xyz\E abc\$xyz abc\$xyz
3955 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
3956
3957 For compatibility with Perl, \Q...\E sequences are recognized inside character
3958 classes as well as outside them.
3959
3960 9. Re-organized 3 code statements in pcretest to avoid "overflow in
3961 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
3962 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
3963 signed/unsigned warnings.
3964
3965 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
3966 option for pcretest, so I've replaced it by a simple function that does just
3967 that job.
3968
3969 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
3970 "pcregrep -".
3971
3972 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
3973 Java package. This provides some syntactic sugar for simple cases of what my
3974 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
3975 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
3976 item, you can use this simplified notation. Note that only makes sense with
3977 greedy quantifiers. Consequently, the use of the possessive quantifier forces
3978 greediness, whatever the setting of the PCRE_UNGREEDY option.
3979
3980 13. A change of greediness default within a pattern was not taking effect at
3981 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
3982 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
3983 was abstracted outside.
3984
3985 14. PCRE now supports the \G assertion. It is true when the current matching
3986 position is at the start point of the match. This differs from \A when the
3987 starting offset is non-zero. Used with the /g option of pcretest (or similar
3988 code), it works in the same way as it does for Perl's /g option. If all
3989 alternatives of a regex begin with \G, the expression is anchored to the start
3990 match position, and the "anchored" flag is set in the compiled expression.
3991
3992 15. Some bugs concerning the handling of certain option changes within patterns
3993 have been fixed. These applied to options other than (?ims). For example,
3994 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
3995 been the other way round. Some of this was related to change 7 above.
3996
3997 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
3998 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
3999 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
4000 POSIX classes only within a class (e.g. /[[:alpha:]]/).
4001
4002 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
4003 mode. Unlike ".", it always matches newline, whatever the setting of
4004 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
4005 assertions. Perl allows it, but it doesn't (in general) work because it can't
4006 calculate the length of the lookbehind. At least, that's the case for Perl
4007 5.8.0 - I've been told they are going to document that it doesn't work in
4008 future.
4009
4010 18. Added an error diagnosis for escapes that PCRE does not support: these are
4011 \L, \l, \N, \P, \p, \U, \u, and \X.
4012
4013 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
4014 reading past the end of the pattern in cases such as /[abcd/.
4015
4016 20. PCRE was getting more memory than necessary for patterns with classes that
4017 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
4018
4019 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
4020 compiling PCRE for use with Virtual Pascal.
4021
4022 22. Small fix to the Makefile to make it work properly if the build is done
4023 outside the source tree.
4024
4025 23. Added a new extension: a condition to go with recursion. If a conditional
4026 subpattern starts with (?(R) the "true" branch is used if recursion has
4027 happened, whereas the "false" branch is used only at the top level.
4028
4029 24. When there was a very long string of literal characters (over 255 bytes
4030 without UTF support, over 250 bytes with UTF support), the computation of how
4031 much memory was required could be incorrect, leading to segfaults or other
4032 strange effects.
4033
4034 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
4035 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
4036 there was a subsequent back reference to those brackets. This meant that, for
4037 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
4038 possible to check for precisely this case. All we can do is abandon the
4039 optimization if .* occurs inside capturing brackets when there are any back
4040 references whatsoever. (See below for a better fix that came later.)
4041
4042 26. The handling of the optimization for finding the first character of a
4043 non-anchored pattern, and for finding a character that is required later in the
4044 match were failing in some cases. This didn't break the matching; it just
4045 failed to optimize when it could. The way this is done has been re-implemented.
4046
4047 27. Fixed typo in error message for invalid (?R item (it said "(?p").
4048
4049 28. Added a new feature that provides some of the functionality that Perl
4050 provides with (?{...}). The facility is termed a "callout". The way it is done
4051 in PCRE is for the caller to provide an optional function, by setting
4052 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
4053 global variable. By default it is unset, which disables all calling out. To get
4054 the function called, the regex must include (?C) at appropriate points. This
4055 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
4056 This provides a means of identifying different callout points. When PCRE
4057 reaches such a point in the regex, if pcre_callout has been set, the external
4058 function is called. It is provided with data in a structure called
4059 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
4060 matching continues; if it returns a non-zero value, the match at the current
4061 point fails. However, backtracking will occur if possible. [This was changed
4062 later and other features added - see item 49 below.]
4063
4064 29. pcretest is upgraded to test the callout functionality. It provides a
4065 callout function that displays information. By default, it shows the start of
4066 the match and the current position in the text. There are some new data escapes
4067 to vary what happens:
4068
4069 \C+ in addition, show current contents of captured substrings
4070 \C- do not supply a callout function
4071 \C!n return 1 when callout number n is reached
4072 \C!n!m return 1 when callout number n is reached for the mth time
4073
4074 30. If pcregrep was called with the -l option and just a single file name, it
4075 output "<stdin>" if a match was found, instead of the file name.
4076
4077 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
4078 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
4079 pcre_exec(). This saves a malloc/free per call. The default value of
4080 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
4081 when configuring.
4082
4083 32. The default maximum size of a compiled pattern is 64K. There have been a
4084 few cases of people hitting this limit. The code now uses macros to handle the
4085 storing of links as offsets within the compiled pattern. It defaults to 2-byte
4086 links, but this can be changed to 3 or 4 bytes by --with-link-size when
4087 configuring. Tests 2 and 5 work only with 2-byte links because they output
4088 debugging information about compiled patterns.
4089
4090 33. Internal code re-arrangements:
4091
4092 (a) Moved the debugging function for printing out a compiled regex into
4093 its own source file (printint.c) and used #include to pull it into
4094 pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
4095 separate copies.
4096
4097 (b) Defined the list of op-code names for debugging as a macro in
4098 internal.h so that it is next to the definition of the opcodes.
4099
4100 (c) Defined a table of op-code lengths for simpler skipping along compiled
4101 code. This is again a macro in internal.h so that it is next to the
4102 definition of the opcodes.
4103
4104 34. Added support for recursive calls to individual subpatterns, along the
4105 lines of Robin Houston's patch (but implemented somewhat differently).
4106
4107 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
4108 allow it to read and process whole directories in Win32. This code was
4109 contributed by Lionel Fourquaux; it has not been tested by me.
4110
4111 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
4112 used to name a group. Names consist of alphanumerics and underscores, and must
4113 be unique. Back references use the syntax (?P=name) and recursive calls use
4114 (?P>name) which is a PCRE extension to the Python extension. Groups still have
4115 numbers. The function pcre_fullinfo() can be used after compilation to extract
4116 a name/number map. There are three relevant calls:
4117
4118 PCRE_INFO_NAMEENTRYSIZE yields the size of each entry in the map
4119 PCRE_INFO_NAMECOUNT yields the number of entries
4120 PCRE_INFO_NAMETABLE yields a pointer to the map.
4121
4122 The map is a vector of fixed-size entries. The size of each entry depends on
4123 the length of the longest name used. The first two bytes of each entry are the
4124 group number, most significant byte first. There follows the corresponding
4125 name, zero terminated. The names are in alphabetical order.
4126
4127 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
4128 case instead of 255. Making it the same both with and without UTF-8 support
4129 means that the same test output works with both.
4130
4131 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
4132 calling malloc() with a zero argument.
4133
4134 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
4135 optimization. I've improved things by keeping a bitmap of backreferences with
4136 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
4137 fact referenced, the optimization can be applied. It is unlikely that a
4138 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
4139 the match to follow \n) will appear inside brackets with a number greater than
4140 31, but if it does, any back reference > 31 suppresses the optimization.
4141
4142 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
4143 of disabling numbered capturing parentheses. Any opening parenthesis that is
4144 not followed by ? behaves as if it were followed by ?: but named parentheses
4145 can still be used for capturing (and they will acquire numbers in the usual
4146 way).
4147
4148 41. Redesigned the return codes from the match() function into yes/no/error so
4149 that errors can be passed back from deep inside the nested calls. A malloc
4150 failure while inside a recursive subpattern call now causes the
4151 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
4152
4153 42. It is now possible to set a limit on the number of times the match()
4154 function is called in a call to pcre_exec(). This facility makes it possible to
4155 limit the amount of recursion and backtracking, though not in a directly
4156 obvious way, because the match() function is used in a number of different
4157 circumstances. The count starts from zero for each position in the subject
4158 string (for non-anchored patterns). The default limit is, for compatibility, a
4159 large number, namely 10 000 000. You can change this in two ways:
4160
4161 (a) When configuring PCRE before making, you can use --with-match-limit=n
4162 to set a default value for the compiled library.
4163
4164 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
4165 a different value is set. See 45 below.
4166
4167 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
4168
4169 43. Added a new function pcre_config(int, void *) to enable run-time extraction
4170 of things that can be changed at compile time. The first argument specifies
4171 what is wanted and the second points to where the information is to be placed.
4172 The current list of available information is:
4173
4174 PCRE_CONFIG_UTF8
4175
4176 The output is an integer that is set to one if UTF-8 support is available;
4177 otherwise it is set to zero.
4178
4179 PCRE_CONFIG_NEWLINE
4180
4181 The output is an integer that it set to the value of the code that is used for
4182 newline. It is either LF (10) or CR (13).
4183
4184 PCRE_CONFIG_LINK_SIZE
4185
4186 The output is an integer that contains the number of bytes used for internal
4187 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
4188
4189 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
4190
4191 The output is an integer that contains the threshold above which the POSIX
4192 interface uses malloc() for output vectors. See item 31 above.
4193
4194 PCRE_CONFIG_MATCH_LIMIT
4195
4196 The output is an unsigned integer that contains the default limit of the number
4197 of match() calls in a pcre_exec() execution. See 42 above.
4198
4199 44. pcretest has been upgraded by the addition of the -C option. This causes it
4200 to extract all the available output from the new pcre_config() function, and to
4201 output it. The program then exits immediately.
4202
4203 45. A need has arisen to pass over additional data with calls to pcre_exec() in
4204 order to support additional features. One way would have been to define
4205 pcre_exec2() (for example) with extra arguments, but this would not have been
4206 extensible, and would also have required all calls to the original function to
4207 be mapped to the new one. Instead, I have chosen to extend the mechanism that
4208 is used for passing in "extra" data from pcre_study().
4209
4210 The pcre_extra structure is now exposed and defined in pcre.h. It currently
4211 contains the following fields:
4212
4213 flags a bitmap indicating which of the following fields are set
4214 study_data opaque data from pcre_study()
4215 match_limit a way of specifying a limit on match() calls for a specific
4216 call to pcre_exec()
4217 callout_data data for callouts (see 49 below)
4218
4219 The flag bits are also defined in pcre.h, and are
4220
4221 PCRE_EXTRA_STUDY_DATA
4222 PCRE_EXTRA_MATCH_LIMIT
4223 PCRE_EXTRA_CALLOUT_DATA
4224
4225 The pcre_study() function now returns one of these new pcre_extra blocks, with
4226 the actual study data pointed to by the study_data field, and the
4227 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
4228 before. That is, this change is entirely upwards-compatible and requires no
4229 change to existing code.
4230
4231 If you want to pass in additional data to pcre_exec(), you can either place it
4232 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
4233 block.
4234
4235 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
4236 data string contains the escape sequence \M, pcretest calls pcre_exec() several
4237 times with different match limits, until it finds the minimum value needed for
4238 pcre_exec() to complete. The value is then output. This can be instructive; for
4239 most simple matches the number is quite small, but for pathological cases it
4240 gets very large very quickly.
4241
4242 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
4243 returns the size of the data block pointed to by the study_data field in a
4244 pcre_extra block, that is, the value that was passed as the argument to
4245 pcre_malloc() when PCRE was getting memory in which to place the information
4246 created by pcre_study(). The fourth argument should point to a size_t variable.
4247 pcretest has been extended so that this information is shown after a successful
4248 pcre_study() call when information about the compiled regex is being displayed.
4249
4250 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
4251 because what follows is always an absolute path. (Later: it turns out that this
4252 is more than cosmetic for MinGW, because it doesn't like empty path
4253 components.)
4254
4255 49. Some changes have been made to the callout feature (see 28 above):
4256
4257 (i) A callout function now has three choices for what it returns:
4258
4259 0 => success, carry on matching
4260 > 0 => failure at this point, but backtrack if possible
4261 < 0 => serious error, return this value from pcre_exec()
4262
4263 Negative values should normally be chosen from the set of PCRE_ERROR_xxx
4264 values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
4265 "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
4266 use by callout functions. It will never be used by PCRE itself.
4267
4268 (ii) The pcre_extra structure (see 45 above) has a void * field called
4269 callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
4270 pcre_callout_block structure has a field of the same name. The contents of
4271 the field passed in the pcre_extra structure are passed to the callout
4272 function in the corresponding field in the callout block. This makes it
4273 easier to use the same callout-containing regex from multiple threads. For
4274 testing, the pcretest program has a new data escape
4275
4276 \C*n pass the number n (may be negative) as callout_data
4277
4278 If the callout function in pcretest receives a non-zero value as
4279 callout_data, it returns that value.
4280
4281 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
4282 there were some redundant $(CFLAGS) in commands that are now specified as
4283 $(LINK), which already includes $(CFLAGS).
4284
4285 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
4286 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
4287 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
4288 one-byte characters throughout. Note that case-insensitive matching applies
4289 only to characters whose values are less than 256. PCRE doesn't support the
4290 notion of cases for higher-valued characters.
4291
4292 (i) A character class whose characters are all within 0-255 is handled as
4293 a bit map, and the map is inverted for negative classes. Previously, a
4294 character > 255 always failed to match such a class; however it should
4295 match if the class was a negative one (e.g. [^ab]). This has been fixed.
4296
4297 (ii) A negated character class with a single character < 255 is coded as
4298 "not this character" (OP_NOT). This wasn't working properly when the test
4299 character was multibyte, either singly or repeated.
4300
4301 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
4302 mode, for example: \x{100}{2,3}.
4303
4304 (iv) The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
4305 singly or repeated) now correctly test multibyte characters. However,
4306 PCRE doesn't recognize any characters with values greater than 255 as
4307 digits, spaces, or word characters. Such characters always match \D, \S,
4308 and \W, and never match \d, \s, or \w.
4309
4310 (v) Classes may now contain characters and character ranges with values
4311 greater than 255. For example: [ab\x{100}-\x{400}].
4312
4313 (vi) pcregrep now has a --utf-8 option (synonym -u) which makes it call
4314 PCRE in UTF-8 mode.
4315
4316 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
4317 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
4318 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
4319 value.)
4320
4321 53. The single man page has become too large. I have therefore split it up into
4322 a number of separate man pages. These also give rise to individual HTML pages;
4323 these are now put in a separate directory, and there is an index.html page that
4324 lists them all. Some hyperlinking between the pages has been installed.
4325
4326 54. Added convenience functions for handling named capturing parentheses.
4327
4328 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
4329 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
4330 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
4331 are faulted.
4332
4333 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
4334 calling configure. These values are used when compiling the dftables.c program
4335 which is run to generate the source of the default character tables. They
4336 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
4337 you will need to set these values.
4338
4339 57. Updated the building process for Windows DLL, as provided by Fred Cox.
4340
4341
4342 Version 3.9 02-Jan-02
4343 ---------------------
4344
4345 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
4346
4347 2. If --disable-static was given, the building process failed when trying to
4348 build pcretest and pcregrep. (For some reason it was using libtool to compile
4349 them, which is not right, as they aren't part of the library.)
4350
4351
4352 Version 3.8 18-Dec-01
4353 ---------------------
4354
4355 1. The experimental UTF-8 code was completely screwed up. It was packing the
4356 bytes in the wrong order. How dumb can you get?
4357
4358
4359 Version 3.7 29-Oct-01
4360 ---------------------
4361
4362 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
4363 This caused pcretest, when used on the test data, to segfault. Unfortunately,
4364 this didn't happen under Solaris 8, where I normally test things.
4365
4366 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
4367 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
4368 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
4369 this fix an hour or so after the initial 3.7 release.)
4370
4371
4372 Version 3.6 23-Oct-01
4373 ---------------------
4374
4375 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
4376 offsets passed as NULL with zero offset count.
4377
4378 2. The config.guess and config.sub files had not been updated when I moved to
4379 the latest autoconf.
4380
4381
4382 Version 3.5 15-Aug-01
4383 ---------------------
4384
4385 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
4386 had been forgotten.
4387
4388 2. By using declared but undefined structures, we can avoid using "void"
4389 definitions in pcre.h while keeping the internal definitions of the structures
4390 private.
4391
4392 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
4393 user point of view, this means that both static and shared libraries are built
4394 by default, but this can be individually controlled. More of the work of
4395 handling this static/shared cases is now inside libtool instead of PCRE's make
4396 file.
4397
4398 4. The pcretest utility is now installed along with pcregrep because it is
4399 useful for users (to test regexs) and by doing this, it automatically gets
4400 relinked by libtool. The documentation has been turned into a man page, so
4401 there are now .1, .txt, and .html versions in /doc.
4402
4403 5. Upgrades to pcregrep:
4404 (i) Added long-form option names like gnu grep.
4405 (ii) Added --help to list all options with an explanatory phrase.
4406 (iii) Added -r, --recursive to recurse into sub-directories.
4407 (iv) Added -f, --file to read patterns from a file.
4408
4409 6. pcre_exec() was referring to its "code" argument before testing that
4410 argument for NULL (and giving an error if it was NULL).
4411
4412 7. Upgraded Makefile.in to allow for compiling in a different directory from
4413 the source directory.
4414
4415 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
4416 options bits, the pointer it was passed was to an int instead of to an unsigned
4417 long int. This mattered only on 64-bit systems.
4418
4419 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
4420 generated) instead of pcre.in, which it its source. Also made the same change
4421 in several of the .c files.
4422
4423 10. A new release of gcc defines printf() as a macro, which broke pcretest
4424 because it had an ifdef in the middle of a string argument for printf(). Fixed
4425 by using separate calls to printf().
4426
4427 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
4428 script, to force use of CR or LF instead of \n in the source. On non-Unix
4429 systems, the value can be set in config.h.
4430
4431 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
4432 absolute limit. Changed the text of the error message to make this clear, and
4433 likewise updated the man page.
4434
4435 13. The limit of 99 on the number of capturing subpatterns has been removed.
4436 The new limit is 65535, which I hope will not be a "real" limit.
4437
4438
4439 Version 3.4 22-Aug-00
4440 ---------------------
4441
4442 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
4443
4444 2. Diagnose condition (?(0) as an error instead of crashing on matching.
4445
4446
4447 Version 3.3 01-Aug-00
4448 ---------------------
4449
4450 1. If an octal character was given, but the value was greater than \377, it
4451 was not getting masked to the least significant bits, as documented. This could
4452 lead to crashes in some systems.
4453
4454 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
4455 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
4456
4457 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
4458 These just pass their arguments on to (pcre_free)(), but they are provided
4459 because some uses of PCRE bind it to non-C systems that can call its functions,
4460 but cannot call free() or pcre_free() directly.
4461
4462 4. Add "make test" as a synonym for "make check". Corrected some comments in
4463 the Makefile.
4464
4465 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
4466 Makefile.
4467
4468 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
4469 command called pgrep for grepping around the active processes.
4470
4471 7. Added the beginnings of support for UTF-8 character strings.
4472
4473 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
4474 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
4475 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
4476 out for the ar command.)
4477
4478
4479 Version 3.2 12-May-00
4480 ---------------------
4481
4482 This is purely a bug fixing release.
4483
4484 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
4485 of ZA. This was just one example of several cases that could provoke this bug,
4486 which was introduced by change 9 of version 2.00. The code for breaking
4487 infinite loops after an iteration that matches an empty string was't working
4488 correctly.
4489
4490 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
4491 when matched against abbab (for example). After matching an empty string, it
4492 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
4493 caused it to match further down the string than it should.
4494
4495 3. The code contained an inclusion of sys/types.h. It isn't clear why this
4496 was there because it doesn't seem to be needed, and it causes trouble on some
4497 systems, as it is not a Standard C header. It has been removed.
4498
4499 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
4500 were reported on the Macintosh. The changes were from
4501
4502 while ((c = *(++ptr)) != 0 && c != '\n');
4503 to
4504 while ((c = *(++ptr)) != 0 && c != '\n') ;
4505
4506 Totally extraordinary, but if that's what it takes...
4507
4508 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
4509 available. Added HAVE_BCOPY and an autoconf test for it; if neither
4510 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
4511 assumes the way PCRE uses memmove() (always moving upwards).
4512
4513 6. PCRE is being used in one environment where strchr() is not available. There
4514 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
4515 faster code anyway.
4516
4517
4518 Version 3.1 09-Feb-00
4519 ---------------------
4520
4521 The only change in this release is the fixing of some bugs in Makefile.in for
4522 the "install" target:
4523
4524 (1) It was failing to install pcreposix.h.
4525
4526 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
4527
4528
4529 Version 3.0 01-Feb-00
4530 ---------------------
4531
4532 1. Add support for the /+ modifier to perltest (to output $` like it does in
4533 pcretest).
4534
4535 2. Add support for the /g modifier to perltest.
4536
4537 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
4538 matches null strings.
4539
4540 4. Fix perltest so that it doesn't do unwanted things when fed an empty
4541 pattern. Perl treats empty patterns specially - it reuses the most recent
4542 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
4543 effect.
4544
4545 5. The POSIX interface was broken in that it was just handing over the POSIX
4546 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
4547 required a bigger vector, with some working space on the end. This means that
4548 the POSIX wrapper now has to get and free some memory, and copy the results.
4549
4550 6. Added some simple autoconf support, placing the test data and the
4551 documentation in separate directories, re-organizing some of the
4552 information files, and making it build pcre-config (a GNU standard). Also added
4553 libtool support for building PCRE as a shared library, which is now the
4554 default.
4555
4556 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
4557 09 are not valid octal constants. Single digits will be used for minor values
4558 less than 10.
4559
4560 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
4561 existing programs that set these in the POSIX interface can use PCRE without
4562 modification.
4563
4564 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
4565 return all that pcre_info() returns, plus additional data. The pcre_info()
4566 function is retained for compatibility, but is considered to be obsolete.
4567
4568 10. Added experimental recursion feature (?R) to handle one common case that
4569 Perl 5.6 will be able to do with (?p{...}).
4570
4571 11. Added support for POSIX character classes like [:alpha:], which Perl is
4572 adopting.
4573
4574
4575 Version 2.08 31-Aug-99
4576 ----------------------
4577
4578 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
4579 trying to match at the startoffset position, but instead was moving forward to
4580 the next newline as if a previous match had failed.
4581
4582 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
4583 and could get into a loop if a null string was matched other than at the start
4584 of the subject.
4585
4586 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
4587 be distinguished at compile time, and for completeness also added PCRE_DATE.
4588
4589 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
4590 in GnuWin32 environments.
4591
4592
4593 Version 2.07 29-Jul-99
4594 ----------------------
4595
4596 1. The documentation is now supplied in plain text form and HTML as well as in
4597 the form of man page sources.
4598
4599 2. C++ compilers don't like assigning (void *) values to other pointer types.
4600 In particular this affects malloc(). Although there is no problem in Standard
4601 C, I've put in casts to keep C++ compilers happy.
4602
4603 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
4604 should be (const char *).
4605
4606 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
4607 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
4608 However, I haven't made this a standard facility. The documentation doesn't
4609 mention it, and the Makefile doesn't support it.
4610
4611 5. The Makefile now contains an "install" target, with editable destinations at
4612 the top of the file. The pcretest program is not installed.
4613
4614 6. pgrep -V now gives the PCRE version number and date.
4615
4616 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
4617 causing the entire string to be ignored, instead of just the last character.
4618
4619 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
4620 non-matching string, it can take a very, very long time, even for strings of
4621 quite modest length, because of the nested recursion. PCRE now does better in
4622 some of these cases. It does this by remembering the last required literal
4623 character in the pattern, and pre-searching the subject to ensure it is present
4624 before running the real match. In other words, it applies a heuristic to detect
4625 some types of certain failure quickly, and in the above example, if presented
4626 with a string that has no trailing " it gives "no match" very quickly.
4627
4628 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
4629 other alternatives are tried instead.
4630
4631
4632 Version 2.06 09-Jun-99
4633 ----------------------
4634
4635 1. Change pcretest's output for amount of store used to show just the code
4636 space, because the remainder (the data block) varies in size between 32-bit and
4637 64-bit systems.
4638
4639 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
4640 start matching at. This allows lookbehinds to work when searching for multiple
4641 occurrences in a string.
4642
4643 3. Added additional options to pcretest for testing multiple occurrences:
4644
4645 /+ outputs the rest of the string that follows a match
4646 /g loops for multiple occurrences, using the new startoffset argument
4647 /G loops for multiple occurrences by passing an incremented pointer
4648
4649 4. PCRE wasn't doing the "first character" optimization for patterns starting
4650 with \b or \B, though it was doing it for other lookbehind assertions. That is,
4651 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
4652 the letter 'x'. On long subject strings, this gives a significant speed-up.
4653
4654
4655 Version 2.05 21-Apr-99
4656 ----------------------
4657
4658 1. Changed the type of magic_number from int to long int so that it works
4659 properly on 16-bit systems.
4660
4661 2. Fixed a bug which caused patterns starting with .* not to work correctly
4662 when the subject string contained newline characters. PCRE was assuming
4663 anchoring for such patterns in all cases, which is not correct because .* will
4664 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
4665 DOTALL is set at top level; otherwise it knows that patterns starting with .*
4666 must be retried after every newline in the subject.
4667
4668
4669 Version 2.04 18-Feb-99
4670 ----------------------
4671
4672 1. For parenthesized subpatterns with repeats whose minimum was zero, the
4673 computation of the store needed to hold the pattern was incorrect (too large).
4674 If such patterns were nested a few deep, this could multiply and become a real
4675 problem.
4676
4677 2. Added /M option to pcretest to show the memory requirement of a specific
4678 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
4679
4680 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
4681 compiled in such a way that the backtracking after subsequent failure was
4682 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
4683 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
4684
4685
4686 Version 2.03 02-Feb-99
4687 ----------------------
4688
4689 1. Fixed typo and small mistake in man page.
4690
4691 2. Added 4th condition (GPL supersedes if conflict) and created separate
4692 LICENCE file containing the conditions.
4693
4694 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
4695 Perl, that is the internal \ allows the delimiter to be included in the
4696 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
4697 the final delimiter, add \ to the end of the pattern (to test the error).
4698
4699 4. Added the convenience functions for extracting substrings after a successful
4700 match. Updated pcretest to make it able to test these functions.
4701
4702
4703 Version 2.02 14-Jan-99
4704 ----------------------
4705
4706 1. Initialized the working variables associated with each extraction so that
4707 their saving and restoring doesn't refer to uninitialized store.
4708
4709 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
4710 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
4711 fix the problem.
4712
4713 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
4714 calls, and wasn't printing the correct value for compiling calls. Increased the
4715 default value of LOOPREPEAT, and the number of significant figures in the
4716 times.
4717
4718 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
4719
4720 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
4721 a building problem on Windows NT with a FAT file system.
4722
4723
4724 Version 2.01 21-Oct-98
4725 ----------------------
4726
4727 1. Changed the API for pcre_compile() to allow for the provision of a pointer
4728 to character tables built by pcre_maketables() in the current locale. If NULL
4729 is passed, the default tables are used.
4730
4731
4732 Version 2.00 24-Sep-98
4733 ----------------------
4734
4735 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
4736 it any more.
4737
4738 2. Allow quantification of (?>) groups, and make it work correctly.
4739
4740 3. The first character computation wasn't working for (?>) groups.
4741
4742 4. Correct the implementation of \Z (it is permitted to match on the \n at the
4743 end of the subject) and add 5.005's \z, which really does match only at the
4744 very end of the subject.
4745
4746 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
4747
4748 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
4749 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
4750 localized options. All options to pcre_study() were also removed.
4751
4752 7. Add other new features from 5.005:
4753
4754 $(?<= positive lookbehind
4755 $(?<! negative lookbehind
4756 (?imsx-imsx) added the unsetting capability
4757 such a setting is global if at outer level; local otherwise
4758 (?imsx-imsx:) non-capturing groups with option setting
4759 (?(cond)re|re) conditional pattern matching
4760
4761 A backreference to itself in a repeated group matches the previous
4762 captured string.
4763
4764 8. General tidying up of studying (both automatic and via "study")
4765 consequential on the addition of new assertions.
4766
4767 9. As in 5.005, unlimited repeated groups that could match an empty substring
4768 are no longer faulted at compile time. Instead, the loop is forcibly broken at
4769 runtime if any iteration does actually match an empty substring.
4770
4771 10. Include the RunTest script in the distribution.
4772
4773 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
4774 discrepancies, some of which were old and were also with respect to 5.004. They
4775 have now been fixed.
4776
4777
4778 Version 1.09 28-Apr-98
4779 ----------------------
4780
4781 1. A negated single character class followed by a quantifier with a minimum
4782 value of one (e.g. [^x]{1,6} ) was not compiled correctly. This could lead to
4783 program crashes, or just wrong answers. This did not apply to negated classes
4784 containing more than one character, or to minima other than one.
4785
4786
4787 Version 1.08 27-Mar-98
4788 ----------------------
4789
4790 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
4791
4792 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
4793 latter must appear before anything that relies on it in the pattern.
4794
4795
4796 Version 1.07 16-Feb-98
4797 ----------------------
4798
4799 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
4800 repeat of a potentially empty string).
4801
4802
4803 Version 1.06 23-Jan-98
4804 ----------------------
4805
4806 1. Added Markus Oberhumer's little patches for C++.
4807
4808 2. Literal strings longer than 255 characters were broken.
4809
4810
4811 Version 1.05 23-Dec-97
4812 ----------------------
4813
4814 1. Negated character classes containing more than one character were failing if
4815 PCRE_CASELESS was set at run time.
4816
4817
4818 Version 1.04 19-Dec-97
4819 ----------------------
4820
4821 1. Corrected the man page, where some "const" qualifiers had been omitted.
4822
4823 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
4824 input syntax.
4825
4826 3. Fixed memory leak which occurred when a regex with back references was
4827 matched with an offsets vector that wasn't big enough. The temporary memory
4828 that is used in this case wasn't being freed if the match failed.
4829
4830 4. Tidied pcretest to ensure it frees memory that it gets.
4831
4832 5. Temporary memory was being obtained in the case where the passed offsets
4833 vector was exactly big enough.
4834
4835 6. Corrected definition of offsetof() from change 5 below.
4836
4837 7. I had screwed up change 6 below and broken the rules for the use of
4838 setjmp(). Now fixed.
4839
4840
4841 Version 1.03 18-Dec-97
4842 ----------------------
4843
4844 1. A erroneous regex with a missing opening parenthesis was correctly
4845 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
4846 on some systems.
4847
4848 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
4849 it was reported that one broken compiler failed on the former because "code" is
4850 also an independent variable.
4851
4852 3. The erroneous regex a[]b caused an array overrun reference.
4853
4854 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
4855 fail on data ending with that character. (It was going on too far, and checking
4856 the next character, typically a binary zero.) This was specific to the
4857 optimized code for single-character negative classes.
4858
4859 5. Added a contributed patch from the TIN world which does the following:
4860
4861 + Add an undef for memmove, in case the the system defines a macro for it.
4862
4863 + Add a definition of offsetof(), in case there isn't one. (I don't know
4864 the reason behind this - offsetof() is part of the ANSI standard - but
4865 it does no harm).
4866
4867 + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
4868 most of the places where whitespace preceded '#'. I have given up and
4869 allowed the remaining 2 cases to be at the margin.
4870
4871 + Rename some variables in pcre to eliminate shadowing. This seems very
4872 pedantic, but does no harm, of course.
4873
4874 6. Moved the call to setjmp() into its own function, to get rid of warnings
4875 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
4876
4877 7. Constructs such as \d{8,} were compiling into the equivalent of
4878 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
4879 outcome, but in this particular case used more store than had been allocated,
4880 which caused the bug to be discovered because it threw up an internal error.
4881
4882 8. The debugging code in both pcre and pcretest for outputting the compiled
4883 form of a regex was going wrong in the case of back references followed by
4884 curly-bracketed repeats.
4885
4886
4887 Version 1.02 12-Dec-97
4888 ----------------------
4889
4890 1. Typos in pcre.3 and comments in the source fixed.
4891
4892 2. Applied a contributed patch to get rid of places where it used to remove
4893 'const' from variables, and fixed some signed/unsigned and uninitialized
4894 variable warnings.
4895
4896 3. Added the "runtest" target to Makefile.
4897
4898 4. Set default compiler flag to -O2 rather than just -O.
4899
4900
4901 Version 1.01 19-Nov-97
4902 ----------------------
4903
4904 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
4905 like /([ab]*)*/, that is, for classes with more than one character in them.
4906
4907 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
4908 as /((?>a*))*/ (a PCRE_EXTRA facility).
4909
4910
4911 Version 1.00 18-Nov-97
4912 ----------------------
4913
4914 1. Added compile-time macros to support systems such as SunOS4 which don't have
4915 memmove() or strerror() but have other things that can be used instead.
4916
4917 2. Arranged that "make clean" removes the executables.
4918
4919
4920 Version 0.99 27-Oct-97
4921 ----------------------
4922
4923 1. Fixed bug in code for optimizing classes with only one character. It was
4924 initializing a 32-byte map regardless, which could cause it to run off the end
4925 of the memory it had got.
4926
4927 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
4928
4929
4930 Version 0.98 22-Oct-97
4931 ----------------------
4932
4933 1. Fixed bug in code for handling temporary memory usage when there are more
4934 back references than supplied space in the ovector. This could cause segfaults.
4935
4936
4937 Version 0.97 21-Oct-97
4938 ----------------------
4939
4940 1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
4941
4942 2. Optimized negated single characters not to use a bit map.
4943
4944 3. Brought error texts together as macro definitions; clarified some of them;
4945 fixed one that was wrong - it said "range out of order" when it meant "invalid
4946 escape sequence".
4947
4948 4. Changed some char * arguments to const char *.
4949
4950 5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
4951
4952 6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
4953 pcretest.
4954
4955
4956 Version 0.96 16-Oct-97
4957 ----------------------
4958
4959 1. Added a simple "pgrep" utility to the distribution.
4960
4961 2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
4962 unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
4963 where "ddd" means "one or more decimal digits".
4964
4965 3. Fixed serious bug. If a pattern had a back reference, but the call to
4966 pcre_exec() didn't supply a large enough ovector to record the related
4967 identifying subpattern, the match always failed. PCRE now remembers the number
4968 of the largest back reference, and gets some temporary memory in which to save
4969 the offsets during matching if necessary, in order to ensure that
4970 backreferences always work.
4971
4972 4. Increased the compatibility with Perl in a number of ways:
4973
4974 (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
4975 to request this handling. The option can be set at compile or exec time.
4976
4977 (b) $ matches before a terminating newline by default; an option
4978 PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
4979 mode). The option can be set at compile or exec time.
4980
4981 (c) The handling of \ followed by a digit other than 0 is now supposed to be
4982 the same as Perl's. If the decimal number it represents is less than 10
4983 or there aren't that many previous left capturing parentheses, an octal
4984 escape is read. Inside a character class, it's always an octal escape,
4985 even if it is a single digit.
4986
4987 (d) An escaped but undefined alphabetic character is taken as a literal,
4988 unless PCRE_EXTRA is set. Currently this just reserves the remaining
4989 escapes.
4990
4991 (e) {0} is now permitted. (The previous item is removed from the compiled
4992 pattern).
4993
4994 5. Changed all the names of code files so that the basic parts are no longer
4995 than 10 characters, and abolished the teeny "globals.c" file.
4996
4997 6. Changed the handling of character classes; they are now done with a 32-byte
4998 bit map always.
4999
5000 7. Added the -d and /D options to pcretest to make it possible to look at the
5001 internals of compilation without having to recompile pcre.
5002
5003
5004 Version 0.95 23-Sep-97
5005 ----------------------
5006
5007 1. Fixed bug in pre-pass concerning escaped "normal" characters such as \x5c or
5008 \x20 at the start of a run of normal characters. These were being treated as
5009 real characters, instead of the source characters being re-checked.
5010
5011
5012 Version 0.94 18-Sep-97
5013 ----------------------
5014
5015 1. The functions are now thread-safe, with the caveat that the global variables
5016 containing pointers to malloc() and free() or alternative functions are the
5017 same for all threads.
5018
5019 2. Get pcre_study() to generate a bitmap of initial characters for non-
5020 anchored patterns when this is possible, and use it if passed to pcre_exec().
5021
5022
5023 Version 0.93 15-Sep-97
5024 ----------------------
5025
5026 1. /(b)|(:+)/ was computing an incorrect first character.
5027
5028 2. Add pcre_study() to the API and the passing of pcre_extra to pcre_exec(),
5029 but not actually doing anything yet.
5030
5031 3. Treat "-" characters in classes that cannot be part of ranges as literals,
5032 as Perl does (e.g. [-az] or [az-]).
5033
5034 4. Set the anchored flag if a branch starts with .* or .*? because that tests
5035 all possible positions.
5036
5037 5. Split up into different modules to avoid including unneeded functions in a
5038 compiled binary. However, compile and exec are still in one module. The "study"
5039 function is split off.
5040
5041 6. The character tables are now in a separate module whose source is generated
5042 by an auxiliary program - but can then be edited by hand if required. There are
5043 now no calls to isalnum(), isspace(), isdigit(), isxdigit(), tolower() or
5044 toupper() in the code.
5045
5046 7. Turn the malloc/free funtions variables into pcre_malloc and pcre_free and
5047 make them global. Abolish the function for setting them, as the caller can now
5048 set them directly.
5049
5050
5051 Version 0.92 11-Sep-97
5052 ----------------------
5053
5054 1. A repeat with a fixed maximum and a minimum of 1 for an ordinary character
5055 (e.g. /a{1,3}/) was broken (I mis-optimized it).
5056
5057 2. Caseless matching was not working in character classes if the characters in
5058 the pattern were in upper case.
5059
5060 3. Make ranges like [W-c] work in the same way as Perl for caseless matching.
5061
5062 4. Make PCRE_ANCHORED public and accept as a compile option.
5063
5064 5. Add an options word to pcre_exec() and accept PCRE_ANCHORED and
5065 PCRE_CASELESS at run time. Add escapes \A and \I to pcretest to cause it to
5066 pass them.
5067
5068 6. Give an error if bad option bits passed at compile or run time.
5069
5070 7. Add PCRE_MULTILINE at compile and exec time, and (?m) as well. Add \M to
5071 pcretest to cause it to pass that flag.
5072
5073 8. Add pcre_info(), to get the number of identifying subpatterns, the stored
5074 options, and the first character, if set.
5075
5076 9. Recognize C+ or C{n,m} where n >= 1 as providing a fixed starting character.
5077
5078
5079 Version 0.91 10-Sep-97
5080 ----------------------
5081
5082 1. PCRE was failing to diagnose unlimited repeats of subpatterns that could
5083 match the empty string as in /(a*)*/. It was looping and ultimately crashing.
5084
5085 2. PCRE was looping on encountering an indefinitely repeated back reference to
5086 a subpattern that had matched an empty string, e.g. /(a|)\1*/. It now does what
5087 Perl does - treats the match as successful.
5088
5089 ****

Properties

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

  ViewVC Help
Powered by ViewVC 1.1.5