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

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1260 - (show annotations)
Wed Feb 27 15:41:22 2013 UTC (6 years, 9 months ago) by ph10
File size: 243399 byte(s)
Error occurred while calculating annotation data.
Add \p{Xuc} to match characters identifiable by Universal Character Names.
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
84 Version 8.32 30-November-2012
85 -----------------------------
86
87 1. Improved JIT compiler optimizations for first character search and single
88 character iterators.
89
90 2. Supporting IBM XL C compilers for PPC architectures in the JIT compiler.
91 Patch by Daniel Richard G.
92
93 3. Single character iterator optimizations in the JIT compiler.
94
95 4. Improved JIT compiler optimizations for character ranges.
96
97 5. Rename the "leave" variable names to "quit" to improve WinCE compatibility.
98 Reported by Giuseppe D'Angelo.
99
100 6. The PCRE_STARTLINE bit, indicating that a match can occur only at the start
101 of a line, was being set incorrectly in cases where .* appeared inside
102 atomic brackets at the start of a pattern, or where there was a subsequent
103 *PRUNE or *SKIP.
104
105 7. Improved instruction cache flush for POWER/PowerPC.
106 Patch by Daniel Richard G.
107
108 8. Fixed a number of issues in pcregrep, making it more compatible with GNU
109 grep:
110
111 (a) There is now no limit to the number of patterns to be matched.
112
113 (b) An error is given if a pattern is too long.
114
115 (c) Multiple uses of --exclude, --exclude-dir, --include, and --include-dir
116 are now supported.
117
118 (d) --exclude-from and --include-from (multiple use) have been added.
119
120 (e) Exclusions and inclusions now apply to all files and directories, not
121 just to those obtained from scanning a directory recursively.
122
123 (f) Multiple uses of -f and --file-list are now supported.
124
125 (g) In a Windows environment, the default for -d has been changed from
126 "read" (the GNU grep default) to "skip", because otherwise the presence
127 of a directory in the file list provokes an error.
128
129 (h) The documentation has been revised and clarified in places.
130
131 9. Improve the matching speed of capturing brackets.
132
133 10. Changed the meaning of \X so that it now matches a Unicode extended
134 grapheme cluster.
135
136 11. Patch by Daniel Richard G to the autoconf files to add a macro for sorting
137 out POSIX threads when JIT support is configured.
138
139 12. Added support for PCRE_STUDY_EXTRA_NEEDED.
140
141 13. In the POSIX wrapper regcomp() function, setting re_nsub field in the preg
142 structure could go wrong in environments where size_t is not the same size
143 as int.
144
145 14. Applied user-supplied patch to pcrecpp.cc to allow PCRE_NO_UTF8_CHECK to be
146 set.
147
148 15. The EBCDIC support had decayed; later updates to the code had included
149 explicit references to (e.g.) \x0a instead of CHAR_LF. There has been a
150 general tidy up of EBCDIC-related issues, and the documentation was also
151 not quite right. There is now a test that can be run on ASCII systems to
152 check some of the EBCDIC-related things (but is it not a full test).
153
154 16. The new PCRE_STUDY_EXTRA_NEEDED option is now used by pcregrep, resulting
155 in a small tidy to the code.
156
157 17. Fix JIT tests when UTF is disabled and both 8 and 16 bit mode are enabled.
158
159 18. If the --only-matching (-o) option in pcregrep is specified multiple
160 times, each one causes appropriate output. For example, -o1 -o2 outputs the
161 substrings matched by the 1st and 2nd capturing parentheses. A separating
162 string can be specified by --om-separator (default empty).
163
164 19. Improving the first n character searches.
165
166 20. Turn case lists for horizontal and vertical white space into macros so that
167 they are defined only once.
168
169 21. This set of changes together give more compatible Unicode case-folding
170 behaviour for characters that have more than one other case when UCP
171 support is available.
172
173 (a) The Unicode property table now has offsets into a new table of sets of
174 three or more characters that are case-equivalent. The MultiStage2.py
175 script that generates these tables (the pcre_ucd.c file) now scans
176 CaseFolding.txt instead of UnicodeData.txt for character case
177 information.
178
179 (b) The code for adding characters or ranges of characters to a character
180 class has been abstracted into a generalized function that also handles
181 case-independence. In UTF-mode with UCP support, this uses the new data
182 to handle characters with more than one other case.
183
184 (c) A bug that is fixed as a result of (b) is that codepoints less than 256
185 whose other case is greater than 256 are now correctly matched
186 caselessly. Previously, the high codepoint matched the low one, but not
187 vice versa.
188
189 (d) The processing of \h, \H, \v, and \ in character classes now makes use
190 of the new class addition function, using character lists defined as
191 macros alongside the case definitions of 20 above.
192
193 (e) Caseless back references now work with characters that have more than
194 one other case.
195
196 (f) General caseless matching of characters with more than one other case
197 is supported.
198
199 22. Unicode character properties were updated from Unicode 6.2.0
200
201 23. Improved CMake support under Windows. Patch by Daniel Richard G.
202
203 24. Add support for 32-bit character strings, and UTF-32
204
205 25. Major JIT compiler update (code refactoring and bugfixing).
206 Experimental Sparc 32 support is added.
207
208 26. Applied a modified version of Daniel Richard G's patch to create
209 pcre.h.generic and config.h.generic by "make" instead of in the
210 PrepareRelease script.
211
212 27. Added a definition for CHAR_NULL (helpful for the z/OS port), and use it in
213 pcre_compile.c when checking for a zero character.
214
215 28. Introducing a native interface for JIT. Through this interface, the compiled
216 machine code can be directly executed. The purpose of this interface is to
217 provide fast pattern matching, so several sanity checks are not performed.
218 However, feature tests are still performed. The new interface provides
219 1.4x speedup compared to the old one.
220
221 29. If pcre_exec() or pcre_dfa_exec() was called with a negative value for
222 the subject string length, the error given was PCRE_ERROR_BADOFFSET, which
223 was confusing. There is now a new error PCRE_ERROR_BADLENGTH for this case.
224
225 30. In 8-bit UTF-8 mode, pcretest failed to give an error for data codepoints
226 greater than 0x7fffffff (which cannot be represented in UTF-8, even under
227 the "old" RFC 2279). Instead, it ended up passing a negative length to
228 pcre_exec().
229
230 31. Add support for GCC's visibility feature to hide internal functions.
231
232 32. Running "pcretest -C pcre8" or "pcretest -C pcre16" gave a spurious error
233 "unknown -C option" after outputting 0 or 1.
234
235 33. There is now support for generating a code coverage report for the test
236 suite in environments where gcc is the compiler and lcov is installed. This
237 is mainly for the benefit of the developers.
238
239 34. If PCRE is built with --enable-valgrind, certain memory regions are marked
240 unaddressable using valgrind annotations, allowing valgrind to detect
241 invalid memory accesses. This is mainly for the benefit of the developers.
242
243 25. (*UTF) can now be used to start a pattern in any of the three libraries.
244
245 26. Give configure error if --enable-cpp but no C++ compiler found.
246
247
248 Version 8.31 06-July-2012
249 -------------------------
250
251 1. Fixing a wrong JIT test case and some compiler warnings.
252
253 2. Removed a bashism from the RunTest script.
254
255 3. Add a cast to pcre_exec.c to fix the warning "unary minus operator applied
256 to unsigned type, result still unsigned" that was given by an MS compiler
257 on encountering the code "-sizeof(xxx)".
258
259 4. Partial matching support is added to the JIT compiler.
260
261 5. Fixed several bugs concerned with partial matching of items that consist
262 of more than one character:
263
264 (a) /^(..)\1/ did not partially match "aba" because checking references was
265 done on an "all or nothing" basis. This also applied to repeated
266 references.
267
268 (b) \R did not give a hard partial match if \r was found at the end of the
269 subject.
270
271 (c) \X did not give a hard partial match after matching one or more
272 characters at the end of the subject.
273
274 (d) When newline was set to CRLF, a pattern such as /a$/ did not recognize
275 a partial match for the string "\r".
276
277 (e) When newline was set to CRLF, the metacharacter "." did not recognize
278 a partial match for a CR character at the end of the subject string.
279
280 6. If JIT is requested using /S++ or -s++ (instead of just /S+ or -s+) when
281 running pcretest, the text "(JIT)" added to the output whenever JIT is
282 actually used to run the match.
283
284 7. Individual JIT compile options can be set in pcretest by following -s+[+]
285 or /S+[+] with a digit between 1 and 7.
286
287 8. OP_NOT now supports any UTF character not just single-byte ones.
288
289 9. (*MARK) control verb is now supported by the JIT compiler.
290
291 10. The command "./RunTest list" lists the available tests without actually
292 running any of them. (Because I keep forgetting what they all are.)
293
294 11. Add PCRE_INFO_MAXLOOKBEHIND.
295
296 12. Applied a (slightly modified) user-supplied patch that improves performance
297 when the heap is used for recursion (compiled with --disable-stack-for-
298 recursion). Instead of malloc and free for each heap frame each time a
299 logical recursion happens, frames are retained on a chain and re-used where
300 possible. This sometimes gives as much as 30% improvement.
301
302 13. As documented, (*COMMIT) is now confined to within a recursive subpattern
303 call.
304
305 14. As documented, (*COMMIT) is now confined to within a positive assertion.
306
307 15. It is now possible to link pcretest with libedit as an alternative to
308 libreadline.
309
310 16. (*COMMIT) control verb is now supported by the JIT compiler.
311
312 17. The Unicode data tables have been updated to Unicode 6.1.0.
313
314 18. Added --file-list option to pcregrep.
315
316 19. Added binary file support to pcregrep, including the -a, --binary-files,
317 -I, and --text options.
318
319 20. The madvise function is renamed for posix_madvise for QNX compatibility
320 reasons. Fixed by Giuseppe D'Angelo.
321
322 21. Fixed a bug for backward assertions with REVERSE 0 in the JIT compiler.
323
324 22. Changed the option for creating symbolic links for 16-bit man pages from
325 -s to -sf so that re-installing does not cause issues.
326
327 23. Support PCRE_NO_START_OPTIMIZE in JIT as (*MARK) support requires it.
328
329 24. Fixed a very old bug in pcretest that caused errors with restarted DFA
330 matches in certain environments (the workspace was not being correctly
331 retained). Also added to pcre_dfa_exec() a simple plausibility check on
332 some of the workspace data at the beginning of a restart.
333
334 25. \s*\R was auto-possessifying the \s* when it should not, whereas \S*\R
335 was not doing so when it should - probably a typo introduced by SVN 528
336 (change 8.10/14).
337
338 26. When PCRE_UCP was not set, \w+\x{c4} was incorrectly auto-possessifying the
339 \w+ when the character tables indicated that \x{c4} was a word character.
340 There were several related cases, all because the tests for doing a table
341 lookup were testing for characters less than 127 instead of 255.
342
343 27. If a pattern contains capturing parentheses that are not used in a match,
344 their slots in the ovector are set to -1. For those that are higher than
345 any matched groups, this happens at the end of processing. In the case when
346 there were back references that the ovector was too small to contain
347 (causing temporary malloc'd memory to be used during matching), and the
348 highest capturing number was not used, memory off the end of the ovector
349 was incorrectly being set to -1. (It was using the size of the temporary
350 memory instead of the true size.)
351
352 28. To catch bugs like 27 using valgrind, when pcretest is asked to specify an
353 ovector size, it uses memory at the end of the block that it has got.
354
355 29. Check for an overlong MARK name and give an error at compile time. The
356 limit is 255 for the 8-bit library and 65535 for the 16-bit library.
357
358 30. JIT compiler update.
359
360 31. JIT is now supported on jailbroken iOS devices. Thanks for Ruiger
361 Rill for the patch.
362
363 32. Put spaces around SLJIT_PRINT_D in the JIT compiler. Required by CXX11.
364
365 33. Variable renamings in the PCRE-JIT compiler. No functionality change.
366
367 34. Fixed typos in pcregrep: in two places there was SUPPORT_LIBZ2 instead of
368 SUPPORT_LIBBZ2. This caused a build problem when bzip2 but not gzip (zlib)
369 was enabled.
370
371 35. Improve JIT code generation for greedy plus quantifier.
372
373 36. When /((?:a?)*)*c/ or /((?>a?)*)*c/ was matched against "aac", it set group
374 1 to "aa" instead of to an empty string. The bug affected repeated groups
375 that could potentially match an empty string.
376
377 37. Optimizing single character iterators in JIT.
378
379 38. Wide characters specified with \uxxxx in JavaScript mode are now subject to
380 the same checks as \x{...} characters in non-JavaScript mode. Specifically,
381 codepoints that are too big for the mode are faulted, and in a UTF mode,
382 disallowed codepoints are also faulted.
383
384 39. If PCRE was compiled with UTF support, in three places in the DFA
385 matcher there was code that should only have been obeyed in UTF mode, but
386 was being obeyed unconditionally. In 8-bit mode this could cause incorrect
387 processing when bytes with values greater than 127 were present. In 16-bit
388 mode the bug would be provoked by values in the range 0xfc00 to 0xdc00. In
389 both cases the values are those that cannot be the first data item in a UTF
390 character. The three items that might have provoked this were recursions,
391 possessively repeated groups, and atomic groups.
392
393 40. Ensure that libpcre is explicitly listed in the link commands for pcretest
394 and pcregrep, because some OS require shared objects to be explicitly
395 passed to ld, causing the link step to fail if they are not.
396
397 41. There were two incorrect #ifdefs in pcre_study.c, meaning that, in 16-bit
398 mode, patterns that started with \h* or \R* might be incorrectly matched.
399
400
401 Version 8.30 04-February-2012
402 -----------------------------
403
404 1. Renamed "isnumber" as "is_a_number" because in some Mac environments this
405 name is defined in ctype.h.
406
407 2. Fixed a bug in fixed-length calculation for lookbehinds that would show up
408 only in quite long subpatterns.
409
410 3. Removed the function pcre_info(), which has been obsolete and deprecated
411 since it was replaced by pcre_fullinfo() in February 2000.
412
413 4. For a non-anchored pattern, if (*SKIP) was given with a name that did not
414 match a (*MARK), and the match failed at the start of the subject, a
415 reference to memory before the start of the subject could occur. This bug
416 was introduced by fix 17 of release 8.21.
417
418 5. A reference to an unset group with zero minimum repetition was giving
419 totally wrong answers (in non-JavaScript-compatibility mode). For example,
420 /(another)?(\1?)test/ matched against "hello world test". This bug was
421 introduced in release 8.13.
422
423 6. Add support for 16-bit character strings (a large amount of work involving
424 many changes and refactorings).
425
426 7. RunGrepTest failed on msys because \r\n was replaced by whitespace when the
427 command "pattern=`printf 'xxx\r\njkl'`" was run. The pattern is now taken
428 from a file.
429
430 8. Ovector size of 2 is also supported by JIT based pcre_exec (the ovector size
431 rounding is not applied in this particular case).
432
433 9. The invalid Unicode surrogate codepoints U+D800 to U+DFFF are now rejected
434 if they appear, or are escaped, in patterns.
435
436 10. Get rid of a number of -Wunused-but-set-variable warnings.
437
438 11. The pattern /(?=(*:x))(q|)/ matches an empty string, and returns the mark
439 "x". The similar pattern /(?=(*:x))((*:y)q|)/ did not return a mark at all.
440 Oddly, Perl behaves the same way. PCRE has been fixed so that this pattern
441 also returns the mark "x". This bug applied to capturing parentheses,
442 non-capturing parentheses, and atomic parentheses. It also applied to some
443 assertions.
444
445 12. Stephen Kelly's patch to CMakeLists.txt allows it to parse the version
446 information out of configure.ac instead of relying on pcre.h.generic, which
447 is not stored in the repository.
448
449 13. Applied Dmitry V. Levin's patch for a more portable method for linking with
450 -lreadline.
451
452 14. ZH added PCRE_CONFIG_JITTARGET; added its output to pcretest -C.
453
454 15. Applied Graycode's patch to put the top-level frame on the stack rather
455 than the heap when not using the stack for recursion. This gives a
456 performance improvement in many cases when recursion is not deep.
457
458 16. Experimental code added to "pcretest -C" to output the stack frame size.
459
460
461 Version 8.21 12-Dec-2011
462 ------------------------
463
464 1. Updating the JIT compiler.
465
466 2. JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
467 are added as well.
468
469 3. Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
470 PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
471 calling _pcre_jit_exec. Some extra comments are added.
472
473 4. (*MARK) settings inside atomic groups that do not contain any capturing
474 parentheses, for example, (?>a(*:m)), were not being passed out. This bug
475 was introduced by change 18 for 8.20.
476
477 5. Supporting of \x, \U and \u in JavaScript compatibility mode based on the
478 ECMA-262 standard.
479
480 6. Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
481 erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
482 This bug was probably introduced by change 9 of 8.13.
483
484 7. While fixing 6 above, I noticed that a number of other items were being
485 incorrectly rejected as "not fixed length". This arose partly because newer
486 opcodes had not been added to the fixed-length checking code. I have (a)
487 corrected the bug and added tests for these items, and (b) arranged for an
488 error to occur if an unknown opcode is encountered while checking for fixed
489 length instead of just assuming "not fixed length". The items that were
490 rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
491 (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
492 repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
493
494 8. A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
495 being incorrectly compiled and would have given unpredicatble results.
496
497 9. A possessively repeated subpattern with minimum repeat count greater than
498 one behaved incorrectly. For example, (A){2,}+ behaved as if it was
499 (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
500 the first (A) could occur when it should not.
501
502 10. Add a cast and remove a redundant test from the code.
503
504 11. JIT should use pcre_malloc/pcre_free for allocation.
505
506 12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
507 best practice nowadays, and helps with cross-compiling. (If the exec_prefix
508 is anything other than /usr, -L is still shown).
509
510 13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
511
512 14. Perl does not support \N without a following name in a [] class; PCRE now
513 also gives an error.
514
515 15. If a forward reference was repeated with an upper limit of around 2000,
516 it caused the error "internal error: overran compiling workspace". The
517 maximum number of forward references (including repeats) was limited by the
518 internal workspace, and dependent on the LINK_SIZE. The code has been
519 rewritten so that the workspace expands (via pcre_malloc) if necessary, and
520 the default depends on LINK_SIZE. There is a new upper limit (for safety)
521 of around 200,000 forward references. While doing this, I also speeded up
522 the filling in of repeated forward references.
523
524 16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
525 incorrectly expecting the subject to contain another "a" after the start.
526
527 17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
528 in the match, the SKIP should be ignored. This was not happening; instead
529 the SKIP was being treated as NOMATCH. For patterns such as
530 /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
531 tested.
532
533 18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
534 now much more compatible with Perl, in particular in cases where the result
535 is a non-match for a non-anchored pattern. For example, if
536 /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
537 "m", where previously it did not return a name. A side effect of this
538 change is that for partial matches, the last encountered mark name is
539 returned, as for non matches. A number of tests that were previously not
540 Perl-compatible have been moved into the Perl-compatible test files. The
541 refactoring has had the pleasing side effect of removing one argument from
542 the match() function, thus reducing its stack requirements.
543
544 19. If the /S+ option was used in pcretest to study a pattern using JIT,
545 subsequent uses of /S (without +) incorrectly behaved like /S+.
546
547 21. Retrieve executable code size support for the JIT compiler and fixing
548 some warnings.
549
550 22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
551 not work when the shorter character appeared right at the end of the
552 subject string.
553
554 23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
555 systems.
556
557 24. Added PCRE_INFO_JITSIZE to pass on the value from (21) above, and also
558 output it when the /M option is used in pcretest.
559
560 25. The CheckMan script was not being included in the distribution. Also, added
561 an explicit "perl" to run Perl scripts from the PrepareRelease script
562 because this is reportedly needed in Windows.
563
564 26. If study data was being save in a file and studying had not found a set of
565 "starts with" bytes for the pattern, the data written to the file (though
566 never used) was taken from uninitialized memory and so caused valgrind to
567 complain.
568
569 27. Updated RunTest.bat as provided by Sheri Pierce.
570
571 28. Fixed a possible uninitialized memory bug in pcre_jit_compile.c.
572
573 29. Computation of memory usage for the table of capturing group names was
574 giving an unnecessarily large value.
575
576
577 Version 8.20 21-Oct-2011
578 ------------------------
579
580 1. Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
581 a POSIX class. After further experiments with Perl, which convinced me that
582 Perl has bugs and confusions, a closing square bracket is no longer allowed
583 in a POSIX name. This bug also affected patterns with classes that started
584 with full stops.
585
586 2. If a pattern such as /(a)b|ac/ is matched against "ac", there is no
587 captured substring, but while checking the failing first alternative,
588 substring 1 is temporarily captured. If the output vector supplied to
589 pcre_exec() was not big enough for this capture, the yield of the function
590 was still zero ("insufficient space for captured substrings"). This cannot
591 be totally fixed without adding another stack variable, which seems a lot
592 of expense for a edge case. However, I have improved the situation in cases
593 such as /(a)(b)x|abc/ matched against "abc", where the return code
594 indicates that fewer than the maximum number of slots in the ovector have
595 been set.
596
597 3. Related to (2) above: when there are more back references in a pattern than
598 slots in the output vector, pcre_exec() uses temporary memory during
599 matching, and copies in the captures as far as possible afterwards. It was
600 using the entire output vector, but this conflicts with the specification
601 that only 2/3 is used for passing back captured substrings. Now it uses
602 only the first 2/3, for compatibility. This is, of course, another edge
603 case.
604
605 4. Zoltan Herczeg's just-in-time compiler support has been integrated into the
606 main code base, and can be used by building with --enable-jit. When this is
607 done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
608 runtime --no-jit option is given.
609
610 5. When the number of matches in a pcre_dfa_exec() run exactly filled the
611 ovector, the return from the function was zero, implying that there were
612 other matches that did not fit. The correct "exactly full" value is now
613 returned.
614
615 6. If a subpattern that was called recursively or as a subroutine contained
616 (*PRUNE) or any other control that caused it to give a non-standard return,
617 invalid errors such as "Error -26 (nested recursion at the same subject
618 position)" or even infinite loops could occur.
619
620 7. If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
621 computing the minimum length on reaching *ACCEPT, and so ended up with the
622 wrong value of 1 rather than 0. Further investigation indicates that
623 computing a minimum subject length in the presence of *ACCEPT is difficult
624 (think back references, subroutine calls), and so I have changed the code
625 so that no minimum is registered for a pattern that contains *ACCEPT.
626
627 8. If (*THEN) was present in the first (true) branch of a conditional group,
628 it was not handled as intended. [But see 16 below.]
629
630 9. Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
631 Sheri Pierce.
632
633 10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
634 the first byte in a match must be "a".
635
636 11. Change 17 for 8.13 increased the recursion depth for patterns like
637 /a(?:.)*?a/ drastically. I've improved things by remembering whether a
638 pattern contains any instances of (*THEN). If it does not, the old
639 optimizations are restored. It would be nice to do this on a per-group
640 basis, but at the moment that is not feasible.
641
642 12. In some environments, the output of pcretest -C is CRLF terminated. This
643 broke RunTest's code that checks for the link size. A single white space
644 character after the value is now allowed for.
645
646 13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
647 For "fr", it uses the Windows-specific input and output files.
648
649 14. If (*THEN) appeared in a group that was called recursively or as a
650 subroutine, it did not work as intended. [But see next item.]
651
652 15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
653 pattern fragments (but not containing any | characters). If A and B are
654 matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
655 was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
656 D. In other words, Perl considers parentheses that do not contain any |
657 characters to be part of a surrounding alternative, whereas PCRE was
658 treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
659 differently. PCRE now behaves in the same way as Perl, except in the case
660 of subroutine/recursion calls such as (?1) which have in any case always
661 been different (but PCRE had them first :-).
662
663 16. Related to 15 above: Perl does not treat the | in a conditional group as
664 creating alternatives. Such a group is treated in the same way as an
665 ordinary group without any | characters when processing (*THEN). PCRE has
666 been changed to match Perl's behaviour.
667
668 17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
669 RunGrepTest script failed.
670
671 18. Change 22 for version 13 caused atomic groups to use more stack. This is
672 inevitable for groups that contain captures, but it can lead to a lot of
673 stack use in large patterns. The old behaviour has been restored for atomic
674 groups that do not contain any capturing parentheses.
675
676 19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
677 suppress the check for a minimum subject length at run time. (If it was
678 given to pcre_exec() or pcre_dfa_exec() it did work.)
679
680 20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
681 fail to work when decoding hex characters in data strings in EBCDIC
682 environments.
683
684 21. It appears that in at least one Mac OS environment, the isxdigit() function
685 is implemented as a macro that evaluates to its argument more than once,
686 contravening the C 90 Standard (I haven't checked a later standard). There
687 was an instance in pcretest which caused it to go wrong when processing
688 \x{...} escapes in subject strings. The has been rewritten to avoid using
689 things like p++ in the argument of isxdigit().
690
691
692 Version 8.13 16-Aug-2011
693 ------------------------
694
695 1. The Unicode data tables have been updated to Unicode 6.0.0.
696
697 2. Two minor typos in pcre_internal.h have been fixed.
698
699 3. Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
700 pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
701 in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
702
703 4. There were a number of related bugs in the code for matching backrefences
704 caselessly in UTF-8 mode when codes for the characters concerned were
705 different numbers of bytes. For example, U+023A and U+2C65 are an upper
706 and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
707 (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
708 code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
709 2-byte code at the end of the subject (it thought there wasn't enough data
710 left).
711
712 5. Comprehensive information about what went wrong is now returned by
713 pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
714 as the output vector has at least 2 elements. The offset of the start of
715 the failing character and a reason code are placed in the vector.
716
717 6. When the UTF-8 string check fails for pcre_compile(), the offset that is
718 now returned is for the first byte of the failing character, instead of the
719 last byte inspected. This is an incompatible change, but I hope it is small
720 enough not to be a problem. It makes the returned offset consistent with
721 pcre_exec() and pcre_dfa_exec().
722
723 7. pcretest now gives a text phrase as well as the error number when
724 pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
725 failure, the offset and reason code are output.
726
727 8. When \R was used with a maximizing quantifier it failed to skip backwards
728 over a \r\n pair if the subsequent match failed. Instead, it just skipped
729 back over a single character (\n). This seems wrong (because it treated the
730 two characters as a single entity when going forwards), conflicts with the
731 documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
732 behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
733 has been changed.
734
735 9. Some internal refactoring has changed the processing so that the handling
736 of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
737 time (the PCRE_DOTALL option was changed this way some time ago: version
738 7.7 change 16). This has made it possible to abolish the OP_OPT op code,
739 which was always a bit of a fudge. It also means that there is one less
740 argument for the match() function, which reduces its stack requirements
741 slightly. This change also fixes an incompatibility with Perl: the pattern
742 (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
743
744 10. More internal refactoring has drastically reduced the number of recursive
745 calls to match() for possessively repeated groups such as (abc)++ when
746 using pcre_exec().
747
748 11. While implementing 10, a number of bugs in the handling of groups were
749 discovered and fixed:
750
751 (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
752 (a|)*(?1) gave a compile-time internal error.
753 ((a|)+)+ did not notice that the outer group could match an empty string.
754 (^a|^)+ was not marked as anchored.
755 (.*a|.*)+ was not marked as matching at start or after a newline.
756
757 12. Yet more internal refactoring has removed another argument from the match()
758 function. Special calls to this function are now indicated by setting a
759 value in a variable in the "match data" data block.
760
761 13. Be more explicit in pcre_study() instead of relying on "default" for
762 opcodes that mean there is no starting character; this means that when new
763 ones are added and accidentally left out of pcre_study(), testing should
764 pick them up.
765
766 14. The -s option of pcretest has been documented for ages as being an old
767 synonym of -m (show memory usage). I have changed it to mean "force study
768 for every regex", that is, assume /S for every regex. This is similar to -i
769 and -d etc. It's slightly incompatible, but I'm hoping nobody is still
770 using it. It makes it easier to run collections of tests with and without
771 study enabled, and thereby test pcre_study() more easily. All the standard
772 tests are now run with and without -s (but some patterns can be marked as
773 "never study" - see 20 below).
774
775 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
776 restoration of the capturing data to the outer values was not happening
777 correctly.
778
779 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
780 empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
781 pattern had matched an empty string, and so incorrectly returned a no
782 match.
783
784 17. There was optimizing code for the last branch of non-capturing parentheses,
785 and also for the obeyed branch of a conditional subexpression, which used
786 tail recursion to cut down on stack usage. Unfortunately, now that there is
787 the possibility of (*THEN) occurring in these branches, tail recursion is
788 no longer possible because the return has to be checked for (*THEN). These
789 two optimizations have therefore been removed. [But see 8.20/11 above.]
790
791 18. If a pattern containing \R was studied, it was assumed that \R always
792 matched two bytes, thus causing the minimum subject length to be
793 incorrectly computed because \R can also match just one byte.
794
795 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
796 was incorrectly computed.
797
798 20. If /S is present twice on a test pattern in pcretest input, it now
799 *disables* studying, thereby overriding the use of -s on the command line
800 (see 14 above). This is necessary for one or two tests to keep the output
801 identical in both cases.
802
803 21. When (*ACCEPT) was used in an assertion that matched an empty string and
804 PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
805
806 22. When an atomic group that contained a capturing parenthesis was
807 successfully matched, but the branch in which it appeared failed, the
808 capturing was not being forgotten if a higher numbered group was later
809 captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
810 group 1 to "a", when in fact it should be unset. This applied to multi-
811 branched capturing and non-capturing groups, repeated or not, and also to
812 positive assertions (capturing in negative assertions does not happen
813 in PCRE) and also to nested atomic groups.
814
815 23. Add the ++ qualifier feature to pcretest, to show the remainder of the
816 subject after a captured substring, to make it easier to tell which of a
817 number of identical substrings has been captured.
818
819 24. The way atomic groups are processed by pcre_exec() has been changed so that
820 if they are repeated, backtracking one repetition now resets captured
821 values correctly. For example, if ((?>(a+)b)+aabab) is matched against
822 "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
823 "aaa". Previously, it would have been "a". As part of this code
824 refactoring, the way recursive calls are handled has also been changed.
825
826 25. If an assertion condition captured any substrings, they were not passed
827 back unless some other capturing happened later. For example, if
828 (?(?=(a))a) was matched against "a", no capturing was returned.
829
830 26. When studying a pattern that contained subroutine calls or assertions,
831 the code for finding the minimum length of a possible match was handling
832 direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
833 group 1 called group 2 while simultaneously a separate group 2 called group
834 1). A stack overflow occurred in this case. I have fixed this by limiting
835 the recursion depth to 10.
836
837 27. Updated RunTest.bat in the distribution to the version supplied by Tom
838 Fortmann. This supports explicit test numbers on the command line, and has
839 argument validation and error reporting.
840
841 28. An instance of \X with an unlimited repeat could fail if at any point the
842 first character it looked at was a mark character.
843
844 29. Some minor code refactoring concerning Unicode properties and scripts
845 should reduce the stack requirement of match() slightly.
846
847 30. Added the '=' option to pcretest to check the setting of unused capturing
848 slots at the end of the pattern, which are documented as being -1, but are
849 not included in the return count.
850
851 31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
852 compiled something random. Now it gives a compile-time error (as does
853 Perl).
854
855 32. A *MARK encountered during the processing of a positive assertion is now
856 recorded and passed back (compatible with Perl).
857
858 33. If --only-matching or --colour was set on a pcregrep call whose pattern
859 had alternative anchored branches, the search for a second match in a line
860 was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
861 matched the line "0102" twice. The same bug affected patterns that started
862 with a backwards assertion. For example /\b01|\b02/ also matched "0102"
863 twice.
864
865 34. Previously, PCRE did not allow quantification of assertions. However, Perl
866 does, and because of capturing effects, quantifying parenthesized
867 assertions may at times be useful. Quantifiers are now allowed for
868 parenthesized assertions.
869
870 35. A minor code tidy in pcre_compile() when checking options for \R usage.
871
872 36. \g was being checked for fancy things in a character class, when it should
873 just be a literal "g".
874
875 37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
876 appearance of a nested POSIX class supersedes an apparent external class.
877 For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
878 unescaped square brackets may also appear as part of class names. For
879 example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
880 more like Perl. (But see 8.20/1 above.)
881
882 38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
883 was because it thought it was \N{name}, which is not supported).
884
885 39. Add minix to OS list not supporting the -S option in pcretest.
886
887 40. PCRE tries to detect cases of infinite recursion at compile time, but it
888 cannot analyze patterns in sufficient detail to catch mutual recursions
889 such as ((?1))((?2)). There is now a runtime test that gives an error if a
890 subgroup is called recursively as a subpattern for a second time at the
891 same position in the subject string. In previous releases this might have
892 been caught by the recursion limit, or it might have run out of stack.
893
894 41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
895 happen only once. PCRE was, however incorrectly giving a compile time error
896 "recursive call could loop indefinitely" because it cannot analyze the
897 pattern in sufficient detail. The compile time test no longer happens when
898 PCRE is compiling a conditional subpattern, but actual runaway loops are
899 now caught at runtime (see 40 above).
900
901 42. It seems that Perl allows any characters other than a closing parenthesis
902 to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
903 has been changed to be the same.
904
905 43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
906 as not to get warnings when autogen.sh is called. Also changed
907 AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
908
909 44. To help people who use pcregrep to scan files containing exceedingly long
910 lines, the following changes have been made:
911
912 (a) The default value of the buffer size parameter has been increased from
913 8K to 20K. (The actual buffer used is three times this size.)
914
915 (b) The default can be changed by ./configure --with-pcregrep-bufsize when
916 PCRE is built.
917
918 (c) A --buffer-size=n option has been added to pcregrep, to allow the size
919 to be set at run time.
920
921 (d) Numerical values in pcregrep options can be followed by K or M, for
922 example --buffer-size=50K.
923
924 (e) If a line being scanned overflows pcregrep's buffer, an error is now
925 given and the return code is set to 2.
926
927 45. Add a pointer to the latest mark to the callout data block.
928
929 46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
930 partial match of an empty string instead of no match. This was specific to
931 the use of ".".
932
933 47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
934 complete match instead of a partial match. This bug was dependent on both
935 the PCRE_UTF8 and PCRE_DOTALL options being set.
936
937 48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
938 starting byte set, because \b was not being ignored.
939
940
941 Version 8.12 15-Jan-2011
942 ------------------------
943
944 1. Fixed some typos in the markup of the man pages, and wrote a script that
945 checks for such things as part of the documentation building process.
946
947 2. On a big-endian 64-bit system, pcregrep did not correctly process the
948 --match-limit and --recursion-limit options (added for 8.11). In
949 particular, this made one of the standard tests fail. (The integer value
950 went into the wrong half of a long int.)
951
952 3. If the --colour option was given to pcregrep with -v (invert match), it
953 did strange things, either producing crazy output, or crashing. It should,
954 of course, ignore a request for colour when reporting lines that do not
955 match.
956
957 4. Another pcregrep bug caused similar problems if --colour was specified with
958 -M (multiline) and the pattern match finished with a line ending.
959
960 5. In pcregrep, when a pattern that ended with a literal newline sequence was
961 matched in multiline mode, the following line was shown as part of the
962 match. This seems wrong, so I have changed it.
963
964 6. Another pcregrep bug in multiline mode, when --colour was specified, caused
965 the check for further matches in the same line (so they could be coloured)
966 to overrun the end of the current line. If another match was found, it was
967 incorrectly shown (and then shown again when found in the next line).
968
969 7. If pcregrep was compiled under Windows, there was a reference to the
970 function pcregrep_exit() before it was defined. I am assuming this was
971 the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
972 reported by a user. I've moved the definition above the reference.
973
974
975 Version 8.11 10-Dec-2010
976 ------------------------
977
978 1. (*THEN) was not working properly if there were untried alternatives prior
979 to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
980 backtracked to try for "b" instead of moving to the next alternative branch
981 at the same level (in this case, to look for "c"). The Perl documentation
982 is clear that when (*THEN) is backtracked onto, it goes to the "next
983 alternative in the innermost enclosing group".
984
985 2. (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
986 such as (A(*COMMIT)B(*THEN)C|D) any failure after matching A should
987 result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
988 (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
989 (*THEN).
990
991 3. If \s appeared in a character class, it removed the VT character from
992 the class, even if it had been included by some previous item, for example
993 in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
994 of \s, but is part of the POSIX "space" class.)
995
996 4. A partial match never returns an empty string (because you can always
997 match an empty string at the end of the subject); however the checking for
998 an empty string was starting at the "start of match" point. This has been
999 changed to the "earliest inspected character" point, because the returned
1000 data for a partial match starts at this character. This means that, for
1001 example, /(?<=abc)def/ gives a partial match for the subject "abc"
1002 (previously it gave "no match").
1003
1004 5. Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
1005 of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
1006 previously a full match would be given. However, setting PCRE_PARTIAL_HARD
1007 has an implication that the given string is incomplete (because a partial
1008 match is preferred over a full match). For this reason, these items now
1009 give a partial match in this situation. [Aside: previously, the one case
1010 /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
1011 match rather than a full match, which was wrong by the old rules, but is
1012 now correct.]
1013
1014 6. There was a bug in the handling of #-introduced comments, recognized when
1015 PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
1016 If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
1017 UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
1018 scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
1019 but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
1020 places in pcre_compile().
1021
1022 7. Related to (6) above, when pcre_compile() was skipping #-introduced
1023 comments when looking ahead for named forward references to subpatterns,
1024 the only newline sequence it recognized was NL. It now handles newlines
1025 according to the set newline convention.
1026
1027 8. SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
1028 former, but used strtoul(), whereas pcretest avoided strtoul() but did not
1029 cater for a lack of strerror(). These oversights have been fixed.
1030
1031 9. Added --match-limit and --recursion-limit to pcregrep.
1032
1033 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
1034
1035 11. When the -o option was used, pcregrep was setting a return code of 1, even
1036 when matches were found, and --line-buffered was not being honoured.
1037
1038 12. Added an optional parentheses number to the -o and --only-matching options
1039 of pcregrep.
1040
1041 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
1042 can match an empty string. The code to do it in pcretest and pcredemo
1043 needed fixing:
1044
1045 (a) When the newline convention was "crlf", pcretest got it wrong, skipping
1046 only one byte after an empty string match just before CRLF (this case
1047 just got forgotten; "any" and "anycrlf" were OK).
1048
1049 (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
1050 mode when an empty string match preceded an ASCII character followed by
1051 a non-ASCII character. (The code for advancing by one character rather
1052 than one byte was nonsense.)
1053
1054 (c) The pcredemo.c sample program did not have any code at all to handle
1055 the cases when CRLF is a valid newline sequence.
1056
1057 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
1058 as a starting offset was within the subject string. There is now a new
1059 error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
1060 negative or greater than the length of the string. In order to test this,
1061 pcretest is extended to allow the setting of negative starting offsets.
1062
1063 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
1064 starting offset points to the beginning of a UTF-8 character was
1065 unnecessarily clumsy. I tidied it up.
1066
1067 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
1068 bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
1069
1070 17. Nobody had reported that the --include_dir option, which was added in
1071 release 7.7 should have been called --include-dir (hyphen, not underscore)
1072 for compatibility with GNU grep. I have changed it to --include-dir, but
1073 left --include_dir as an undocumented synonym, and the same for
1074 --exclude-dir, though that is not available in GNU grep, at least as of
1075 release 2.5.4.
1076
1077 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
1078 characters from a string of bytes) have been redefined so as not to use
1079 loops, in order to improve performance in some environments. At the same
1080 time, I abstracted some of the common code into auxiliary macros to save
1081 repetition (this should not affect the compiled code).
1082
1083 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
1084 compile-time error is now given if \c is not followed by an ASCII
1085 character, that is, a byte less than 128. (In EBCDIC mode, the code is
1086 different, and any byte value is allowed.)
1087
1088 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
1089 START_OPTIMIZE option, which is now allowed at compile time - but just
1090 passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
1091 to pcregrep and other applications that have no direct access to PCRE
1092 options. The new /Y option in pcretest sets this option when calling
1093 pcre_compile().
1094
1095 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
1096 back references. Groups containing recursive back references were forced to
1097 be atomic by that change, but in the case of named groups, the amount of
1098 memory required was incorrectly computed, leading to "Failed: internal
1099 error: code overflow". This has been fixed.
1100
1101 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
1102 pcretest.c, to avoid build problems in some Borland environments.
1103
1104
1105 Version 8.10 25-Jun-2010
1106 ------------------------
1107
1108 1. Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
1109 THEN.
1110
1111 2. (*ACCEPT) was not working when inside an atomic group.
1112
1113 3. Inside a character class, \B is treated as a literal by default, but
1114 faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
1115 causes the error). The code is unchanged, but I tidied the documentation.
1116
1117 4. Inside a character class, PCRE always treated \R and \X as literals,
1118 whereas Perl faults them if its -w option is set. I have changed PCRE so
1119 that it faults them when PCRE_EXTRA is set.
1120
1121 5. Added support for \N, which always matches any character other than
1122 newline. (It is the same as "." when PCRE_DOTALL is not set.)
1123
1124 6. When compiling pcregrep with newer versions of gcc which may have
1125 FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
1126 declared with attribute warn_unused_result" were given. Just casting the
1127 result to (void) does not stop the warnings; a more elaborate fudge is
1128 needed. I've used a macro to implement this.
1129
1130 7. Minor change to pcretest.c to avoid a compiler warning.
1131
1132 8. Added four artifical Unicode properties to help with an option to make
1133 \s etc use properties (see next item). The new properties are: Xan
1134 (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
1135
1136 9. Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
1137 use Unicode properties. (*UCP) at the start of a pattern can be used to set
1138 this option. Modified pcretest to add /W to test this facility. Added
1139 REG_UCP to make it available via the POSIX interface.
1140
1141 10. Added --line-buffered to pcregrep.
1142
1143 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
1144 studied, and the match started with a letter with a code point greater than
1145 127 whose first byte was different to the first byte of the other case of
1146 the letter, the other case of this starting letter was not recognized
1147 (#976).
1148
1149 12. If a pattern that was studied started with a repeated Unicode property
1150 test, for example, \p{Nd}+, there was the theoretical possibility of
1151 setting up an incorrect bitmap of starting bytes, but fortunately it could
1152 not have actually happened in practice until change 8 above was made (it
1153 added property types that matched character-matching opcodes).
1154
1155 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
1156 possible starting bytes for non-anchored patterns.
1157
1158 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
1159 \R, and also a number of cases that involve Unicode properties, both
1160 explicit and implicit when PCRE_UCP is set.
1161
1162 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
1163 input, it could crash or give wrong results if characters with values
1164 greater than 0xc0 were present in the subject string. (Detail: it assumed
1165 UTF-8 input when processing these items.)
1166
1167 16. Added a lot of (int) casts to avoid compiler warnings in systems where
1168 size_t is 64-bit (#991).
1169
1170 17. Added a check for running out of memory when PCRE is compiled with
1171 --disable-stack-for-recursion (#990).
1172
1173 18. If the last data line in a file for pcretest does not have a newline on
1174 the end, a newline was missing in the output.
1175
1176 19. The default pcre_chartables.c file recognizes only ASCII characters (values
1177 less than 128) in its various bitmaps. However, there is a facility for
1178 generating tables according to the current locale when PCRE is compiled. It
1179 turns out that in some environments, 0x85 and 0xa0, which are Unicode space
1180 characters, are recognized by isspace() and therefore were getting set in
1181 these tables, and indeed these tables seem to approximate to ISO 8859. This
1182 caused a problem in UTF-8 mode when pcre_study() was used to create a list
1183 of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
1184 which of course cannot start UTF-8 characters. I have changed the code so
1185 that only real ASCII characters (less than 128) and the correct starting
1186 bytes for UTF-8 encodings are set for characters greater than 127 when in
1187 UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
1188 altogether.)
1189
1190 20. Added the /T option to pcretest so as to be able to run tests with non-
1191 standard character tables, thus making it possible to include the tests
1192 used for 19 above in the standard set of tests.
1193
1194 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
1195 reference to a subpattern the other side of a comment that contains an
1196 opening parenthesis caused either an internal compiling error, or a
1197 reference to the wrong subpattern.
1198
1199
1200 Version 8.02 19-Mar-2010
1201 ------------------------
1202
1203 1. The Unicode data tables have been updated to Unicode 5.2.0.
1204
1205 2. Added the option --libs-cpp to pcre-config, but only when C++ support is
1206 configured.
1207
1208 3. Updated the licensing terms in the pcregexp.pas file, as agreed with the
1209 original author of that file, following a query about its status.
1210
1211 4. On systems that do not have stdint.h (e.g. Solaris), check for and include
1212 inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
1213
1214 5. A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
1215 quantifier applied to a forward-referencing subroutine call, could compile
1216 incorrect code or give the error "internal error: previously-checked
1217 referenced subpattern not found".
1218
1219 6. Both MS Visual Studio and Symbian OS have problems with initializing
1220 variables to point to external functions. For these systems, therefore,
1221 pcre_malloc etc. are now initialized to local functions that call the
1222 relevant global functions.
1223
1224 7. There were two entries missing in the vectors called coptable and poptable
1225 in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
1226 I've fixed the data, and added a kludgy way of testing at compile time that
1227 the lengths are correct (equal to the number of opcodes).
1228
1229 8. Following on from 7, I added a similar kludge to check the length of the
1230 eint vector in pcreposix.c.
1231
1232 9. Error texts for pcre_compile() are held as one long string to avoid too
1233 much relocation at load time. To find a text, the string is searched,
1234 counting zeros. There was no check for running off the end of the string,
1235 which could happen if a new error number was added without updating the
1236 string.
1237
1238 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
1239
1240 11. \K was not working if it appeared in an atomic group or in a group that
1241 was called as a "subroutine", or in an assertion. Perl 5.11 documents that
1242 \K is "not well defined" if used in an assertion. PCRE now accepts it if
1243 the assertion is positive, but not if it is negative.
1244
1245 12. Change 11 fortuitously reduced the size of the stack frame used in the
1246 "match()" function of pcre_exec.c by one pointer. Forthcoming
1247 implementation of support for (*MARK) will need an extra pointer on the
1248 stack; I have reserved it now, so that the stack frame size does not
1249 decrease.
1250
1251 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
1252 item in branch that calls a recursion is a subroutine call - as in the
1253 second branch in the above example - was incorrectly given the compile-
1254 time error "recursive call could loop indefinitely" because pcre_compile()
1255 was not correctly checking the subroutine for matching a non-empty string.
1256
1257 14. The checks for overrunning compiling workspace could trigger after an
1258 overrun had occurred. This is a "should never occur" error, but it can be
1259 triggered by pathological patterns such as hundreds of nested parentheses.
1260 The checks now trigger 100 bytes before the end of the workspace.
1261
1262 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
1263
1264
1265 Version 8.01 19-Jan-2010
1266 ------------------------
1267
1268 1. If a pattern contained a conditional subpattern with only one branch (in
1269 particular, this includes all (*DEFINE) patterns), a call to pcre_study()
1270 computed the wrong minimum data length (which is of course zero for such
1271 subpatterns). This could cause incorrect "no match" results.
1272
1273 2. For patterns such as (?i)a(?-i)b|c where an option setting at the start of
1274 the pattern is reset in the first branch, pcre_compile() failed with
1275 "internal error: code overflow at offset...". This happened only when
1276 the reset was to the original external option setting. (An optimization
1277 abstracts leading options settings into an external setting, which was the
1278 cause of this.)
1279
1280 3. A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
1281 of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
1282 assertion pattern did not match (meaning that the assertion was true), it
1283 was incorrectly treated as false if the SKIP had been reached during the
1284 matching. This also applied to assertions used as conditions.
1285
1286 4. If an item that is not supported by pcre_dfa_exec() was encountered in an
1287 assertion subpattern, including such a pattern used as a condition,
1288 unpredictable results occurred, instead of the error return
1289 PCRE_ERROR_DFA_UITEM.
1290
1291 5. The C++ GlobalReplace function was not working like Perl for the special
1292 situation when an empty string is matched. It now does the fancy magic
1293 stuff that is necessary.
1294
1295 6. In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
1296 removed. (These were left over from very, very early versions of PCRE.)
1297
1298 7. Some cosmetic changes to the code to make life easier when compiling it
1299 as part of something else:
1300
1301 (a) Change DEBUG to PCRE_DEBUG.
1302
1303 (b) In pcre_compile(), rename the member of the "branch_chain" structure
1304 called "current" as "current_branch", to prevent a collision with the
1305 Linux macro when compiled as a kernel module.
1306
1307 (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
1308 prevent a collision with the Linux macro when compiled as a kernel
1309 module.
1310
1311 8. In pcre_compile() there are some checks for integer overflows that used to
1312 cast potentially large values to (double). This has been changed to that
1313 when building, a check for int64_t is made, and if it is found, it is used
1314 instead, thus avoiding the use of floating point arithmetic. (There is no
1315 other use of FP in PCRE.) If int64_t is not found, the fallback is to
1316 double.
1317
1318 9. Added two casts to avoid signed/unsigned warnings from VS Studio Express
1319 2005 (difference between two addresses compared to an unsigned value).
1320
1321 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
1322 custom one, because of the following reported problem in Windows:
1323
1324 - libbz2 uses the Pascal calling convention (WINAPI) for the functions
1325 under Win32.
1326 - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
1327 therefore missing the function definition.
1328 - The compiler thus generates a "C" signature for the test function.
1329 - The linker fails to find the "C" function.
1330 - PCRE fails to configure if asked to do so against libbz2.
1331
1332 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
1333 messages were output:
1334
1335 Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
1336 rerunning libtoolize, to keep the correct libtool macros in-tree.
1337 Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
1338
1339 I have done both of these things.
1340
1341 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
1342 most of the time, it *can* run out if it is given a pattern that contains a
1343 runaway infinite recursion. I updated the discussion in the pcrestack man
1344 page.
1345
1346 13. Now that we have gone to the x.xx style of version numbers, the minor
1347 version may start with zero. Using 08 or 09 is a bad idea because users
1348 might check the value of PCRE_MINOR in their code, and 08 or 09 may be
1349 interpreted as invalid octal numbers. I've updated the previous comment in
1350 configure.ac, and also added a check that gives an error if 08 or 09 are
1351 used.
1352
1353 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
1354 causing partial matching to fail when the end of the subject matched \W
1355 in a UTF-8 pattern where \W was quantified with a minimum of 3.
1356
1357 15. There were some discrepancies between the declarations in pcre_internal.h
1358 of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
1359 their definitions. The declarations used "const uschar *" and the
1360 definitions used USPTR. Even though USPTR is normally defined as "const
1361 unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
1362 reported that: "This difference in casting confuses some C++ compilers, for
1363 example, SunCC recognizes above declarations as different functions and
1364 generates broken code for hbpcre." I have changed the declarations to use
1365 USPTR.
1366
1367 16. GNU libtool is named differently on some systems. The autogen.sh script now
1368 tries several variants such as glibtoolize (MacOSX) and libtoolize1x
1369 (FreeBSD).
1370
1371 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
1372 (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
1373 comment: "Figure out how to create a longlong from a string: strtoll and
1374 equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
1375 instance, but it only takes 2 args instead of 3!"
1376
1377 18. A subtle bug concerned with back references has been fixed by a change of
1378 specification, with a corresponding code fix. A pattern such as
1379 ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
1380 refers, was giving matches when it shouldn't. For example, xa=xaaa would
1381 match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
1382 same bug. Such groups have to be quantified to be useful, or contained
1383 inside another quantified group. (If there's no repetition, the reference
1384 can never match.) The problem arises because, having left the group and
1385 moved on to the rest of the pattern, a later failure that backtracks into
1386 the group uses the captured value from the final iteration of the group
1387 rather than the correct earlier one. I have fixed this in PCRE by forcing
1388 any group that contains a reference to itself to be an atomic group; that
1389 is, there cannot be any backtracking into it once it has completed. This is
1390 similar to recursive and subroutine calls.
1391
1392
1393 Version 8.00 19-Oct-09
1394 ----------------------
1395
1396 1. The table for translating pcre_compile() error codes into POSIX error codes
1397 was out-of-date, and there was no check on the pcre_compile() error code
1398 being within the table. This could lead to an OK return being given in
1399 error.
1400
1401 2. Changed the call to open a subject file in pcregrep from fopen(pathname,
1402 "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
1403 in a Windows environment.
1404
1405 3. The pcregrep --count option prints the count for each file even when it is
1406 zero, as does GNU grep. However, pcregrep was also printing all files when
1407 --files-with-matches was added. Now, when both options are given, it prints
1408 counts only for those files that have at least one match. (GNU grep just
1409 prints the file name in this circumstance, but including the count seems
1410 more useful - otherwise, why use --count?) Also ensured that the
1411 combination -clh just lists non-zero counts, with no names.
1412
1413 4. The long form of the pcregrep -F option was incorrectly implemented as
1414 --fixed_strings instead of --fixed-strings. This is an incompatible change,
1415 but it seems right to fix it, and I didn't think it was worth preserving
1416 the old behaviour.
1417
1418 5. The command line items --regex=pattern and --regexp=pattern were not
1419 recognized by pcregrep, which required --regex pattern or --regexp pattern
1420 (with a space rather than an '='). The man page documented the '=' forms,
1421 which are compatible with GNU grep; these now work.
1422
1423 6. No libpcreposix.pc file was created for pkg-config; there was just
1424 libpcre.pc and libpcrecpp.pc. The omission has been rectified.
1425
1426 7. Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
1427 when UCP support is not needed, by modifying the Python script that
1428 generates it from Unicode data files. This should not matter if the module
1429 is correctly used as a library, but I received one complaint about 50K of
1430 unwanted data. My guess is that the person linked everything into his
1431 program rather than using a library. Anyway, it does no harm.
1432
1433 8. A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
1434 was a minimum greater than 1 for a wide character in a possessive
1435 repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
1436 which had an unlimited repeat of a nested, fixed maximum repeat of a wide
1437 character. Chaos in the form of incorrect output or a compiling loop could
1438 result.
1439
1440 9. The restrictions on what a pattern can contain when partial matching is
1441 requested for pcre_exec() have been removed. All patterns can now be
1442 partially matched by this function. In addition, if there are at least two
1443 slots in the offset vector, the offset of the earliest inspected character
1444 for the match and the offset of the end of the subject are set in them when
1445 PCRE_ERROR_PARTIAL is returned.
1446
1447 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
1448 synonymous with PCRE_PARTIAL, for backwards compatibility, and
1449 PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
1450 and may be more useful for multi-segment matching.
1451
1452 11. Partial matching with pcre_exec() is now more intuitive. A partial match
1453 used to be given if ever the end of the subject was reached; now it is
1454 given only if matching could not proceed because another character was
1455 needed. This makes a difference in some odd cases such as Z(*FAIL) with the
1456 string "Z", which now yields "no match" instead of "partial match". In the
1457 case of pcre_dfa_exec(), "no match" is given if every matching path for the
1458 final character ended with (*FAIL).
1459
1460 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
1461 if the pattern had a "must contain" character that was already found in the
1462 earlier partial match, unless partial matching was again requested. For
1463 example, with the pattern /dog.(body)?/, the "must contain" character is
1464 "g". If the first part-match was for the string "dog", restarting with
1465 "sbody" failed. This bug has been fixed.
1466
1467 13. The string returned by pcre_dfa_exec() after a partial match has been
1468 changed so that it starts at the first inspected character rather than the
1469 first character of the match. This makes a difference only if the pattern
1470 starts with a lookbehind assertion or \b or \B (\K is not supported by
1471 pcre_dfa_exec()). It's an incompatible change, but it makes the two
1472 matching functions compatible, and I think it's the right thing to do.
1473
1474 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
1475 so that the demonstration program is easily available in environments where
1476 PCRE has not been installed from source.
1477
1478 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
1479 libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
1480 library.
1481
1482 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
1483 It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
1484 is not the first non-POSIX option to be added. Clearly some people find
1485 these options useful.
1486
1487 17. If a caller to the POSIX matching function regexec() passes a non-zero
1488 value for nmatch with a NULL value for pmatch, the value of
1489 nmatch is forced to zero.
1490
1491 18. RunGrepTest did not have a test for the availability of the -u option of
1492 the diff command, as RunTest does. It now checks in the same way as
1493 RunTest, and also checks for the -b option.
1494
1495 19. If an odd number of negated classes containing just a single character
1496 interposed, within parentheses, between a forward reference to a named
1497 subpattern and the definition of the subpattern, compilation crashed with
1498 an internal error, complaining that it could not find the referenced
1499 subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
1500 [The bug was that it was starting one character too far in when skipping
1501 over the character class, thus treating the ] as data rather than
1502 terminating the class. This meant it could skip too much.]
1503
1504 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
1505 /g option in pcretest when the pattern contains \K, which makes it possible
1506 to have an empty string match not at the start, even when the pattern is
1507 anchored. Updated pcretest and pcredemo to use this option.
1508
1509 21. If the maximum number of capturing subpatterns in a recursion was greater
1510 than the maximum at the outer level, the higher number was returned, but
1511 with unset values at the outer level. The correct (outer level) value is
1512 now given.
1513
1514 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
1515 PCRE did not set those parentheses (unlike Perl). I have now found a way to
1516 make it do so. The string so far is captured, making this feature
1517 compatible with Perl.
1518
1519 23. The tests have been re-organized, adding tests 11 and 12, to make it
1520 possible to check the Perl 5.10 features against Perl 5.10.
1521
1522 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
1523 pattern matches a fixed length string. PCRE did not allow this; now it
1524 does. Neither allows recursion.
1525
1526 25. I finally figured out how to implement a request to provide the minimum
1527 length of subject string that was needed in order to match a given pattern.
1528 (It was back references and recursion that I had previously got hung up
1529 on.) This code has now been added to pcre_study(); it finds a lower bound
1530 to the length of subject needed. It is not necessarily the greatest lower
1531 bound, but using it to avoid searching strings that are too short does give
1532 some useful speed-ups. The value is available to calling programs via
1533 pcre_fullinfo().
1534
1535 26. While implementing 25, I discovered to my embarrassment that pcretest had
1536 not been passing the result of pcre_study() to pcre_dfa_exec(), so the
1537 study optimizations had never been tested with that matching function.
1538 Oops. What is worse, even when it was passed study data, there was a bug in
1539 pcre_dfa_exec() that meant it never actually used it. Double oops. There
1540 were also very few tests of studied patterns with pcre_dfa_exec().
1541
1542 27. If (?| is used to create subpatterns with duplicate numbers, they are now
1543 allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
1544 on the other side of the coin, they are no longer allowed to have different
1545 names, because these cannot be distinguished in PCRE, and this has caused
1546 confusion. (This is a difference from Perl.)
1547
1548 28. When duplicate subpattern names are present (necessarily with different
1549 numbers, as required by 27 above), and a test is made by name in a
1550 conditional pattern, either for a subpattern having been matched, or for
1551 recursion in such a pattern, all the associated numbered subpatterns are
1552 tested, and the overall condition is true if the condition is true for any
1553 one of them. This is the way Perl works, and is also more like the way
1554 testing by number works.
1555
1556
1557 Version 7.9 11-Apr-09
1558 ---------------------
1559
1560 1. When building with support for bzlib/zlib (pcregrep) and/or readline
1561 (pcretest), all targets were linked against these libraries. This included
1562 libpcre, libpcreposix, and libpcrecpp, even though they do not use these
1563 libraries. This caused unwanted dependencies to be created. This problem
1564 has been fixed, and now only pcregrep is linked with bzlib/zlib and only
1565 pcretest is linked with readline.
1566
1567 2. The "typedef int BOOL" in pcre_internal.h that was included inside the
1568 "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
1569 moved outside it again, because FALSE and TRUE are already defined in AIX,
1570 but BOOL is not.
1571
1572 3. The pcre_config() function was treating the PCRE_MATCH_LIMIT and
1573 PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
1574
1575 4. The pcregrep documentation said spaces were inserted as well as colons (or
1576 hyphens) following file names and line numbers when outputting matching
1577 lines. This is not true; no spaces are inserted. I have also clarified the
1578 wording for the --colour (or --color) option.
1579
1580 5. In pcregrep, when --colour was used with -o, the list of matching strings
1581 was not coloured; this is different to GNU grep, so I have changed it to be
1582 the same.
1583
1584 6. When --colo(u)r was used in pcregrep, only the first matching substring in
1585 each matching line was coloured. Now it goes on to look for further matches
1586 of any of the test patterns, which is the same behaviour as GNU grep.
1587
1588 7. A pattern that could match an empty string could cause pcregrep to loop; it
1589 doesn't make sense to accept an empty string match in pcregrep, so I have
1590 locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
1591 seems to be how GNU grep behaves.
1592
1593 8. The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
1594 start or after a newline", because the conditional assertion was not being
1595 correctly handled. The rule now is that both the assertion and what follows
1596 in the first alternative must satisfy the test.
1597
1598 9. If auto-callout was enabled in a pattern with a conditional group whose
1599 condition was an assertion, PCRE could crash during matching, both with
1600 pcre_exec() and pcre_dfa_exec().
1601
1602 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
1603 used for matching.
1604
1605 11. Unicode property support in character classes was not working for
1606 characters (bytes) greater than 127 when not in UTF-8 mode.
1607
1608 12. Added the -M command line option to pcretest.
1609
1610 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
1611
1612 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
1613
1614 16. Added comments and documentation about mis-use of no_arg in the C++
1615 wrapper.
1616
1617 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
1618 from Martin Jerabek that uses macro names for all relevant character and
1619 string constants.
1620
1621 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
1622 SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
1623 SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
1624 these, but not everybody uses configure.
1625
1626 19. A conditional group that had only one branch was not being correctly
1627 recognized as an item that could match an empty string. This meant that an
1628 enclosing group might also not be so recognized, causing infinite looping
1629 (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
1630 with the subject "ab", where knowledge that the repeated group can match
1631 nothing is needed in order to break the loop.
1632
1633 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
1634 exec(), but without supplying a callout function, matching went wrong.
1635
1636 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
1637 leak if the size of the offset vector was greater than 30. When the vector
1638 is smaller, the saved offsets during recursion go onto a local stack
1639 vector, but for larger vectors malloc() is used. It was failing to free
1640 when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
1641 error, in fact).
1642
1643 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
1644 heapframe that is used only when UTF-8 support is enabled. This caused no
1645 problem, but was untidy.
1646
1647 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
1648 CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
1649 included within another project.
1650
1651 24. Steven Van Ingelgem's patches to add more options to the CMake support,
1652 slightly modified by me:
1653
1654 (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
1655 not building pcregrep.
1656
1657 (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
1658 if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
1659
1660 25. Forward references, both numeric and by name, in patterns that made use of
1661 duplicate group numbers, could behave incorrectly or give incorrect errors,
1662 because when scanning forward to find the reference group, PCRE was not
1663 taking into account the duplicate group numbers. A pattern such as
1664 ^X(?3)(a)(?|(b)|(q))(Y) is an example.
1665
1666 26. Changed a few more instances of "const unsigned char *" to USPTR, making
1667 the feature of a custom pointer more persuasive (as requested by a user).
1668
1669 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
1670 pcretest.c, inside #ifndefs, because it seems they are sometimes already
1671 pre-defined.
1672
1673 28. Added support for (*UTF8) at the start of a pattern.
1674
1675 29. Arrange for flags added by the "release type" setting in CMake to be shown
1676 in the configuration summary.
1677
1678
1679 Version 7.8 05-Sep-08
1680 ---------------------
1681
1682 1. Replaced UCP searching code with optimized version as implemented for Ad
1683 Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
1684 stage table and inline lookup instead of a function, giving speed ups of 2
1685 to 5 times on some simple patterns that I tested. Permission was given to
1686 distribute the MultiStage2.py script that generates the tables (it's not in
1687 the tarball, but is in the Subversion repository).
1688
1689 2. Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
1690 scripts.
1691
1692 3. Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
1693 a group with a zero qualifier. The result of the study could be incorrect,
1694 or the function might crash, depending on the pattern.
1695
1696 4. Caseless matching was not working for non-ASCII characters in back
1697 references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
1698 It now works when Unicode Property Support is available.
1699
1700 5. In pcretest, an escape such as \x{de} in the data was always generating
1701 a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
1702 non-UTF-8 mode. If the value is greater than 255, it gives a warning about
1703 truncation.
1704
1705 6. Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
1706
1707 7. Added two (int) casts to pcregrep when printing the difference of two
1708 pointers, in case they are 64-bit values.
1709
1710 8. Added comments about Mac OS X stack usage to the pcrestack man page and to
1711 test 2 if it fails.
1712
1713 9. Added PCRE_CALL_CONVENTION just before the names of all exported functions,
1714 and a #define of that name to empty if it is not externally set. This is to
1715 allow users of MSVC to set it if necessary.
1716
1717 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
1718 the convenience functions in the pcre_get.c source file.
1719
1720 11. An option change at the start of a pattern that had top-level alternatives
1721 could cause overwriting and/or a crash. This command provoked a crash in
1722 some environments:
1723
1724 printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
1725
1726 This potential security problem was recorded as CVE-2008-2371.
1727
1728 12. For a pattern where the match had to start at the beginning or immediately
1729 after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
1730 pcre_dfa_exec() could read past the end of the passed subject if there was
1731 no match. To help with detecting such bugs (e.g. with valgrind), I modified
1732 pcretest so that it places the subject at the end of its malloc-ed buffer.
1733
1734 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
1735 exec() might read past the end of the data buffer in UTF-8 mode.
1736
1737 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
1738 the data contained the byte 0x85 as part of a UTF-8 character within its
1739 first line. This applied both to normal and DFA matching.
1740
1741 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
1742 /^[^d]*?$/8 failed to match "abc".
1743
1744 16. Added a missing copyright notice to pcrecpp_internal.h.
1745
1746 17. Make it more clear in the documentation that values returned from
1747 pcre_exec() in ovector are byte offsets, not character counts.
1748
1749 18. Tidied a few places to stop certain compilers from issuing warnings.
1750
1751 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
1752 supplied by Stefan Weber. I made a further small update for 7.8 because
1753 there is a change of source arrangements: the pcre_searchfuncs.c module is
1754 replaced by pcre_ucd.c.
1755
1756
1757 Version 7.7 07-May-08
1758 ---------------------
1759
1760 1. Applied Craig's patch to sort out a long long problem: "If we can't convert
1761 a string to a long long, pretend we don't even have a long long." This is
1762 done by checking for the strtoq, strtoll, and _strtoi64 functions.
1763
1764 2. Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
1765 pre-7.6 versions, which defined a global no_arg variable instead of putting
1766 it in the RE class. (See also #8 below.)
1767
1768 3. Remove a line of dead code, identified by coverity and reported by Nuno
1769 Lopes.
1770
1771 4. Fixed two related pcregrep bugs involving -r with --include or --exclude:
1772
1773 (1) The include/exclude patterns were being applied to the whole pathnames
1774 of files, instead of just to the final components.
1775
1776 (2) If there was more than one level of directory, the subdirectories were
1777 skipped unless they satisfied the include/exclude conditions. This is
1778 inconsistent with GNU grep (and could even be seen as contrary to the
1779 pcregrep specification - which I improved to make it absolutely clear).
1780 The action now is always to scan all levels of directory, and just
1781 apply the include/exclude patterns to regular files.
1782
1783 5. Added the --include_dir and --exclude_dir patterns to pcregrep, and used
1784 --exclude_dir in the tests to avoid scanning .svn directories.
1785
1786 6. Applied Craig's patch to the QuoteMeta function so that it escapes the
1787 NUL character as backslash + 0 rather than backslash + NUL, because PCRE
1788 doesn't support NULs in patterns.
1789
1790 7. Added some missing "const"s to declarations of static tables in
1791 pcre_compile.c and pcre_dfa_exec.c.
1792
1793 8. Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
1794 caused by fix #2 above. (Subsequently also a second patch to fix the
1795 first patch. And a third patch - this was a messy problem.)
1796
1797 9. Applied Craig's patch to remove the use of push_back().
1798
1799 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
1800 matching function regexec().
1801
1802 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
1803 which, however, unlike Perl's \g{...}, are subroutine calls, not back
1804 references. PCRE supports relative numbers with this syntax (I don't think
1805 Oniguruma does).
1806
1807 12. Previously, a group with a zero repeat such as (...){0} was completely
1808 omitted from the compiled regex. However, this means that if the group
1809 was called as a subroutine from elsewhere in the pattern, things went wrong
1810 (an internal error was given). Such groups are now left in the compiled
1811 pattern, with a new opcode that causes them to be skipped at execution
1812 time.
1813
1814 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
1815 to the way PCRE behaves:
1816
1817 (a) A lone ] character is dis-allowed (Perl treats it as data).
1818
1819 (b) A back reference to an unmatched subpattern matches an empty string
1820 (Perl fails the current match path).
1821
1822 (c) A data ] in a character class must be notated as \] because if the
1823 first data character in a class is ], it defines an empty class. (In
1824 Perl it is not possible to have an empty class.) The empty class []
1825 never matches; it forces failure and is equivalent to (*FAIL) or (?!).
1826 The negative empty class [^] matches any one character, independently
1827 of the DOTALL setting.
1828
1829 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
1830 non-existent subpattern following a character class starting with ']' and
1831 containing () gave an internal compiling error instead of "reference to
1832 non-existent subpattern". Fortunately, when the pattern did exist, the
1833 compiled code was correct. (When scanning forwards to check for the
1834 existencd of the subpattern, it was treating the data ']' as terminating
1835 the class, so got the count wrong. When actually compiling, the reference
1836 was subsequently set up correctly.)
1837
1838 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
1839 it was being rejected as not supported by pcre_dfa_exec(), even though
1840 other assertions are supported. I have made pcre_dfa_exec() support
1841 (*FAIL).
1842
1843 16. The implementation of 13c above involved the invention of a new opcode,
1844 OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
1845 cannot be changed at match time, I realized I could make a small
1846 improvement to matching performance by compiling OP_ALLANY instead of
1847 OP_ANY for "." when DOTALL was set, and then removing the runtime tests
1848 on the OP_ANY path.
1849
1850 17. Compiling pcretest on Windows with readline support failed without the
1851 following two fixes: (1) Make the unistd.h include conditional on
1852 HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
1853
1854 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
1855 ncurses library to be included for pcretest when ReadLine support is
1856 requested, but also to allow for it to be overridden. This patch came from
1857 Daniel Bergström.
1858
1859 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
1860 as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
1861 any errors with the current Unicode tables. Thanks to Peter Kankowski for
1862 spotting this.
1863
1864
1865 Version 7.6 28-Jan-08
1866 ---------------------
1867
1868 1. A character class containing a very large number of characters with
1869 codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
1870 overflow.
1871
1872 2. Patch to cut out the "long long" test in pcrecpp_unittest when
1873 HAVE_LONG_LONG is not defined.
1874
1875 3. Applied Christian Ehrlicher's patch to update the CMake build files to
1876 bring them up to date and include new features. This patch includes:
1877
1878 - Fixed PH's badly added libz and libbz2 support.
1879 - Fixed a problem with static linking.
1880 - Added pcredemo. [But later removed - see 7 below.]
1881 - Fixed dftables problem and added an option.
1882 - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
1883 HAVE_LONG_LONG.
1884 - Added readline support for pcretest.
1885 - Added an listing of the option settings after cmake has run.
1886
1887 4. A user submitted a patch to Makefile that makes it easy to create
1888 "pcre.dll" under mingw when using Configure/Make. I added stuff to
1889 Makefile.am that cause it to include this special target, without
1890 affecting anything else. Note that the same mingw target plus all
1891 the other distribution libraries and programs are now supported
1892 when configuring with CMake (see 6 below) instead of with
1893 Configure/Make.
1894
1895 5. Applied Craig's patch that moves no_arg into the RE class in the C++ code.
1896 This is an attempt to solve the reported problem "pcrecpp::no_arg is not
1897 exported in the Windows port". It has not yet been confirmed that the patch
1898 solves the problem, but it does no harm.
1899
1900 6. Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
1901 NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
1902 with CMake, and also correct the comment about stack recursion.
1903
1904 7. Remove the automatic building of pcredemo from the ./configure system and
1905 from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
1906 of a program that users should build themselves after PCRE is installed, so
1907 building it automatically is not really right. What is more, it gave
1908 trouble in some build environments.
1909
1910 8. Further tidies to CMakeLists.txt from Sheri and Christian.
1911
1912
1913 Version 7.5 10-Jan-08
1914 ---------------------
1915
1916 1. Applied a patch from Craig: "This patch makes it possible to 'ignore'
1917 values in parens when parsing an RE using the C++ wrapper."
1918
1919 2. Negative specials like \S did not work in character classes in UTF-8 mode.
1920 Characters greater than 255 were excluded from the class instead of being
1921 included.
1922
1923 3. The same bug as (2) above applied to negated POSIX classes such as
1924 [:^space:].
1925
1926 4. PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
1927 defined or documented. It seems to have been a typo for PCRE_STATIC, so
1928 I have changed it.
1929
1930 5. The construct (?&) was not diagnosed as a syntax error (it referenced the
1931 first named subpattern) and a construct such as (?&a) would reference the
1932 first named subpattern whose name started with "a" (in other words, the
1933 length check was missing). Both these problems are fixed. "Subpattern name
1934 expected" is now given for (?&) (a zero-length name), and this patch also
1935 makes it give the same error for \k'' (previously it complained that that
1936 was a reference to a non-existent subpattern).
1937
1938 6. The erroneous patterns (?+-a) and (?-+a) give different error messages;
1939 this is right because (?- can be followed by option settings as well as by
1940 digits. I have, however, made the messages clearer.
1941
1942 7. Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
1943 than the number used in the conditional) now cause a compile-time error.
1944 This is actually not compatible with Perl, which accepts such patterns, but
1945 treats the conditional as always being FALSE (as PCRE used to), but it
1946 seems to me that giving a diagnostic is better.
1947
1948 8. Change "alphameric" to the more common word "alphanumeric" in comments
1949 and messages.
1950
1951 9. Fix two occurrences of "backslash" in comments that should have been
1952 "backspace".
1953
1954 10. Remove two redundant lines of code that can never be obeyed (their function
1955 was moved elsewhere).
1956
1957 11. The program that makes PCRE's Unicode character property table had a bug
1958 which caused it to generate incorrect table entries for sequences of
1959 characters that have the same character type, but are in different scripts.
1960 It amalgamated them into a single range, with the script of the first of
1961 them. In other words, some characters were in the wrong script. There were
1962 thirteen such cases, affecting characters in the following ranges:
1963
1964 U+002b0 - U+002c1
1965 U+0060c - U+0060d
1966 U+0061e - U+00612
1967 U+0064b - U+0065e
1968 U+0074d - U+0076d
1969 U+01800 - U+01805
1970 U+01d00 - U+01d77
1971 U+01d9b - U+01dbf
1972 U+0200b - U+0200f
1973 U+030fc - U+030fe
1974 U+03260 - U+0327f
1975 U+0fb46 - U+0fbb1
1976 U+10450 - U+1049d
1977
1978 12. The -o option (show only the matching part of a line) for pcregrep was not
1979 compatible with GNU grep in that, if there was more than one match in a
1980 line, it showed only the first of them. It now behaves in the same way as
1981 GNU grep.
1982
1983 13. If the -o and -v options were combined for pcregrep, it printed a blank
1984 line for every non-matching line. GNU grep prints nothing, and pcregrep now
1985 does the same. The return code can be used to tell if there were any
1986 non-matching lines.
1987
1988 14. Added --file-offsets and --line-offsets to pcregrep.
1989
1990 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
1991 infinitely looping recursion. The bug was that positive lookaheads were not
1992 being skipped when checking for a possible empty match (negative lookaheads
1993 and both kinds of lookbehind were skipped).
1994
1995 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
1996 inclusion of <windows.h> to before rather than after the definition of
1997 INVALID_FILE_ATTRIBUTES (patch from David Byron).
1998
1999 17. Specifying a possessive quantifier with a specific limit for a Unicode
2000 character property caused pcre_compile() to compile bad code, which led at
2001 runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
2002 are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
2003 caused the error; without that there was no problem.
2004
2005 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
2006
2007 19. Added --enable-pcretest-libreadline.
2008
2009 20. In pcrecpp.cc, the variable 'count' was incremented twice in
2010 RE::GlobalReplace(). As a result, the number of replacements returned was
2011 double what it should be. I removed one of the increments, but Craig sent a
2012 later patch that removed the other one (the right fix) and added unit tests
2013 that check the return values (which was not done before).
2014
2015 21. Several CMake things:
2016
2017 (1) Arranged that, when cmake is used on Unix, the libraries end up with
2018 the names libpcre and libpcreposix, not just pcre and pcreposix.
2019
2020 (2) The above change means that pcretest and pcregrep are now correctly
2021 linked with the newly-built libraries, not previously installed ones.
2022
2023 (3) Added PCRE_SUPPORT_LIBREADLINE, PCRE_SUPPORT_LIBZ, PCRE_SUPPORT_LIBBZ2.
2024
2025 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
2026 crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
2027 UTF-8 newline character). The key issue is that the pattern starts .*;
2028 this means that the match must be either at the beginning, or after a
2029 newline. The bug was in the code for advancing after a failed match and
2030 checking that the new position followed a newline. It was not taking
2031 account of UTF-8 characters correctly.
2032
2033 23. PCRE was behaving differently from Perl in the way it recognized POSIX
2034 character classes. PCRE was not treating the sequence [:...:] as a
2035 character class unless the ... were all letters. Perl, however, seems to
2036 allow any characters between [: and :], though of course it rejects as
2037 unknown any "names" that contain non-letters, because all the known class
2038 names consist only of letters. Thus, Perl gives an error for [[:1234:]],
2039 for example, whereas PCRE did not - it did not recognize a POSIX character
2040 class. This seemed a bit dangerous, so the code has been changed to be
2041 closer to Perl. The behaviour is not identical to Perl, because PCRE will
2042 diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
2043 treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
2044 Perl does, and where it didn't before.
2045
2046 24. Rewrite so as to remove the single use of %n from pcregrep because in some
2047 Windows environments %n is disabled by default.
2048
2049
2050 Version 7.4 21-Sep-07
2051 ---------------------
2052
2053 1. Change 7.3/28 was implemented for classes by looking at the bitmap. This
2054 means that a class such as [\s] counted as "explicit reference to CR or
2055 LF". That isn't really right - the whole point of the change was to try to
2056 help when there was an actual mention of one of the two characters. So now
2057 the change happens only if \r or \n (or a literal CR or LF) character is
2058 encountered.
2059
2060 2. The 32-bit options word was also used for 6 internal flags, but the numbers
2061 of both had grown to the point where there were only 3 bits left.
2062 Fortunately, there was spare space in the data structure, and so I have
2063 moved the internal flags into a new 16-bit field to free up more option
2064 bits.
2065
2066 3. The appearance of (?J) at the start of a pattern set the DUPNAMES option,
2067 but did not set the internal JCHANGED flag - either of these is enough to
2068 control the way the "get" function works - but the PCRE_INFO_JCHANGED
2069 facility is supposed to tell if (?J) was ever used, so now (?J) at the
2070 start sets both bits.
2071
2072 4. Added options (at build time, compile time, exec time) to change \R from
2073 matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
2074
2075 5. doc/pcresyntax.html was missing from the distribution.
2076
2077 6. Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
2078 compatibility, even though it is no longer used.
2079
2080 7. Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
2081 strtoull to pcrecpp.cc to select the available functions in WIN32 when the
2082 windows.h file is present (where different names are used). [This was
2083 reversed later after testing - see 16 below.]
2084
2085 8. Changed all #include <config.h> to #include "config.h". There were also
2086 some further <pcre.h> cases that I changed to "pcre.h".
2087
2088 9. When pcregrep was used with the --colour option, it missed the line ending
2089 sequence off the lines that it output.
2090
2091 10. It was pointed out to me that arrays of string pointers cause lots of
2092 relocations when a shared library is dynamically loaded. A technique of
2093 using a single long string with a table of offsets can drastically reduce
2094 these. I have refactored PCRE in four places to do this. The result is
2095 dramatic:
2096
2097 Originally: 290
2098 After changing UCP table: 187
2099 After changing error message table: 43
2100 After changing table of "verbs" 36
2101 After changing table of Posix names 22
2102
2103 Thanks to the folks working on Gregex for glib for this insight.
2104
2105 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
2106 unicode-properties was also set.
2107
2108 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
2109
2110 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
2111 checked only for CRLF.
2112
2113 14. Added casts to pcretest.c to avoid compiler warnings.
2114
2115 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
2116
2117 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
2118 and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
2119 entirely. This removes changes made in 7 above.
2120
2121 17. The CMake files have been updated, and there is now more information about
2122 building with CMake in the NON-UNIX-USE document.
2123
2124
2125 Version 7.3 28-Aug-07
2126 ---------------------
2127
2128 1. In the rejigging of the build system that eventually resulted in 7.1, the
2129 line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
2130 brackets there is not right, since it causes compilers to look for an
2131 installed pcre.h, not the version that is in the source that is being
2132 compiled (which of course may be different). I have changed it back to:
2133
2134 #include "pcre.h"
2135
2136 I have a vague recollection that the change was concerned with compiling in
2137 different directories, but in the new build system, that is taken care of
2138 by the VPATH setting the Makefile.
2139
2140 2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
2141 when the subject happened to end in the byte 0x85 (e.g. if the last
2142 character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
2143 characters but of course it shouldn't be taken as a newline when it is part
2144 of another character. The bug was that, for an unlimited repeat of . in
2145 not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
2146 characters when looking for a newline.
2147
2148 3. A small performance improvement in the DOTALL UTF-8 mode .* case.
2149
2150 4. Debugging: adjusted the names of opcodes for different kinds of parentheses
2151 in debug output.
2152
2153 5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
2154 long printing in the pcrecpp unittest when running under MinGW.
2155
2156 6. ESC_K was left out of the EBCDIC table.
2157
2158 7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
2159 parentheses; I made it 1000, which seemed large enough. Unfortunately, the
2160 limit also applies to "virtual nesting" when a pattern is recursive, and in
2161 this case 1000 isn't so big. I have been able to remove this limit at the
2162 expense of backing off one optimization in certain circumstances. Normally,
2163 when pcre_exec() would call its internal match() function recursively and
2164 immediately return the result unconditionally, it uses a "tail recursion"
2165 feature to save stack. However, when a subpattern that can match an empty
2166 string has an unlimited repetition quantifier, it no longer makes this
2167 optimization. That gives it a stack frame in which to save the data for
2168 checking that an empty string has been matched. Previously this was taken
2169 from the 1000-entry workspace that had been reserved. So now there is no
2170 explicit limit, but more stack is used.
2171
2172 8. Applied Daniel's patches to solve problems with the import/export magic
2173 syntax that is required for Windows, and which was going wrong for the
2174 pcreposix and pcrecpp parts of the library. These were overlooked when this
2175 problem was solved for the main library.
2176
2177 9. There were some crude static tests to avoid integer overflow when computing
2178 the size of patterns that contain repeated groups with explicit upper
2179 limits. As the maximum quantifier is 65535, the maximum group length was
2180 set at 30,000 so that the product of these two numbers did not overflow a
2181 32-bit integer. However, it turns out that people want to use groups that
2182 are longer than 30,000 bytes (though not repeat them that many times).
2183 Change 7.0/17 (the refactoring of the way the pattern size is computed) has
2184 made it possible to implement the integer overflow checks in a much more
2185 dynamic way, which I have now done. The artificial limitation on group
2186 length has been removed - we now have only the limit on the total length of
2187 the compiled pattern, which depends on the LINK_SIZE setting.
2188
2189 10. Fixed a bug in the documentation for get/copy named substring when
2190 duplicate names are permitted. If none of the named substrings are set, the
2191 functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
2192 empty string.
2193
2194 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
2195 instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
2196 because the ] is interpreted as the first data character and the
2197 terminating ] is not found. PCRE has been made compatible with Perl in this
2198 regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
2199 cause memory overwriting.
2200
2201 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
2202 string has been matched (to stop an infinite loop). It was not recognizing
2203 a conditional subpattern that could match an empty string if that
2204 subpattern was within another subpattern. For example, it looped when
2205 trying to match (((?(1)X|))*) but it was OK with ((?(1)X|)*) where the
2206 condition was not nested. This bug has been fixed.
2207
2208 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
2209 past the start of the subject in the presence of bytes with the top bit
2210 set, for example "\x8aBCD".
2211
2212 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
2213 (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
2214
2215 14. Optimized (?!) to (*FAIL).
2216
2217 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
2218 This restricts code points to be within the range 0 to 0x10FFFF, excluding
2219 the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
2220 full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
2221 does: it's just the validity check that is more restrictive.
2222
2223 16. Inserted checks for integer overflows during escape sequence (backslash)
2224 processing, and also fixed erroneous offset values for syntax errors during
2225 backslash processing.
2226
2227 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
2228 for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
2229
2230 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
2231 caused an overrun.
2232
2233 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
2234 something other than just ASCII characters) inside a group that had an
2235 unlimited repeat caused a loop at compile time (while checking to see
2236 whether the group could match an empty string).
2237
2238 20. Debugging a pattern containing \p or \P could cause a crash. For example,
2239 [\P{Any}] did so. (Error in the code for printing property names.)
2240
2241 21. An orphan \E inside a character class could cause a crash.
2242
2243 22. A repeated capturing bracket such as (A)? could cause a wild memory
2244 reference during compilation.
2245
2246 23. There are several functions in pcre_compile() that scan along a compiled
2247 expression for various reasons (e.g. to see if it's fixed length for look
2248 behind). There were bugs in these functions when a repeated \p or \P was
2249 present in the pattern. These operators have additional parameters compared
2250 with \d, etc, and these were not being taken into account when moving along
2251 the compiled data. Specifically:
2252
2253 (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
2254 length.
2255
2256 (b) An item such as \pL+ within a repeated group could cause crashes or
2257 loops.
2258
2259 (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
2260 "reference to non-existent subpattern" error.
2261
2262 (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
2263
2264 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
2265 characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
2266
2267 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
2268
2269 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
2270 character were causing crashes (broken optimization).
2271
2272 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
2273 \p or \P) caused a compile-time loop.
2274
2275 28. More problems have arisen in unanchored patterns when CRLF is a valid line
2276 break. For example, the unstudied pattern [\r\n]A does not match the string
2277 "\r\nA" because change 7.0/46 below moves the current point on by two
2278 characters after failing to match at the start. However, the pattern \nA
2279 *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
2280 the same is true. There doesn't seem any very clean way out of this, but
2281 what I have chosen to do makes the common cases work: PCRE now takes note
2282 of whether there can be an explicit match for \r or \n anywhere in the
2283 pattern, and if so, 7.0/46 no longer applies. As part of this change,
2284 there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
2285 pattern has explicit CR or LF references.
2286
2287 29. Added (*CR) etc for changing newline setting at start of pattern.
2288
2289
2290 Version 7.2 19-Jun-07
2291 ---------------------
2292
2293 1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
2294 which is apparently normally available under Windows.
2295
2296 2. Re-jig the pcregrep tests with different newline settings in an attempt
2297 to make them independent of the local environment's newline setting.
2298
2299 3. Add code to configure.ac to remove -g from the CFLAGS default settings.
2300
2301 4. Some of the "internals" tests were previously cut out when the link size
2302 was not 2, because the output contained actual offsets. The recent new
2303 "Z" feature of pcretest means that these can be cut out, making the tests
2304 usable with all link sizes.
2305
2306 5. Implemented Stan Switzer's goto replacement for longjmp() when not using
2307 stack recursion. This gives a massive performance boost under BSD, but just
2308 a small improvement under Linux. However, it saves one field in the frame
2309 in all cases.
2310
2311 6. Added more features from the forthcoming Perl 5.10:
2312
2313 (a) (?-n) (where n is a string of digits) is a relative subroutine or
2314 recursion call. It refers to the nth most recently opened parentheses.
2315
2316 (b) (?+n) is also a relative subroutine call; it refers to the nth next
2317 to be opened parentheses.
2318
2319 (c) Conditions that refer to capturing parentheses can be specified
2320 relatively, for example, (?(-2)... or (?(+3)...
2321
2322 (d) \K resets the start of the current match so that everything before
2323 is not part of it.
2324
2325 (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
2326
2327 (f) \g{name} is another synonym - part of Perl 5.10's unification of
2328 reference syntax.
2329
2330 (g) (?| introduces a group in which the numbering of parentheses in each
2331 alternative starts with the same number.
2332
2333 (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
2334
2335 7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
2336 PCRE_INFO_JCHANGED.
2337
2338 8. A pattern such as (.*(.)?)* caused pcre_exec() to fail by either not
2339 terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
2340 for detecting groups that can match an empty string.
2341
2342 9. A pattern with a very large number of alternatives (more than several
2343 hundred) was running out of internal workspace during the pre-compile
2344 phase, where pcre_compile() figures out how much memory will be needed. A
2345 bit of new cunning has reduced the workspace needed for groups with
2346 alternatives. The 1000-alternative test pattern now uses 12 bytes of
2347 workspace instead of running out of the 4096 that are available.
2348
2349 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
2350
2351 11. Applied patch from Google to remove an optimization that didn't quite work.
2352 The report of the bug said:
2353
2354 pcrecpp::RE("a*").FullMatch("aaa") matches, while
2355 pcrecpp::RE("a*?").FullMatch("aaa") does not, and
2356 pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
2357
2358 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
2359 it matched the wrong number of bytes.
2360
2361
2362 Version 7.1 24-Apr-07
2363 ---------------------
2364
2365 1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
2366 that is more "standard", making use of automake and other Autotools. There
2367 is some re-arrangement of the files and adjustment of comments consequent
2368 on this.
2369
2370 2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
2371 for recursive directory scanning broke on some systems because the files
2372 are not scanned in any specific order and on different systems the order
2373 was different. A call to "sort" has been inserted into RunGrepTest for the
2374 approprate test as a short-term fix. In the longer term there may be an
2375 alternative.
2376
2377 3. I had an email from Eric Raymond about problems translating some of PCRE's
2378 man pages to HTML (despite the fact that I distribute HTML pages, some
2379 people do their own conversions for various reasons). The problems
2380 concerned the use of low-level troff macros .br and .in. I have therefore
2381 removed all such uses from the man pages (some were redundant, some could
2382 be replaced by .nf/.fi pairs). The 132html script that I use to generate
2383 HTML has been updated to handle .nf/.fi and to complain if it encounters
2384 .br or .in.
2385
2386 4. Updated comments in configure.ac that get placed in config.h.in and also
2387 arranged for config.h to be included in the distribution, with the name
2388 config.h.generic, for the benefit of those who have to compile without
2389 Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
2390
2391 5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
2392 Weber: (1) pcre_internal.h was missing some function renames; (2) updated
2393 makevp.bat for the current PCRE, using the additional files
2394 makevp_c.txt, makevp_l.txt, and pcregexp.pas.
2395
2396 6. A Windows user reported a minor discrepancy with test 2, which turned out
2397 to be caused by a trailing space on an input line that had got lost in his
2398 copy. The trailing space was an accident, so I've just removed it.
2399
2400 7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
2401 that is needed.
2402
2403 8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
2404 as "const" (a) because they are and (b) because it helps the PHP
2405 maintainers who have recently made a script to detect big data structures
2406 in the php code that should be moved to the .rodata section. I remembered
2407 to update Builducptable as well, so it won't revert if ucptable.h is ever
2408 re-created.
2409
2410 9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
2411 pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
2412 order to be able to cut out the UTF-8 tables in the latter when UTF-8
2413 support is not required. This saves 1.5-2K of code, which is important in
2414 some applications.
2415
2416 Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
2417 so as not to refer to the tables, even though these functions will never be
2418 called when UTF-8 support is disabled. Otherwise there are problems with a
2419 shared library.
2420
2421 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
2422
2423 (a) It was defining its arguments as char * instead of void *.
2424
2425 (b) It was assuming that all moves were upwards in memory; this was true
2426 a long time ago when I wrote it, but is no longer the case.
2427
2428 The emulated memove() is provided for those environments that have neither
2429 memmove() nor bcopy(). I didn't think anyone used it these days, but that
2430 is clearly not the case, as these two bugs were recently reported.
2431
2432 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
2433 and Detrail to create the HTML documentation, the .txt form of the man
2434 pages, and it removes trailing spaces from listed files. It also creates
2435 pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
2436 case, it wraps all the #defines with #ifndefs. This script should be run
2437 before "make dist".
2438
2439 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
2440 with Unicode property support.
2441
2442 (a) For a maximizing quantifier, if the two different cases of the
2443 character were of different lengths in their UTF-8 codings (there are
2444 some cases like this - I found 11), and the matching function had to
2445 back up over a mixture of the two cases, it incorrectly assumed they
2446 were both the same length.
2447
2448 (b) When PCRE was configured to use the heap rather than the stack for
2449 recursion during matching, it was not correctly preserving the data for
2450 the other case of a UTF-8 character when checking ahead for a match
2451 while processing a minimizing repeat. If the check also involved
2452 matching a wide character, but failed, corruption could cause an
2453 erroneous result when trying to check for a repeat of the original
2454 character.
2455
2456 13. Some tidying changes to the testing mechanism:
2457
2458 (a) The RunTest script now detects the internal link size and whether there
2459 is UTF-8 and UCP support by running ./pcretest -C instead of relying on
2460 values substituted by "configure". (The RunGrepTest script already did
2461 this for UTF-8.) The configure.ac script no longer substitutes the
2462 relevant variables.
2463
2464 (b) The debugging options /B and /D in pcretest show the compiled bytecode
2465 with length and offset values. This means that the output is different
2466 for different internal link sizes. Test 2 is skipped for link sizes
2467 other than 2 because of this, bypassing the problem. Unfortunately,
2468 there was also a test in test 3 (the locale tests) that used /B and
2469 failed for link sizes other than 2. Rather than cut the whole test out,
2470 I have added a new /Z option to pcretest that replaces the length and
2471 offset values with spaces. This is now used to make test 3 independent
2472 of link size. (Test 2 will be tidied up later.)
2473
2474 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
2475 segmentation fault instead of returning the appropriate error message.
2476
2477 15. In multiline mode when the newline sequence was set to "any", the pattern
2478 ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
2479 This doesn't seem right; it now treats the CRLF combination as the line
2480 ending, and so does not match in that case. It's only a pattern such as ^$
2481 that would hit this one: something like ^ABC$ would have failed after \r
2482 and then tried again after \r\n.
2483
2484 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
2485 in an attempt to make files that differ only in their line terminators
2486 compare equal. This works on Linux.
2487
2488 17. Under certain error circumstances pcregrep might try to free random memory
2489 as it exited. This is now fixed, thanks to valgrind.
2490
2491 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
2492 "abc\r\n\r\n", it found an unwanted second match after the second \r. This
2493 was because its rules for how to advance for /g after matching an empty
2494 string at the end of a line did not allow for this case. They now check for
2495 it specially.
2496
2497 20. pcretest is supposed to handle patterns and data of any length, by
2498 extending its buffers when necessary. It was getting this wrong when the
2499 buffer for a data line had to be extended.
2500
2501 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
2502 CRLF as a newline sequence.
2503
2504 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
2505 out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
2506 I have nevertheless tidied it up.
2507
2508 23. Added some casts to kill warnings from HP-UX ia64 compiler.
2509
2510 24. Added a man page for pcre-config.
2511
2512
2513 Version 7.0 19-Dec-06
2514 ---------------------
2515
2516 1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
2517 moving to gcc 4.1.1.
2518
2519 2. The -S option for pcretest uses setrlimit(); I had omitted to #include
2520 sys/time.h, which is documented as needed for this function. It doesn't
2521 seem to matter on Linux, but it showed up on some releases of OS X.
2522
2523 3. It seems that there are systems where bytes whose values are greater than
2524 127 match isprint() in the "C" locale. The "C" locale should be the
2525 default when a C program starts up. In most systems, only ASCII printing
2526 characters match isprint(). This difference caused the output from pcretest
2527 to vary, making some of the tests fail. I have changed pcretest so that:
2528
2529 (a) When it is outputting text in the compiled version of a pattern, bytes
2530 other than 32-126 are always shown as hex escapes.
2531
2532 (b) When it is outputting text that is a matched part of a subject string,
2533 it does the same, unless a different locale has been set for the match
2534 (using the /L modifier). In this case, it uses isprint() to decide.
2535
2536 4. Fixed a major bug that caused incorrect computation of the amount of memory
2537 required for a compiled pattern when options that changed within the
2538 pattern affected the logic of the preliminary scan that determines the
2539 length. The relevant options are -x, and -i in UTF-8 mode. The result was
2540 that the computed length was too small. The symptoms of this bug were
2541 either the PCRE error "internal error: code overflow" from pcre_compile(),
2542 or a glibc crash with a message such as "pcretest: free(): invalid next
2543 size (fast)". Examples of patterns that provoked this bug (shown in
2544 pcretest format) are:
2545
2546 /(?-x: )/x
2547 /(?x)(?-x: \s*#\s*)/
2548 /((?i)[\x{c0}])/8
2549 /(?i:[\x{c0}])/8
2550
2551 HOWEVER: Change 17 below makes this fix obsolete as the memory computation
2552 is now done differently.
2553
2554 5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
2555 wrapper classes; (b) implement a new function in the C++ scanner that is
2556 more efficient than the old way of doing things because it avoids levels of
2557 recursion in the regex matching; (c) add a paragraph to the documentation
2558 for the FullMatch() function.
2559
2560 6. The escape sequence \n was being treated as whatever was defined as
2561 "newline". Not only was this contrary to the documentation, which states
2562 that \n is character 10 (hex 0A), but it also went horribly wrong when
2563 "newline" was defined as CRLF. This has been fixed.
2564
2565 7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
2566 was being set to -1 for the "end of line" case (supposedly a value that no
2567 character can have). Though this value is never used (the check for end of
2568 line is "zero bytes in current character"), it caused compiler complaints.
2569 I've changed it to 0xffffffff.
2570
2571 8. In pcre_version.c, the version string was being built by a sequence of
2572 C macros that, in the event of PCRE_PRERELEASE being defined as an empty
2573 string (as it is for production releases) called a macro with an empty
2574 argument. The C standard says the result of this is undefined. The gcc
2575 compiler treats it as an empty string (which was what was wanted) but it is
2576 reported that Visual C gives an error. The source has been hacked around to
2577 avoid this problem.
2578
2579 9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
2580 builds of pcretest, and changed the call to _setmode() to use _O_BINARY
2581 instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
2582 of them did).
2583
2584 10. Originally, pcretest opened its input and output without "b"; then I was
2585 told that "b" was needed in some environments, so it was added for release
2586 5.0 to both the input and output. (It makes no difference on Unix-like
2587 systems.) Later I was told that it is wrong for the input on Windows. I've
2588 now abstracted the modes into two macros, to make it easier to fiddle with
2589 them, and removed "b" from the input mode under Windows.
2590
2591 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
2592
2593 12. Added -help and --help to pcretest as an official way of being reminded
2594 of the options.
2595
2596 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
2597 and pcrecpp.cc because they annoy compilers at high warning levels.
2598
2599 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
2600
2601 15. Fixed an occurrence of == in configure.ac that should have been = (shell
2602 scripts are not C programs :-) and which was not noticed because it works
2603 on Linux.
2604
2605 16. pcretest is supposed to handle any length of pattern and data line (as one
2606 line or as a continued sequence of lines) by extending its input buffer if
2607 necessary. This feature was broken for very long pattern lines, leading to
2608 a string of junk being passed to pcre_compile() if the pattern was longer
2609 than about 50K.
2610
2611 17. I have done a major re-factoring of the way pcre_compile() computes the
2612 amount of memory needed for a compiled pattern. Previously, there was code
2613 that made a preliminary scan of the pattern in order to do this. That was
2614 OK when PCRE was new, but as the facilities have expanded, it has become
2615 harder and harder to keep it in step with the real compile phase, and there
2616 have been a number of bugs (see for example, 4 above). I have now found a
2617 cunning way of running the real compile function in a "fake" mode that
2618 enables it to compute how much memory it would need, while actually only
2619 ever using a few hundred bytes of working memory and without too many
2620 tests of the mode. This should make future maintenance and development
2621 easier. A side effect of this work is that the limit of 200 on the nesting
2622 depth of parentheses has been removed (though this was never a serious
2623 limitation, I suspect). However, there is a downside: pcre_compile() now
2624 runs more slowly than before (30% or more, depending on the pattern). I
2625 hope this isn't a big issue. There is no effect on runtime performance.
2626
2627 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
2628 newline (only possible for the last line of a file) and it was a
2629 pattern that set a locale (followed by /Lsomething), pcretest crashed.
2630
2631 19. Added additional timing features to pcretest. (1) The -tm option now times
2632 matching only, not compiling. (2) Both -t and -tm can be followed, as a
2633 separate command line item, by a number that specifies the number of
2634 repeats to use when timing. The default is 50000; this gives better
2635 precision, but takes uncomfortably long for very large patterns.
2636
2637 20. Extended pcre_study() to be more clever in cases where a branch of a
2638 subpattern has no definite first character. For example, (a*|b*)[cd] would
2639 previously give no result from pcre_study(). Now it recognizes that the
2640 first character must be a, b, c, or d.
2641
2642 21. There was an incorrect error "recursive call could loop indefinitely" if
2643 a subpattern (or the entire pattern) that was being tested for matching an
2644 empty string contained only one non-empty item after a nested subpattern.
2645 For example, the pattern (?>\x{100}*)\d(?R) provoked this error
2646 incorrectly, because the \d was being skipped in the check.
2647
2648 22. The pcretest program now has a new pattern option /B and a command line
2649 option -b, which is equivalent to adding /B to every pattern. This causes
2650 it to show the compiled bytecode, without the additional information that
2651 -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
2652 is the same as /B/I).
2653
2654 23. A new optimization is now able automatically to treat some sequences such
2655 as a*b as a*+b. More specifically, if something simple (such as a character
2656 or a simple class like \d) has an unlimited quantifier, and is followed by
2657 something that cannot possibly match the quantified thing, the quantifier
2658 is automatically "possessified".
2659
2660 24. A recursive reference to a subpattern whose number was greater than 39
2661 went wrong under certain circumstances in UTF-8 mode. This bug could also
2662 have affected the operation of pcre_study().
2663
2664 25. Realized that a little bit of performance could be had by replacing
2665 (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
2666
2667 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
2668
2669 27. Possessive quantifiers such as a++ were previously implemented by turning
2670 them into atomic groups such as ($>a+). Now they have their own opcodes,
2671 which improves performance. This includes the automatically created ones
2672 from 23 above.
2673
2674 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
2675 lookahead was broken if it was not anchored. PCRE was mistakenly expecting
2676 the first matched character to be a colon. This applied both to named and
2677 numbered groups.
2678
2679 29. The ucpinternal.h header file was missing its idempotency #ifdef.
2680
2681 30. I was sent a "project" file called libpcre.a.dev which I understand makes
2682 building PCRE on Windows easier, so I have included it in the distribution.
2683
2684 31. There is now a check in pcretest against a ridiculously large number being
2685 returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
2686 loop, the loop is abandoned.
2687
2688 32. Forward references to subpatterns in conditions such as (?(2)...) where
2689 subpattern 2 is defined later cause pcre_compile() to search forwards in
2690 the pattern for the relevant set of parentheses. This search went wrong
2691 when there were unescaped parentheses in a character class, parentheses
2692 escaped with \Q...\E, or parentheses in a #-comment in /x mode.
2693
2694 33. "Subroutine" calls and backreferences were previously restricted to
2695 referencing subpatterns earlier in the regex. This restriction has now
2696 been removed.
2697
2698 34. Added a number of extra features that are going to be in Perl 5.10. On the
2699 whole, these are just syntactic alternatives for features that PCRE had
2700 previously implemented using the Python syntax or my own invention. The
2701 other formats are all retained for compatibility.
2702
2703 (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
2704 as (?P<name>...). The new forms, as well as being in Perl 5.10, are
2705 also .NET compatible.
2706
2707 (b) A recursion or subroutine call to a named group can now be defined as
2708 (?&name) as well as (?P>name).
2709
2710 (c) A backreference to a named group can now be defined as \k<name> or
2711 \k'name' as well as (?P=name). The new forms, as well as being in Perl
2712 5.10, are also .NET compatible.
2713
2714 (d) A conditional reference to a named group can now use the syntax
2715 (?(<name>) or (?('name') as well as (?(name).
2716
2717 (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
2718 groups (named and numbered) that are never evaluated inline, but can be
2719 called as "subroutines" from elsewhere. In effect, the DEFINE condition
2720 is always false. There may be only one alternative in such a group.
2721
2722 (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
2723 as the simple (?(R). The condition is true only if the most recent
2724 recursion is that of the given number or name. It does not search out
2725 through the entire recursion stack.
2726
2727 (g) The escape \gN or \g{N} has been added, where N is a positive or
2728 negative number, specifying an absolute or relative reference.
2729
2730 35. Tidied to get rid of some further signed/unsigned compiler warnings and
2731 some "unreachable code" warnings.
2732
2733 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
2734 things, this adds five new scripts.
2735
2736 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
2737 There were also incompatibilities regarding the handling of \Q..\E inside
2738 character classes, for example with patterns like [\Qa\E-\Qz\E] where the
2739 hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
2740
2741 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
2742 matches an empty string, and forcibly breaks the loop. There were bugs in
2743 this code in non-simple cases. For a pattern such as ^(a()*)* matched
2744 against aaaa the result was just "a" rather than "aaaa", for example. Two
2745 separate and independent bugs (that affected different cases) have been
2746 fixed.
2747
2748 39. Refactored the code to abolish the use of different opcodes for small
2749 capturing bracket numbers. This is a tidy that I avoided doing when I
2750 removed the limit on the number of capturing brackets for 3.5 back in 2001.
2751 The new approach is not only tidier, it makes it possible to reduce the
2752 memory needed to fix the previous bug (38).
2753
2754 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
2755 sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
2756 processing dot, circumflex, or dollar metacharacters, or #-comments in /x
2757 mode.
2758
2759 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
2760 report.
2761
2762 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
2763 copy construction and assignment in the C++ wrapper.
2764
2765 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
2766 couple of bugs that could have given wrong results in the "--newline=crlf"
2767 case.
2768
2769 44. Added a number of casts and did some reorganization of signed/unsigned int
2770 variables following suggestions from Dair Grant. Also renamed the variable
2771 "this" as "item" because it is a C++ keyword.
2772
2773 45. Arranged for dftables to add
2774
2775 #include "pcre_internal.h"
2776
2777 to pcre_chartables.c because without it, gcc 4.x may remove the array
2778 definition from the final binary if PCRE is built into a static library and
2779 dead code stripping is activated.
2780
2781 46. For an unanchored pattern, if a match attempt fails at the start of a
2782 newline sequence, and the newline setting is CRLF or ANY, and the next two
2783 characters are CRLF, advance by two characters instead of one.
2784
2785
2786 Version 6.7 04-Jul-06
2787 ---------------------
2788
2789 1. In order to handle tests when input lines are enormously long, pcretest has
2790 been re-factored so that it automatically extends its buffers when
2791 necessary. The code is crude, but this _is_ just a test program. The
2792 default size has been increased from 32K to 50K.
2793
2794 2. The code in pcre_study() was using the value of the re argument before
2795 testing it for NULL. (Of course, in any sensible call of the function, it
2796 won't be NULL.)
2797
2798 3. The memmove() emulation function in pcre_internal.h, which is used on
2799 systems that lack both memmove() and bcopy() - that is, hardly ever -
2800 was missing a "static" storage class specifier.
2801
2802 4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
2803 containing an extended class (one that cannot be represented by a bitmap
2804 because it contains high-valued characters or Unicode property items, e.g.
2805 [\pZ]). Almost always one would set UTF-8 mode when processing such a
2806 pattern, but PCRE should not loop if you do not (it no longer does).
2807 [Detail: two cases were found: (a) a repeated subpattern containing an
2808 extended class; (b) a recursive reference to a subpattern that followed a
2809 previous extended class. It wasn't skipping over the extended class
2810 correctly when UTF-8 mode was not set.]
2811
2812 5. A negated single-character class was not being recognized as fixed-length
2813 in lookbehind assertions such as (?<=[^f]), leading to an incorrect
2814 compile error "lookbehind assertion is not fixed length".
2815
2816 6. The RunPerlTest auxiliary script was showing an unexpected difference
2817 between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
2818 write a Perl script that can interpret lines of an input file either as
2819 byte characters or as UTF-8, which is what "perltest" was being required to
2820 do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
2821 can't do is switch easily at run time between having the "use utf8;" pragma
2822 or not. In the end, I fudged it by using the RunPerlTest script to insert
2823 "use utf8;" explicitly for the UTF-8 tests.
2824
2825 7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
2826 the end of the subject string, contrary to the documentation and to what
2827 Perl does. This was true of both matching functions. Now it matches only at
2828 the start of the subject and immediately after *internal* newlines.
2829
2830 8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
2831 a pointer to an int instead of a pointer to an unsigned long int. This
2832 caused problems on 64-bit systems.
2833
2834 9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
2835 instance of the 'standard' template library not being so standard".
2836
2837 10. There was no check on the number of named subpatterns nor the maximum
2838 length of a subpattern name. The product of these values is used to compute
2839 the size of the memory block for a compiled pattern. By supplying a very
2840 long subpattern name and a large number of named subpatterns, the size
2841 computation could be caused to overflow. This is now prevented by limiting
2842 the length of names to 32 characters, and the number of named subpatterns
2843 to 10,000.
2844
2845 11. Subpatterns that are repeated with specific counts have to be replicated in
2846 the compiled pattern. The size of memory for this was computed from the
2847 length of the subpattern and the repeat count. The latter is limited to
2848 65535, but there was no limit on the former, meaning that integer overflow
2849 could in principle occur. The compiled length of a repeated subpattern is
2850 now limited to 30,000 bytes in order to prevent this.
2851
2852 12. Added the optional facility to have named substrings with the same name.
2853
2854 13. Added the ability to use a named substring as a condition, using the
2855 Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
2856 are numbers (not recommended). Forward references are permitted.
2857
2858 14. Added forward references in named backreferences (if you see what I mean).
2859
2860 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
2861 pattern could run off the end of the subject. For example, the pattern
2862 "(?s)(.{1,5})"8 did this with the subject "ab".
2863
2864 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
2865 PCRE_CASELESS was set when matching characters that were quantified with ?
2866 or *.
2867
2868 17. A character class other than a single negated character that had a minimum
2869 but no maximum quantifier - for example [ab]{6,} - was not handled
2870 correctly by pce_dfa_exec(). It would match only one character.
2871
2872 18. A valid (though odd) pattern that looked like a POSIX character
2873 class but used an invalid character after [ (for example [[,abc,]]) caused
2874 pcre_compile() to give the error "Failed: internal error: code overflow" or
2875 in some cases to crash with a glibc free() error. This could even happen if
2876 the pattern terminated after [[ but there just happened to be a sequence of
2877 letters, a binary zero, and a closing ] in the memory that followed.
2878
2879 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
2880 over the years. Originally (before any Unicode support), just the bottom 8
2881 bits were taken. Thus, for example, \500 really meant \100. Nowadays the
2882 output from "man perlunicode" includes this:
2883
2884 The regular expression compiler produces polymorphic opcodes. That
2885 is, the pattern adapts to the data and automatically switches to
2886 the Unicode character scheme when presented with Unicode data--or
2887 instead uses a traditional byte scheme when presented with byte
2888 data.
2889
2890 Sadly, a wide octal escape does not cause a switch, and in a string with
2891 no other multibyte characters, these octal escapes are treated as before.
2892 Thus, in Perl, the pattern /\500/ actually matches \100 but the pattern
2893 /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
2894 Unicode string.
2895
2896 I have not perpetrated such confusion in PCRE. Up till now, it took just
2897 the bottom 8 bits, as in old Perl. I have now made octal escapes with
2898 values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
2899 translate to the appropriate multibyte character.
2900
2901 29. Applied some refactoring to reduce the number of warnings from Microsoft
2902 and Borland compilers. This has included removing the fudge introduced
2903 seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
2904 a warning about an unused variable.
2905
2906 21. PCRE has not included VT (character 0x0b) in the set of whitespace
2907 characters since release 4.0, because Perl (from release 5.004) does not.
2908 [Or at least, is documented not to: some releases seem to be in conflict
2909 with the documentation.] However, when a pattern was studied with
2910 pcre_study() and all its branches started with \s, PCRE still included VT
2911 as a possible starting character. Of course, this did no harm; it just
2912 caused an unnecessary match attempt.
2913
2914 22. Removed a now-redundant internal flag bit that recorded the fact that case
2915 dependency changed within the pattern. This was once needed for "required
2916 byte" processing, but is no longer used. This recovers a now-scarce options
2917 bit. Also moved the least significant internal flag bit to the most-
2918 significant bit of the word, which was not previously used (hangover from
2919 the days when it was an int rather than a uint) to free up another bit for
2920 the future.
2921
2922 23. Added support for CRLF line endings as well as CR and LF. As well as the
2923 default being selectable at build time, it can now be changed at runtime
2924 via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
2925 specify that it is scanning data with non-default line endings.
2926
2927 24. Changed the definition of CXXLINK to make it agree with the definition of
2928 LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
2929
2930 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
2931 recursions. This makes a big different to stack usage for some patterns.
2932
2933 26. If a subpattern containing a named recursion or subroutine reference such
2934 as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
2935 the space required for the compiled pattern went wrong and gave too small a
2936 value. Depending on the environment, this could lead to "Failed: internal
2937 error: code overflow at offset 49" or "glibc detected double free or
2938 corruption" errors.
2939
2940 27. Applied patches from Google (a) to support the new newline modes and (b) to
2941 advance over multibyte UTF-8 characters in GlobalReplace.
2942
2943 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
2944 difference for some implementation of PCRE in some Windows version.
2945
2946 29. Added some extra testing facilities to pcretest:
2947
2948 \q<number> in a data line sets the "match limit" value
2949 \Q<number> in a data line sets the "match recursion limt" value
2950 -S <number> sets the stack size, where <number> is in megabytes
2951
2952 The -S option isn't available for Windows.
2953
2954
2955 Version 6.6 06-Feb-06
2956 ---------------------
2957
2958 1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
2959 in pcreposix.h. I have copied the definition from pcre.h.
2960
2961 2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
2962 because pcre.h is no longer a built file.
2963
2964 3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
2965 not normally included in the compiled code.
2966
2967
2968 Version 6.5 01-Feb-06
2969 ---------------------
2970
2971 1. When using the partial match feature with pcre_dfa_exec(), it was not
2972 anchoring the second and subsequent partial matches at the new starting
2973 point. This could lead to incorrect results. For example, with the pattern
2974 /1234/, partially matching against "123" and then "a4" gave a match.
2975
2976 2. Changes to pcregrep:
2977
2978 (a) All non-match returns from pcre_exec() were being treated as failures
2979 to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
2980 error message is output. Some extra information is given for the
2981 PCRE_ERROR_MATCHLIMIT and PCRE_ERROR_RECURSIONLIMIT errors, which are
2982 probably the only errors that are likely to be caused by users (by
2983 specifying a regex that has nested indefinite repeats, for instance).
2984 If there are more than 20 of these errors, pcregrep is abandoned.
2985
2986 (b) A binary zero was treated as data while matching, but terminated the
2987 output line if it was written out. This has been fixed: binary zeroes
2988 are now no different to any other data bytes.
2989
2990 (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
2991 used to set a locale for matching. The --locale=xxxx long option has
2992 been added (no short equivalent) to specify a locale explicitly on the
2993 pcregrep command, overriding the environment variables.
2994
2995 (d) When -B was used with -n, some line numbers in the output were one less
2996 than they should have been.
2997
2998 (e) Added the -o (--only-matching) option.
2999
3000 (f) If -A or -C was used with -c (count only), some lines of context were
3001 accidentally printed for the final match.
3002
3003 (g) Added the -H (--with-filename) option.
3004
3005 (h) The combination of options -rh failed to suppress file names for files
3006 that were found from directory arguments.
3007
3008 (i) Added the -D (--devices) and -d (--directories) options.
3009
3010 (j) Added the -F (--fixed-strings) option.
3011
3012 (k) Allow "-" to be used as a file name for -f as well as for a data file.
3013
3014 (l) Added the --colo(u)r option.
3015
3016 (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
3017 is not present by default.
3018
3019 3. A nasty bug was discovered in the handling of recursive patterns, that is,
3020 items such as (?R) or (?1), when the recursion could match a number of
3021 alternatives. If it matched one of the alternatives, but subsequently,
3022 outside the recursion, there was a failure, the code tried to back up into
3023 the recursion. However, because of the way PCRE is implemented, this is not
3024 possible, and the result was an incorrect result from the match.
3025
3026 In order to prevent this happening, the specification of recursion has
3027 been changed so that all such subpatterns are automatically treated as
3028 atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
3029
3030 4. I had overlooked the fact that, in some locales, there are characters for
3031 which isalpha() is true but neither isupper() nor islower() are true. In
3032 the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
3033 and ordfeminine) are like this. This affected the treatment of \w and \W
3034 when they appeared in character classes, but not when they appeared outside
3035 a character class. The bit map for "word" characters is now created
3036 separately from the results of isalnum() instead of just taking it from the
3037 upper, lower, and digit maps. (Plus the underscore character, of course.)
3038
3039 5. The above bug also affected the handling of POSIX character classes such as
3040 [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
3041 permanent tables. Instead, the bit maps for such a class were previously
3042 created as the appropriate unions of the upper, lower, and digit bitmaps.
3043 Now they are created by subtraction from the [[:word:]] class, which has
3044 its own bitmap.
3045
3046 6. The [[:blank:]] character class matches horizontal, but not vertical space.
3047 It is created by subtracting the vertical space characters (\x09, \x0a,
3048 \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
3049 subtraction was done in the overall bitmap for a character class, meaning
3050 that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
3051 be recognized. This bug has been fixed.
3052
3053 7. Patches from the folks at Google:
3054
3055 (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
3056 real life, but is still worth protecting against".
3057
3058 (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
3059 regular expressions".
3060
3061 (c) pcre_scanner.cc: avoid use of std::count() because not all systems
3062 have it.
3063
3064 (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
3065 "configure" and the latter not, in order to fix a problem somebody had
3066 with compiling the Arg class on HP-UX.
3067
3068 (e) Improve the error-handling of the C++ wrapper a little bit.
3069
3070 (f) New tests for checking recursion limiting.
3071
3072 8. The pcre_memmove() function, which is used only if the environment does not
3073 have a standard memmove() function (and is therefore rarely compiled),
3074 contained two bugs: (a) use of int instead of size_t, and (b) it was not
3075 returning a result (though PCRE never actually uses the result).
3076
3077 9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
3078 large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
3079 returned instead of calling malloc() with an overflowing number that would
3080 most likely cause subsequent chaos.
3081
3082 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
3083
3084 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
3085 with this option is matched, the nmatch and pmatch options of regexec() are
3086 ignored.
3087
3088 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
3089 provided in case anyone wants to the the POSIX interface with UTF-8
3090 strings.
3091
3092 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
3093 C++ linking (needed for some HP-UX environments).
3094
3095 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
3096 (unused parameter) and in the pcre_printint() function (omitted "default"
3097 switch label when the default is to do nothing).
3098
3099 15. Added some code to make it possible, when PCRE is compiled as a C++
3100 library, to replace subject pointers for pcre_exec() with a smart pointer
3101 class, thus making it possible to process discontinuous strings.
3102
3103 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
3104 much the same function. They were added by different people who were trying
3105 to make PCRE easy to compile on non-Unix systems. It has been suggested
3106 that PCRE_EXPORT be abolished now that there is more automatic apparatus
3107 for compiling on Windows systems. I have therefore replaced it with
3108 PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
3109 defaults to "extern" for C or "extern C" for C++, which works fine on
3110 Unix-like systems. It is now possible to override the value of PCRE_DATA_
3111 SCOPE with something explicit in config.h. In addition:
3112
3113 (a) pcreposix.h still had just "extern" instead of either of these macros;
3114 I have replaced it with PCRE_DATA_SCOPE.
3115
3116 (b) Functions such as _pcre_xclass(), which are internal to the library,
3117 but external in the C sense, all had PCRE_EXPORT in their definitions.
3118 This is apparently wrong for the Windows case, so I have removed it.
3119 (It makes no difference on Unix-like systems.)
3120
3121 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
3122 of recursive calls to match(). This is different to MATCH_LIMIT because
3123 that limits the total number of calls to match(), not all of which increase
3124 the depth of recursion. Limiting the recursion depth limits the amount of
3125 stack (or heap if NO_RECURSE is set) that is used. The default can be set
3126 when PCRE is compiled, and changed at run time. A patch from Google adds
3127 this functionality to the C++ interface.
3128
3129 18. Changes to the handling of Unicode character properties:
3130
3131 (a) Updated the table to Unicode 4.1.0.
3132
3133 (b) Recognize characters that are not in the table as "Cn" (undefined).
3134
3135 (c) I revised the way the table is implemented to a much improved format
3136 which includes recognition of ranges. It now supports the ranges that
3137 are defined in UnicodeData.txt, and it also amalgamates other
3138 characters into ranges. This has reduced the number of entries in the
3139 table from around 16,000 to around 3,000, thus reducing its size
3140 considerably. I realized I did not need to use a tree structure after
3141 all - a binary chop search is just as efficient. Having reduced the
3142 number of entries, I extended their size from 6 bytes to 8 bytes to
3143 allow for more data.
3144
3145 (d) Added support for Unicode script names via properties such as \p{Han}.
3146
3147 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
3148 matching that character.
3149
3150 20. When matching a repeated Unicode property with a minimum greater than zero,
3151 (for example \pL{2,}), PCRE could look past the end of the subject if it
3152 reached it while seeking the minimum number of characters. This could
3153 happen only if some of the characters were more than one byte long, because
3154 there is a check for at least the minimum number of bytes.
3155
3156 21. Refactored the implementation of \p and \P so as to be more general, to
3157 allow for more different types of property in future. This has changed the
3158 compiled form incompatibly. Anybody with saved compiled patterns that use
3159 \p or \P will have to recompile them.
3160
3161 22. Added "Any" and "L&" to the supported property types.
3162
3163 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
3164 but give a compile time error if the value is greater than 0xff.
3165
3166 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
3167 accidentally not being installed or uninstalled.
3168
3169 25. The pcre.h file was built from pcre.h.in, but the only changes that were
3170 made were to insert the current release number. This seemed silly, because
3171 it made things harder for people building PCRE on systems that don't run
3172 "configure". I have turned pcre.h into a distributed file, no longer built
3173 by "configure", with the version identification directly included. There is
3174 no longer a pcre.h.in file.
3175
3176 However, this change necessitated a change to the pcre-config script as
3177 well. It is built from pcre-config.in, and one of the substitutions was the
3178 release number. I have updated configure.ac so that ./configure now finds
3179 the release number by grepping pcre.h.
3180
3181 26. Added the ability to run the tests under valgrind.
3182
3183
3184 Version 6.4 05-Sep-05
3185 ---------------------
3186
3187 1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
3188 "--" to be printed when multiple files were scanned, even when none of the
3189 -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
3190 consider it to be a bug, and have restored the previous behaviour.
3191
3192 2. A couple of code tidies to get rid of compiler warnings.
3193
3194 3. The pcretest program used to cheat by referring to symbols in the library
3195 whose names begin with _pcre_. These are internal symbols that are not
3196 really supposed to be visible externally, and in some environments it is
3197 possible to suppress them. The cheating is now confined to including
3198 certain files from the library's source, which is a bit cleaner.
3199
3200 4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
3201 file's purpose clearer.
3202
3203 5. Reorganized pcre_ucp_findchar().
3204
3205
3206 Version 6.3 15-Aug-05
3207 ---------------------
3208
3209 1. The file libpcre.pc.in did not have general read permission in the tarball.
3210
3211 2. There were some problems when building without C++ support:
3212
3213 (a) If C++ support was not built, "make install" and "make test" still
3214 tried to test it.
3215
3216 (b) There were problems when the value of CXX was explicitly set. Some
3217 changes have been made to try to fix these, and ...
3218
3219 (c) --disable-cpp can now be used to explicitly disable C++ support.
3220
3221 (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
3222 backslash in a target when C++ was disabled. This confuses some
3223 versions of "make", apparently. Using an intermediate variable solves
3224 this. (Same for CPP_LOBJ.)
3225
3226 3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
3227 (non-Windows) now includes $(CFLAGS) because these flags are sometimes
3228 necessary on certain architectures.
3229
3230 4. Added a setting of -export-symbols-regex to the link command to remove
3231 those symbols that are exported in the C sense, but actually are local
3232 within the library, and not documented. Their names all begin with
3233 "_pcre_". This is not a perfect job, because (a) we have to except some
3234 symbols that pcretest ("illegally") uses, and (b) the facility isn't always
3235 available (and never for static libraries). I have made a note to try to
3236 find a way round (a) in the future.
3237
3238
3239 Version 6.2 01-Aug-05
3240 ---------------------
3241
3242 1. There was no test for integer overflow of quantifier values. A construction
3243 such as {1111111111111111} would give undefined results. What is worse, if
3244 a minimum quantifier for a parenthesized subpattern overflowed and became
3245 negative, the calculation of the memory size went wrong. This could have
3246 led to memory overwriting.
3247
3248 2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
3249
3250 3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
3251 operating environments where this matters.
3252
3253 4. Applied Giuseppe Maxia's patch to add additional features for controlling
3254 PCRE options from within the C++ wrapper.
3255
3256 5. Named capturing subpatterns were not being correctly counted when a pattern
3257 was compiled. This caused two problems: (a) If there were more than 100
3258 such subpatterns, the calculation of the memory needed for the whole
3259 compiled pattern went wrong, leading to an overflow error. (b) Numerical
3260 back references of the form \12, where the number was greater than 9, were
3261 not recognized as back references, even though there were sufficient
3262 previous subpatterns.
3263
3264 6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
3265 versions of gcc, e.g. 2.95.4.
3266
3267
3268 Version 6.1 21-Jun-05
3269 ---------------------
3270
3271 1. There was one reference to the variable "posix" in pcretest.c that was not
3272 surrounded by "#if !defined NOPOSIX".
3273
3274 2. Make it possible to compile pcretest without DFA support, UTF8 support, or
3275 the cross-check on the old pcre_info() function, for the benefit of the
3276 cut-down version of PCRE that is currently imported into Exim.
3277
3278 3. A (silly) pattern starting with (?i)(?-i) caused an internal space
3279 allocation error. I've done the easy fix, which wastes 2 bytes for sensible
3280 patterns that start (?i) but I don't think that matters. The use of (?i) is
3281 just an example; this all applies to the other options as well.
3282
3283 4. Since libtool seems to echo the compile commands it is issuing, the output
3284 from "make" can be reduced a bit by putting "@" in front of each libtool
3285 compile command.
3286
3287 5. Patch from the folks at Google for configure.in to be a bit more thorough
3288 in checking for a suitable C++ installation before trying to compile the
3289 C++ stuff. This should fix a reported problem when a compiler was present,
3290 but no suitable headers.
3291
3292 6. The man pages all had just "PCRE" as their title. I have changed them to
3293 be the relevant file name. I have also arranged that these names are
3294 retained in the file doc/pcre.txt, which is a concatenation in text format
3295 of all the man pages except the little individual ones for each function.
3296
3297 7. The NON-UNIX-USE file had not been updated for the different set of source
3298 files that come with release 6. I also added a few comments about the C++
3299 wrapper.
3300
3301
3302 Version 6.0 07-Jun-05
3303 ---------------------
3304
3305 1. Some minor internal re-organization to help with my DFA experiments.
3306
3307 2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
3308 didn't matter for the library itself when fully configured, but did matter
3309 when compiling without UCP support, or within Exim, where the ucp files are
3310 not imported.
3311
3312 3. Refactoring of the library code to split up the various functions into
3313 different source modules. The addition of the new DFA matching code (see
3314 below) to a single monolithic source would have made it really too
3315 unwieldy, quite apart from causing all the code to be include in a
3316 statically linked application, when only some functions are used. This is
3317 relevant even without the DFA addition now that patterns can be compiled in
3318 one application and matched in another.
3319
3320 The downside of splitting up is that there have to be some external
3321 functions and data tables that are used internally in different modules of
3322 the library but which are not part of the API. These have all had their
3323 names changed to start with "_pcre_" so that they are unlikely to clash
3324 with other external names.
3325
3326 4. Added an alternate matching function, pcre_dfa_exec(), which matches using
3327 a different (DFA) algorithm. Although it is slower than the original
3328 function, it does have some advantages for certain types of matching
3329 problem.
3330
3331 5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
3332 including restarting after a partial match.
3333
3334 6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
3335 defined when compiling for Windows was sent to me. I have put it into the
3336 code, though I have no means of testing or verifying it.
3337
3338 7. Added the pcre_refcount() auxiliary function.
3339
3340 8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
3341 match before or at the first newline in the subject string. In pcretest,
3342 the /f option on a pattern can be used to set this.
3343
3344 9. A repeated \w when used in UTF-8 mode with characters greater than 256
3345 would behave wrongly. This has been present in PCRE since release 4.0.
3346
3347 10. A number of changes to the pcregrep command:
3348
3349 (a) Refactored how -x works; insert ^(...)$ instead of setting
3350 PCRE_ANCHORED and checking the length, in preparation for adding
3351 something similar for -w.
3352
3353 (b) Added the -w (match as a word) option.
3354
3355 (c) Refactored the way lines are read and buffered so as to have more
3356 than one at a time available.
3357
3358 (d) Implemented a pcregrep test script.
3359
3360 (e) Added the -M (multiline match) option. This allows patterns to match
3361 over several lines of the subject. The buffering ensures that at least
3362 8K, or the rest of the document (whichever is the shorter) is available
3363 for matching (and similarly the previous 8K for lookbehind assertions).
3364
3365 (f) Changed the --help output so that it now says
3366
3367 -w, --word-regex(p)
3368
3369 instead of two lines, one with "regex" and the other with "regexp"
3370 because that confused at least one person since the short forms are the
3371 same. (This required a bit of code, as the output is generated
3372 automatically from a table. It wasn't just a text change.)
3373
3374 (g) -- can be used to terminate pcregrep options if the next thing isn't an
3375 option but starts with a hyphen. Could be a pattern or a path name
3376 starting with a hyphen, for instance.
3377
3378 (h) "-" can be given as a file name to represent stdin.
3379
3380 (i) When file names are being printed, "(standard input)" is used for
3381 the standard input, for compatibility with GNU grep. Previously
3382 "<stdin>" was used.
3383
3384 (j) The option --label=xxx can be used to supply a name to be used for
3385 stdin when file names are being printed. There is no short form.
3386
3387 (k) Re-factored the options decoding logic because we are going to add
3388 two more options that take data. Such options can now be given in four
3389 different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
3390
3391 (l) Added the -A, -B, and -C options for requesting that lines of context
3392 around matches be printed.
3393
3394 (m) Added the -L option to print the names of files that do not contain
3395 any matching lines, that is, the complement of -l.
3396
3397 (n) The return code is 2 if any file cannot be opened, but pcregrep does
3398 continue to scan other files.
3399
3400 (o) The -s option was incorrectly implemented. For compatibility with other
3401 greps, it now suppresses the error message for a non-existent or non-
3402 accessible file (but not the return code). There is a new option called
3403 -q that suppresses the output of matching lines, which was what -s was
3404 previously doing.
3405
3406 (p) Added --include and --exclude options to specify files for inclusion
3407 and exclusion when recursing.
3408
3409 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
3410 Hopefully, it now does.
3411
3412 12. Missing cast in pcre_study().
3413
3414 13. Added an "uninstall" target to the makefile.
3415
3416 14. Replaced "extern" in the function prototypes in Makefile.in with
3417 "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
3418 world, but is set differently for Windows.
3419
3420 15. Added a second compiling function called pcre_compile2(). The only
3421 difference is that it has an extra argument, which is a pointer to an
3422 integer error code. When there is a compile-time failure, this is set
3423 non-zero, in addition to the error test pointer being set to point to an
3424 error message. The new argument may be NULL if no error number is required
3425 (but then you may as well call pcre_compile(), which is now just a
3426 wrapper). This facility is provided because some applications need a
3427 numeric error indication, but it has also enabled me to tidy up the way
3428 compile-time errors are handled in the POSIX wrapper.
3429
3430 16. Added VPATH=.libs to the makefile; this should help when building with one
3431 prefix path and installing with another. (Or so I'm told by someone who
3432 knows more about this stuff than I do.)
3433
3434 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
3435 passes PCRE_DOTALL to the pcre_compile() function, making the "." character
3436 match everything, including newlines. This is not POSIX-compatible, but
3437 somebody wanted the feature. From pcretest it can be activated by using
3438 both the P and the s flags.
3439
3440 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
3441
3442 19. libpcre.pc was being incorrectly installed as executable.
3443
3444 20. A couple of places in pcretest check for end-of-line by looking for '\n';
3445 it now also looks for '\r' so that it will work unmodified on Windows.
3446
3447 21. Added Google's contributed C++ wrapper to the distribution.
3448
3449 22. Added some untidy missing memory free() calls in pcretest, to keep
3450 Electric Fence happy when testing.
3451
3452
3453
3454 Version 5.0 13-Sep-04
3455 ---------------------
3456
3457 1. Internal change: literal characters are no longer packed up into items
3458 containing multiple characters in a single byte-string. Each character
3459 is now matched using a separate opcode. However, there may be more than one
3460 byte in the character in UTF-8 mode.
3461
3462 2. The pcre_callout_block structure has two new fields: pattern_position and
3463 next_item_length. These contain the offset in the pattern to the next match
3464 item, and its length, respectively.
3465
3466 3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
3467 insertion of callouts before each pattern item. Added the /C option to
3468 pcretest to make use of this.
3469
3470 4. On the advice of a Windows user, the lines
3471
3472 #if defined(_WIN32) || defined(WIN32)
3473 _setmode( _fileno( stdout ), 0x8000 );
3474 #endif /* defined(_WIN32) || defined(WIN32) */
3475
3476 have been added to the source of pcretest. This apparently does useful
3477 magic in relation to line terminators.
3478
3479 5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
3480 for the benefit of those environments where the "b" makes a difference.
3481
3482 6. The icc compiler has the same options as gcc, but "configure" doesn't seem
3483 to know about it. I have put a hack into configure.in that adds in code
3484 to set GCC=yes if CC=icc. This seems to end up at a point in the
3485 generated configure script that is early enough to affect the setting of
3486 compiler options, which is what is needed, but I have no means of testing
3487 whether it really works. (The user who reported this had patched the
3488 generated configure script, which of course I cannot do.)
3489
3490 LATER: After change 22 below (new libtool files), the configure script
3491 seems to know about icc (and also ecc). Therefore, I have commented out
3492 this hack in configure.in.
3493
3494 7. Added support for pkg-config (2 patches were sent in).
3495
3496 8. Negated POSIX character classes that used a combination of internal tables
3497 were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
3498 [[:^ascii]]. Typically, they would match almost any characters. The other
3499 POSIX classes were not broken in this way.
3500
3501 9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
3502 to find the match, as PCRE was deluded into thinking that the match had to
3503 start at the start point or following a newline. The same bug applied to
3504 patterns with negative forward assertions or any backward assertions
3505 preceding ".*" at the start, unless the pattern required a fixed first
3506 character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
3507
3508 10. In UTF-8 mode, when moving forwards in the subject after a failed match
3509 starting at the last subject character, bytes beyond the end of the subject
3510 string were read.
3511
3512 11. Renamed the variable "class" as "classbits" to make life easier for C++
3513 users. (Previously there was a macro definition, but it apparently wasn't
3514 enough.)
3515
3516 12. Added the new field "tables" to the extra data so that tables can be passed
3517 in at exec time, or the internal tables can be re-selected. This allows
3518 a compiled regex to be saved and re-used at a later time by a different
3519 program that might have everything at different addresses.
3520
3521 13. Modified the pcre-config script so that, when run on Solaris, it shows a
3522 -R library as well as a -L library.
3523
3524 14. The debugging options of pcretest (-d on the command line or D on a
3525 pattern) showed incorrect output for anything following an extended class
3526 that contained multibyte characters and which was followed by a quantifier.
3527
3528 15. Added optional support for general category Unicode character properties
3529 via the \p, \P, and \X escapes. Unicode property support implies UTF-8
3530 support. It adds about 90K to the size of the library. The meanings of the
3531 inbuilt class escapes such as \d and \s have NOT been changed.
3532
3533 16. Updated pcredemo.c to include calls to free() to release the memory for the
3534 compiled pattern.
3535
3536 17. The generated file chartables.c was being created in the source directory
3537 instead of in the building directory. This caused the build to fail if the
3538 source directory was different from the building directory, and was
3539 read-only.
3540
3541 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
3542 file. No doubt somebody will tell me if they don't make sense... Also added
3543 Dan Mooney's comments about building on OpenVMS.
3544
3545 19. Added support for partial matching via the PCRE_PARTIAL option for
3546 pcre_exec() and the \P data escape in pcretest.
3547
3548 20. Extended pcretest with 3 new pattern features:
3549
3550 (i) A pattern option of the form ">rest-of-line" causes pcretest to
3551 write the compiled pattern to the file whose name is "rest-of-line".
3552 This is a straight binary dump of the data, with the saved pointer to
3553 the character tables forced to be NULL. The study data, if any, is
3554 written too. After writing, pcretest reads a new pattern.
3555
3556 (ii) If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
3557 compiled pattern from the given file. There must not be any
3558 occurrences of "<" in the file name (pretty unlikely); if there are,
3559 pcretest will instead treat the initial "<" as a pattern delimiter.
3560 After reading in the pattern, pcretest goes on to read data lines as
3561 usual.
3562
3563 (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
3564 and 16-bit fields in a compiled pattern, to simulate a pattern that
3565 was compiled on a host of opposite endianness.
3566
3567 21. The pcre-exec() function can now cope with patterns that were compiled on
3568 hosts of opposite endianness, with this restriction:
3569
3570 As for any compiled expression that is saved and used later, the tables
3571 pointer field cannot be preserved; the extra_data field in the arguments
3572 to pcre_exec() should be used to pass in a tables address if a value
3573 other than the default internal tables were used at compile time.
3574
3575 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
3576 now diagnosed as an error. Previously, most of the time, a negative number
3577 would have been treated as zero, but if in addition "ovector" was passed as
3578 NULL, a crash could occur.
3579
3580 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
3581 new versions from the libtool 1.5 distribution (the last one is a copy of
3582 a file called libtool.m4). This seems to have fixed the need to patch
3583 "configure" to support Darwin 1.3 (which I used to do). However, I still
3584 had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
3585 workstation).
3586
3587 24. Changed the PCRE licence to be the more standard "BSD" licence.
3588
3589
3590 Version 4.5 01-Dec-03
3591 ---------------------
3592
3593 1. There has been some re-arrangement of the code for the match() function so
3594 that it can be compiled in a version that does not call itself recursively.
3595 Instead, it keeps those local variables that need separate instances for
3596 each "recursion" in a frame on the heap, and gets/frees frames whenever it
3597 needs to "recurse". Keeping track of where control must go is done by means
3598 of setjmp/longjmp. The whole thing is implemented by a set of macros that
3599 hide most of the details from the main code, and operates only if
3600 NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
3601 "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
3602 operating.
3603
3604 To make it easier for callers to provide specially tailored get/free
3605 functions for this usage, two new functions, pcre_stack_malloc, and
3606 pcre_stack_free, are used. They are always called in strict stacking order,
3607 and the size of block requested is always the same.
3608
3609 The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
3610 PCRE has been compiled to use the stack or the heap for recursion. The
3611 -C option of pcretest uses this to show which version is compiled.
3612
3613 A new data escape \S, is added to pcretest; it causes the amounts of store
3614 obtained and freed by both kinds of malloc/free at match time to be added
3615 to the output.
3616
3617 2. Changed the locale test to use "fr_FR" instead of "fr" because that's
3618 what's available on my current Linux desktop machine.
3619
3620 3. When matching a UTF-8 string, the test for a valid string at the start has
3621 been extended. If start_offset is not zero, PCRE now checks that it points
3622 to a byte that is the start of a UTF-8 character. If not, it returns
3623 PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
3624 this is necessary because there may be backward assertions in the pattern.
3625 When matching the same subject several times, it may save resources to use
3626 PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
3627
3628 4. The code for checking the validity of UTF-8 strings has been tightened so
3629 that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
3630 containing "overlong sequences".
3631
3632 5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
3633 I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
3634 should have been "&", but it just so happened that all the cases this let
3635 through by mistake were picked up later in the function.
3636
3637 6. I had used a variable called "isblank" - this is a C99 function, causing
3638 some compilers to warn. To avoid this, I renamed it (as "blankclass").
3639
3640 7. Cosmetic: (a) only output another newline at the end of pcretest if it is
3641 prompting; (b) run "./pcretest /dev/null" at the start of the test script
3642 so the version is shown; (c) stop "make test" echoing "./RunTest".
3643
3644 8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
3645
3646 9. The prototype for memmove() for systems that don't have it was using
3647 size_t, but the inclusion of the header that defines size_t was later. I've
3648 moved the #includes for the C headers earlier to avoid this.
3649
3650 10. Added some adjustments to the code to make it easier to compiler on certain
3651 special systems:
3652
3653 (a) Some "const" qualifiers were missing.
3654 (b) Added the macro EXPORT before all exported functions; by default this
3655 is defined to be empty.
3656 (c) Changed the dftables auxiliary program (that builds chartables.c) so
3657 that it reads its output file name as an argument instead of writing
3658 to the standard output and assuming this can be redirected.
3659
3660 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
3661 class containing characters with values greater than 255, PCRE compilation
3662 went into a loop.
3663
3664 12. A recursive reference to a subpattern that was within another subpattern
3665 that had a minimum quantifier of zero caused PCRE to crash. For example,
3666 (x(y(?2))z)? provoked this bug with a subject that got as far as the
3667 recursion. If the recursively-called subpattern itself had a zero repeat,
3668 that was OK.
3669
3670 13. In pcretest, the buffer for reading a data line was set at 30K, but the
3671 buffer into which it was copied (for escape processing) was still set at
3672 1024, so long lines caused crashes.
3673
3674 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
3675 "internal error: code overflow...". This applied to any character class
3676 that was followed by a possessive quantifier.
3677
3678 15. Modified the Makefile to add libpcre.la as a prerequisite for
3679 libpcreposix.la because I was told this is needed for a parallel build to
3680 work.
3681
3682 16. If a pattern that contained .* following optional items at the start was
3683 studied, the wrong optimizing data was generated, leading to matching
3684 errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
3685 matching string must start with a or b or c. The correct conclusion for
3686 this pattern is that a match can start with any character.
3687
3688
3689 Version 4.4 13-Aug-03
3690 ---------------------
3691
3692 1. In UTF-8 mode, a character class containing characters with values between
3693 127 and 255 was not handled correctly if the compiled pattern was studied.
3694 In fixing this, I have also improved the studying algorithm for such
3695 classes (slightly).
3696
3697 2. Three internal functions had redundant arguments passed to them. Removal
3698 might give a very teeny performance improvement.
3699
3700 3. Documentation bug: the value of the capture_top field in a callout is *one
3701 more than* the number of the hightest numbered captured substring.
3702
3703 4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
3704 in incorrectly linking with a previously installed version. They now link
3705 explicitly with libpcre.la.
3706
3707 5. configure.in no longer needs to recognize Cygwin specially.
3708
3709 6. A problem in pcre.in for Windows platforms is fixed.
3710
3711 7. If a pattern was successfully studied, and the -d (or /D) flag was given to
3712 pcretest, it used to include the size of the study block as part of its
3713 output. Unfortunately, the structure contains a field that has a different
3714 size on different hardware architectures. This meant that the tests that
3715 showed this size failed. As the block is currently always of a fixed size,
3716 this information isn't actually particularly useful in pcretest output, so
3717 I have just removed it.
3718
3719 8. Three pre-processor statements accidentally did not start in column 1.
3720 Sadly, there are *still* compilers around that complain, even though
3721 standard C has not required this for well over a decade. Sigh.
3722
3723 9. In pcretest, the code for checking callouts passed small integers in the
3724 callout_data field, which is a void * field. However, some picky compilers
3725 complained about the casts involved for this on 64-bit systems. Now
3726 pcretest passes the address of the small integer instead, which should get
3727 rid of the warnings.
3728
3729 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
3730 both compile and run time, and gives an error if an invalid UTF-8 sequence
3731 is found. There is a option for disabling this check in cases where the
3732 string is known to be correct and/or the maximum performance is wanted.
3733
3734 11. In response to a bug report, I changed one line in Makefile.in from
3735
3736 -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
3737 to
3738 -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
3739
3740 to look similar to other lines, but I have no way of telling whether this
3741 is the right thing to do, as I do not use Windows. No doubt I'll get told
3742 if it's wrong...
3743
3744
3745 Version 4.3 21-May-03
3746 ---------------------
3747
3748 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
3749 Makefile.
3750
3751 2. Some refactoring to improve the quality of the code:
3752
3753 (i) The utf8_table... variables are now declared "const".
3754
3755 (ii) The code for \cx, which used the "case flipping" table to upper case
3756 lower case letters, now just substracts 32. This is ASCII-specific,
3757 but the whole concept of \cx is ASCII-specific, so it seems
3758 reasonable.
3759
3760 (iii) PCRE was using its character types table to recognize decimal and
3761 hexadecimal digits in the pattern. This is silly, because it handles
3762 only 0-9, a-f, and A-F, but the character types table is locale-
3763 specific, which means strange things might happen. A private
3764 table is now used for this - though it costs 256 bytes, a table is
3765 much faster than multiple explicit tests. Of course, the standard
3766 character types table is still used for matching digits in subject
3767 strings against \d.
3768
3769 (iv) Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
3770 ending in _t are). So I've renamed it as ESC_tee.
3771
3772 3. The first argument for regexec() in the POSIX wrapper should have been
3773 defined as "const".
3774
3775 4. Changed pcretest to use malloc() for its buffers so that they can be
3776 Electric Fenced for debugging.
3777
3778 5. There were several places in the code where, in UTF-8 mode, PCRE would try
3779 to read one or more bytes before the start of the subject string. Often this
3780 had no effect on PCRE's behaviour, but in some circumstances it could
3781 provoke a segmentation fault.
3782
3783 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
3784 to try to read one or more bytes before the start of the subject string.
3785
3786 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
3787 UTF-8 support could misbehave in various ways if the subject string
3788 contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
3789 area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
3790 back over UTF-8 characters.)
3791
3792
3793 Version 4.2 14-Apr-03
3794 ---------------------
3795
3796 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
3797
3798 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
3799 [ON_WINDOWS]: new variable, "#" on non-Windows platforms
3800 [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
3801 [WIN_PREFIX]: new variable, "cyg" for Cygwin
3802 * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
3803 and BUILD_EXEEXT
3804 Note: automatic setting of the BUILD variables is not yet working
3805 set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
3806 compile-time but not at link-time
3807 [LINK]: use for linking executables only
3808 make different versions for Windows and non-Windows
3809 [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
3810 libraries
3811 [LINK_FOR_BUILD]: new variable
3812 [OBJEXT]: use throughout
3813 [EXEEXT]: use throughout
3814 <winshared>: new target
3815 <wininstall>: new target
3816 <dftables.o>: use native compiler
3817 <dftables>: use native linker
3818 <install>: handle Windows platform correctly
3819 <clean>: ditto
3820 <check>: ditto
3821 copy DLL to top builddir before testing
3822
3823 As part of these changes, -no-undefined was removed again. This was reported
3824 to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
3825 in any case.
3826
3827 3. Some tidies to get rid of compiler warnings:
3828
3829 . In the match_data structure, match_limit was an unsigned long int, whereas
3830 match_call_count was an int. I've made them both unsigned long ints.
3831
3832 . In pcretest the fact that a const uschar * doesn't automatically cast to
3833 a void * provoked a warning.
3834
3835 . Turning on some more compiler warnings threw up some "shadow" variables
3836 and a few more missing casts.
3837
3838 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
3839 option, a class that contained a single character with a value between 128
3840 and 255 (e.g. /[\xFF]/) caused PCRE to crash.
3841
3842 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
3843 option, a class that contained several characters, but with at least one
3844 whose value was between 128 and 255 caused PCRE to crash.
3845
3846
3847 Version 4.1 12-Mar-03
3848 ---------------------
3849
3850 1. Compiling with gcc -pedantic found a couple of places where casts were
3851 needed, and a string in dftables.c that was longer than standard compilers are
3852 required to support.
3853
3854 2. Compiling with Sun's compiler found a few more places where the code could
3855 be tidied up in order to avoid warnings.
3856
3857 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
3858 first of these names is deprecated in the latest Autoconf in favour of the name
3859 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
3860 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
3861 analogy I have changed it to CFLAGS_FOR_BUILD.
3862
3863 4. Added -no-undefined to the linking command in the Makefile, because this is
3864 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
3865 linking step for the pcreposix library.
3866
3867 5. PCRE was failing to diagnose the case of two named groups with the same
3868 name.
3869
3870 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
3871 literal character that is needed in the subject for a match, and scans along to
3872 ensure that it is present before embarking on the full matching process. This
3873 saves time in cases of nested unlimited repeats that are never going to match.
3874 Problem: the scan can take a lot of time if the subject is very long (e.g.
3875 megabytes), thus penalizing straightforward matches. It is now done only if the
3876 amount of subject to be scanned is less than 1000 bytes.
3877
3878 7. A lesser problem with the same optimization is that it was recording the
3879 first character of an anchored pattern as "needed", thus provoking a search
3880 right along the subject, even when the first match of the pattern was going to
3881 fail. The "needed" character is now not set for anchored patterns, unless it
3882 follows something in the pattern that is of non-fixed length. Thus, it still
3883 fulfils its original purpose of finding quick non-matches in cases of nested
3884 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
3885
3886
3887 Version 4.0 17-Feb-03
3888 ---------------------
3889
3890 1. If a comment in an extended regex that started immediately after a meta-item
3891 extended to the end of string, PCRE compiled incorrect data. This could lead to
3892 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
3893
3894 2. Moved to autoconf 2.53 and libtool 1.4.2.
3895
3896 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
3897 the special perltest8 script is no longer needed - all the tests can be run
3898 from a single perltest script.
3899
3900 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
3901 by \s. It has now been removed in PCRE. This means it isn't recognized as
3902 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
3903 class [:space:] *does* include VT, thereby creating a mess.
3904
3905 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
3906 space and tab.
3907
3908 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
3909 its new features into the main test script, reducing the number of scripts.
3910
3911 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
3912 were backward compatible, and made the (?i) apply to the whole pattern, as if
3913 /i were given. Now it behaves more logically, and applies the option setting
3914 only to what follows. PCRE has been changed to follow suit. However, if it
3915 finds options settings right at the start of the pattern, it extracts them into
3916 the global options, as before. Thus, they show up in the info data.
3917
3918 8. Added support for the \Q...\E escape sequence. Characters in between are
3919 treated as literals. This is slightly different from Perl in that $ and @ are
3920 also handled as literals inside the quotes. In Perl, they will cause variable
3921 interpolation. Note the following examples:
3922
3923 Pattern PCRE matches Perl matches
3924
3925 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
3926 \Qabc\$xyz\E abc\$xyz abc\$xyz
3927 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
3928
3929 For compatibility with Perl, \Q...\E sequences are recognized inside character
3930 classes as well as outside them.
3931
3932 9. Re-organized 3 code statements in pcretest to avoid "overflow in
3933 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
3934 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
3935 signed/unsigned warnings.
3936
3937 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
3938 option for pcretest, so I've replaced it by a simple function that does just
3939 that job.
3940
3941 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
3942 "pcregrep -".
3943
3944 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
3945 Java package. This provides some syntactic sugar for simple cases of what my
3946 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
3947 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
3948 item, you can use this simplified notation. Note that only makes sense with
3949 greedy quantifiers. Consequently, the use of the possessive quantifier forces
3950 greediness, whatever the setting of the PCRE_UNGREEDY option.
3951
3952 13. A change of greediness default within a pattern was not taking effect at
3953 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
3954 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
3955 was abstracted outside.
3956
3957 14. PCRE now supports the \G assertion. It is true when the current matching
3958 position is at the start point of the match. This differs from \A when the
3959 starting offset is non-zero. Used with the /g option of pcretest (or similar
3960 code), it works in the same way as it does for Perl's /g option. If all
3961 alternatives of a regex begin with \G, the expression is anchored to the start
3962 match position, and the "anchored" flag is set in the compiled expression.
3963
3964 15. Some bugs concerning the handling of certain option changes within patterns
3965 have been fixed. These applied to options other than (?ims). For example,
3966 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
3967 been the other way round. Some of this was related to change 7 above.
3968
3969 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
3970 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
3971 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
3972 POSIX classes only within a class (e.g. /[[:alpha:]]/).
3973
3974 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
3975 mode. Unlike ".", it always matches newline, whatever the setting of
3976 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
3977 assertions. Perl allows it, but it doesn't (in general) work because it can't
3978 calculate the length of the lookbehind. At least, that's the case for Perl
3979 5.8.0 - I've been told they are going to document that it doesn't work in
3980 future.
3981
3982 18. Added an error diagnosis for escapes that PCRE does not support: these are
3983 \L, \l, \N, \P, \p, \U, \u, and \X.
3984
3985 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
3986 reading past the end of the pattern in cases such as /[abcd/.
3987
3988 20. PCRE was getting more memory than necessary for patterns with classes that
3989 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
3990
3991 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
3992 compiling PCRE for use with Virtual Pascal.
3993
3994 22. Small fix to the Makefile to make it work properly if the build is done
3995 outside the source tree.
3996
3997 23. Added a new extension: a condition to go with recursion. If a conditional
3998 subpattern starts with (?(R) the "true" branch is used if recursion has
3999 happened, whereas the "false" branch is used only at the top level.
4000
4001 24. When there was a very long string of literal characters (over 255 bytes
4002 without UTF support, over 250 bytes with UTF support), the computation of how
4003 much memory was required could be incorrect, leading to segfaults or other
4004 strange effects.
4005
4006 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
4007 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
4008 there was a subsequent back reference to those brackets. This meant that, for
4009 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
4010 possible to check for precisely this case. All we can do is abandon the
4011 optimization if .* occurs inside capturing brackets when there are any back
4012 references whatsoever. (See below for a better fix that came later.)
4013
4014 26. The handling of the optimization for finding the first character of a
4015 non-anchored pattern, and for finding a character that is required later in the
4016 match were failing in some cases. This didn't break the matching; it just
4017 failed to optimize when it could. The way this is done has been re-implemented.
4018
4019 27. Fixed typo in error message for invalid (?R item (it said "(?p").
4020
4021 28. Added a new feature that provides some of the functionality that Perl
4022 provides with (?{...}). The facility is termed a "callout". The way it is done
4023 in PCRE is for the caller to provide an optional function, by setting
4024 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
4025 global variable. By default it is unset, which disables all calling out. To get
4026 the function called, the regex must include (?C) at appropriate points. This
4027 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
4028 This provides a means of identifying different callout points. When PCRE
4029 reaches such a point in the regex, if pcre_callout has been set, the external
4030 function is called. It is provided with data in a structure called
4031 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
4032 matching continues; if it returns a non-zero value, the match at the current
4033 point fails. However, backtracking will occur if possible. [This was changed
4034 later and other features added - see item 49 below.]
4035
4036 29. pcretest is upgraded to test the callout functionality. It provides a
4037 callout function that displays information. By default, it shows the start of
4038 the match and the current position in the text. There are some new data escapes
4039 to vary what happens:
4040
4041 \C+ in addition, show current contents of captured substrings
4042 \C- do not supply a callout function
4043 \C!n return 1 when callout number n is reached
4044 \C!n!m return 1 when callout number n is reached for the mth time
4045
4046 30. If pcregrep was called with the -l option and just a single file name, it
4047 output "<stdin>" if a match was found, instead of the file name.
4048
4049 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
4050 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
4051 pcre_exec(). This saves a malloc/free per call. The default value of
4052 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
4053 when configuring.
4054
4055 32. The default maximum size of a compiled pattern is 64K. There have been a
4056 few cases of people hitting this limit. The code now uses macros to handle the
4057 storing of links as offsets within the compiled pattern. It defaults to 2-byte
4058 links, but this can be changed to 3 or 4 bytes by --with-link-size when
4059 configuring. Tests 2 and 5 work only with 2-byte links because they output
4060 debugging information about compiled patterns.
4061
4062 33. Internal code re-arrangements:
4063
4064 (a) Moved the debugging function for printing out a compiled regex into
4065 its own source file (printint.c) and used #include to pull it into
4066 pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
4067 separate copies.
4068
4069 (b) Defined the list of op-code names for debugging as a macro in
4070 internal.h so that it is next to the definition of the opcodes.
4071
4072 (c) Defined a table of op-code lengths for simpler skipping along compiled
4073 code. This is again a macro in internal.h so that it is next to the
4074 definition of the opcodes.
4075
4076 34. Added support for recursive calls to individual subpatterns, along the
4077 lines of Robin Houston's patch (but implemented somewhat differently).
4078
4079 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
4080 allow it to read and process whole directories in Win32. This code was
4081 contributed by Lionel Fourquaux; it has not been tested by me.
4082
4083 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
4084 used to name a group. Names consist of alphanumerics and underscores, and must
4085 be unique. Back references use the syntax (?P=name) and recursive calls use
4086 (?P>name) which is a PCRE extension to the Python extension. Groups still have
4087 numbers. The function pcre_fullinfo() can be used after compilation to extract
4088 a name/number map. There are three relevant calls:
4089
4090 PCRE_INFO_NAMEENTRYSIZE yields the size of each entry in the map
4091 PCRE_INFO_NAMECOUNT yields the number of entries
4092 PCRE_INFO_NAMETABLE yields a pointer to the map.
4093
4094 The map is a vector of fixed-size entries. The size of each entry depends on
4095 the length of the longest name used. The first two bytes of each entry are the
4096 group number, most significant byte first. There follows the corresponding
4097 name, zero terminated. The names are in alphabetical order.
4098
4099 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
4100 case instead of 255. Making it the same both with and without UTF-8 support
4101 means that the same test output works with both.
4102
4103 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
4104 calling malloc() with a zero argument.
4105
4106 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
4107 optimization. I've improved things by keeping a bitmap of backreferences with
4108 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
4109 fact referenced, the optimization can be applied. It is unlikely that a
4110 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
4111 the match to follow \n) will appear inside brackets with a number greater than
4112 31, but if it does, any back reference > 31 suppresses the optimization.
4113
4114 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
4115 of disabling numbered capturing parentheses. Any opening parenthesis that is
4116 not followed by ? behaves as if it were followed by ?: but named parentheses
4117 can still be used for capturing (and they will acquire numbers in the usual
4118 way).
4119
4120 41. Redesigned the return codes from the match() function into yes/no/error so
4121 that errors can be passed back from deep inside the nested calls. A malloc
4122 failure while inside a recursive subpattern call now causes the
4123 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
4124
4125 42. It is now possible to set a limit on the number of times the match()
4126 function is called in a call to pcre_exec(). This facility makes it possible to
4127 limit the amount of recursion and backtracking, though not in a directly
4128 obvious way, because the match() function is used in a number of different
4129 circumstances. The count starts from zero for each position in the subject
4130 string (for non-anchored patterns). The default limit is, for compatibility, a
4131 large number, namely 10 000 000. You can change this in two ways:
4132
4133 (a) When configuring PCRE before making, you can use --with-match-limit=n
4134 to set a default value for the compiled library.
4135
4136 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
4137 a different value is set. See 45 below.
4138
4139 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
4140
4141 43. Added a new function pcre_config(int, void *) to enable run-time extraction
4142 of things that can be changed at compile time. The first argument specifies
4143 what is wanted and the second points to where the information is to be placed.
4144 The current list of available information is:
4145
4146 PCRE_CONFIG_UTF8
4147
4148 The output is an integer that is set to one if UTF-8 support is available;
4149 otherwise it is set to zero.
4150
4151 PCRE_CONFIG_NEWLINE
4152
4153 The output is an integer that it set to the value of the code that is used for
4154 newline. It is either LF (10) or CR (13).
4155
4156 PCRE_CONFIG_LINK_SIZE
4157
4158 The output is an integer that contains the number of bytes used for internal
4159 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
4160
4161 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
4162
4163 The output is an integer that contains the threshold above which the POSIX
4164 interface uses malloc() for output vectors. See item 31 above.
4165
4166 PCRE_CONFIG_MATCH_LIMIT
4167
4168 The output is an unsigned integer that contains the default limit of the number
4169 of match() calls in a pcre_exec() execution. See 42 above.
4170
4171 44. pcretest has been upgraded by the addition of the -C option. This causes it
4172 to extract all the available output from the new pcre_config() function, and to
4173 output it. The program then exits immediately.
4174
4175 45. A need has arisen to pass over additional data with calls to pcre_exec() in
4176 order to support additional features. One way would have been to define
4177 pcre_exec2() (for example) with extra arguments, but this would not have been
4178 extensible, and would also have required all calls to the original function to
4179 be mapped to the new one. Instead, I have chosen to extend the mechanism that
4180 is used for passing in "extra" data from pcre_study().
4181
4182 The pcre_extra structure is now exposed and defined in pcre.h. It currently
4183 contains the following fields:
4184
4185 flags a bitmap indicating which of the following fields are set
4186 study_data opaque data from pcre_study()
4187 match_limit a way of specifying a limit on match() calls for a specific
4188 call to pcre_exec()
4189 callout_data data for callouts (see 49 below)
4190
4191 The flag bits are also defined in pcre.h, and are
4192
4193 PCRE_EXTRA_STUDY_DATA
4194 PCRE_EXTRA_MATCH_LIMIT
4195 PCRE_EXTRA_CALLOUT_DATA
4196
4197 The pcre_study() function now returns one of these new pcre_extra blocks, with
4198 the actual study data pointed to by the study_data field, and the
4199 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
4200 before. That is, this change is entirely upwards-compatible and requires no
4201 change to existing code.
4202
4203 If you want to pass in additional data to pcre_exec(), you can either place it
4204 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
4205 block.
4206
4207 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
4208 data string contains the escape sequence \M, pcretest calls pcre_exec() several
4209 times with different match limits, until it finds the minimum value needed for
4210 pcre_exec() to complete. The value is then output. This can be instructive; for
4211 most simple matches the number is quite small, but for pathological cases it
4212 gets very large very quickly.
4213
4214 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
4215 returns the size of the data block pointed to by the study_data field in a
4216 pcre_extra block, that is, the value that was passed as the argument to
4217 pcre_malloc() when PCRE was getting memory in which to place the information
4218 created by pcre_study(). The fourth argument should point to a size_t variable.
4219 pcretest has been extended so that this information is shown after a successful
4220 pcre_study() call when information about the compiled regex is being displayed.
4221
4222 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
4223 because what follows is always an absolute path. (Later: it turns out that this
4224 is more than cosmetic for MinGW, because it doesn't like empty path
4225 components.)
4226
4227 49. Some changes have been made to the callout feature (see 28 above):
4228
4229 (i) A callout function now has three choices for what it returns:
4230
4231 0 => success, carry on matching
4232 > 0 => failure at this point, but backtrack if possible
4233 < 0 => serious error, return this value from pcre_exec()
4234
4235 Negative values should normally be chosen from the set of PCRE_ERROR_xxx
4236 values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
4237 "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
4238 use by callout functions. It will never be used by PCRE itself.
4239
4240 (ii) The pcre_extra structure (see 45 above) has a void * field called
4241 callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
4242 pcre_callout_block structure has a field of the same name. The contents of
4243 the field passed in the pcre_extra structure are passed to the callout
4244 function in the corresponding field in the callout block. This makes it
4245 easier to use the same callout-containing regex from multiple threads. For
4246 testing, the pcretest program has a new data escape
4247
4248 \C*n pass the number n (may be negative) as callout_data
4249
4250 If the callout function in pcretest receives a non-zero value as
4251 callout_data, it returns that value.
4252
4253 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
4254 there were some redundant $(CFLAGS) in commands that are now specified as
4255 $(LINK), which already includes $(CFLAGS).
4256
4257 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
4258 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
4259 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
4260 one-byte characters throughout. Note that case-insensitive matching applies
4261 only to characters whose values are less than 256. PCRE doesn't support the
4262 notion of cases for higher-valued characters.
4263
4264 (i) A character class whose characters are all within 0-255 is handled as
4265 a bit map, and the map is inverted for negative classes. Previously, a
4266 character > 255 always failed to match such a class; however it should
4267 match if the class was a negative one (e.g. [^ab]). This has been fixed.
4268
4269 (ii) A negated character class with a single character < 255 is coded as
4270 "not this character" (OP_NOT). This wasn't working properly when the test
4271 character was multibyte, either singly or repeated.
4272
4273 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
4274 mode, for example: \x{100}{2,3}.
4275
4276 (iv) The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
4277 singly or repeated) now correctly test multibyte characters. However,
4278 PCRE doesn't recognize any characters with values greater than 255 as
4279 digits, spaces, or word characters. Such characters always match \D, \S,
4280 and \W, and never match \d, \s, or \w.
4281
4282 (v) Classes may now contain characters and character ranges with values
4283 greater than 255. For example: [ab\x{100}-\x{400}].
4284
4285 (vi) pcregrep now has a --utf-8 option (synonym -u) which makes it call
4286 PCRE in UTF-8 mode.
4287
4288 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
4289 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
4290 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
4291 value.)
4292
4293 53. The single man page has become too large. I have therefore split it up into
4294 a number of separate man pages. These also give rise to individual HTML pages;
4295 these are now put in a separate directory, and there is an index.html page that
4296 lists them all. Some hyperlinking between the pages has been installed.
4297
4298 54. Added convenience functions for handling named capturing parentheses.
4299
4300 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
4301 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
4302 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
4303 are faulted.
4304
4305 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
4306 calling configure. These values are used when compiling the dftables.c program
4307 which is run to generate the source of the default character tables. They
4308 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
4309 you will need to set these values.
4310
4311 57. Updated the building process for Windows DLL, as provided by Fred Cox.
4312
4313
4314 Version 3.9 02-Jan-02
4315 ---------------------
4316
4317 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
4318
4319 2. If --disable-static was given, the building process failed when trying to
4320 build pcretest and pcregrep. (For some reason it was using libtool to compile
4321 them, which is not right, as they aren't part of the library.)
4322
4323
4324 Version 3.8 18-Dec-01
4325 ---------------------
4326
4327 1. The experimental UTF-8 code was completely screwed up. It was packing the
4328 bytes in the wrong order. How dumb can you get?
4329
4330
4331 Version 3.7 29-Oct-01
4332 ---------------------
4333
4334 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
4335 This caused pcretest, when used on the test data, to segfault. Unfortunately,
4336 this didn't happen under Solaris 8, where I normally test things.
4337
4338 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
4339 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
4340 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
4341 this fix an hour or so after the initial 3.7 release.)
4342
4343
4344 Version 3.6 23-Oct-01
4345 ---------------------
4346
4347 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
4348 offsets passed as NULL with zero offset count.
4349
4350 2. The config.guess and config.sub files had not been updated when I moved to
4351 the latest autoconf.
4352
4353
4354 Version 3.5 15-Aug-01
4355 ---------------------
4356
4357 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
4358 had been forgotten.
4359
4360 2. By using declared but undefined structures, we can avoid using "void"
4361 definitions in pcre.h while keeping the internal definitions of the structures
4362 private.
4363
4364 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
4365 user point of view, this means that both static and shared libraries are built
4366 by default, but this can be individually controlled. More of the work of
4367 handling this static/shared cases is now inside libtool instead of PCRE's make
4368 file.
4369
4370 4. The pcretest utility is now installed along with pcregrep because it is
4371 useful for users (to test regexs) and by doing this, it automatically gets
4372 relinked by libtool. The documentation has been turned into a man page, so
4373 there are now .1, .txt, and .html versions in /doc.
4374
4375 5. Upgrades to pcregrep:
4376 (i) Added long-form option names like gnu grep.
4377 (ii) Added --help to list all options with an explanatory phrase.
4378 (iii) Added -r, --recursive to recurse into sub-directories.
4379 (iv) Added -f, --file to read patterns from a file.
4380
4381 6. pcre_exec() was referring to its "code" argument before testing that
4382 argument for NULL (and giving an error if it was NULL).
4383
4384 7. Upgraded Makefile.in to allow for compiling in a different directory from
4385 the source directory.
4386
4387 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
4388 options bits, the pointer it was passed was to an int instead of to an unsigned
4389 long int. This mattered only on 64-bit systems.
4390
4391 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
4392 generated) instead of pcre.in, which it its source. Also made the same change
4393 in several of the .c files.
4394
4395 10. A new release of gcc defines printf() as a macro, which broke pcretest
4396 because it had an ifdef in the middle of a string argument for printf(). Fixed
4397 by using separate calls to printf().
4398
4399 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
4400 script, to force use of CR or LF instead of \n in the source. On non-Unix
4401 systems, the value can be set in config.h.
4402
4403 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
4404 absolute limit. Changed the text of the error message to make this clear, and
4405 likewise updated the man page.
4406
4407 13. The limit of 99 on the number of capturing subpatterns has been removed.
4408 The new limit is 65535, which I hope will not be a "real" limit.
4409
4410
4411 Version 3.4 22-Aug-00
4412 ---------------------
4413
4414 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
4415
4416 2. Diagnose condition (?(0) as an error instead of crashing on matching.
4417
4418
4419 Version 3.3 01-Aug-00
4420 ---------------------
4421
4422 1. If an octal character was given, but the value was greater than \377, it
4423 was not getting masked to the least significant bits, as documented. This could
4424 lead to crashes in some systems.
4425
4426 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
4427 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
4428
4429 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
4430 These just pass their arguments on to (pcre_free)(), but they are provided
4431 because some uses of PCRE bind it to non-C systems that can call its functions,
4432 but cannot call free() or pcre_free() directly.
4433
4434 4. Add "make test" as a synonym for "make check". Corrected some comments in
4435 the Makefile.
4436
4437 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
4438 Makefile.
4439
4440 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
4441 command called pgrep for grepping around the active processes.
4442
4443 7. Added the beginnings of support for UTF-8 character strings.
4444
4445 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
4446 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
4447 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
4448 out for the ar command.)
4449
4450
4451 Version 3.2 12-May-00
4452 ---------------------
4453
4454 This is purely a bug fixing release.
4455
4456 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
4457 of ZA. This was just one example of several cases that could provoke this bug,
4458 which was introduced by change 9 of version 2.00. The code for breaking
4459 infinite loops after an iteration that matches an empty string was't working
4460 correctly.
4461
4462 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
4463 when matched against abbab (for example). After matching an empty string, it
4464 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
4465 caused it to match further down the string than it should.
4466
4467 3. The code contained an inclusion of sys/types.h. It isn't clear why this
4468 was there because it doesn't seem to be needed, and it causes trouble on some
4469 systems, as it is not a Standard C header. It has been removed.
4470
4471 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
4472 were reported on the Macintosh. The changes were from
4473
4474 while ((c = *(++ptr)) != 0 && c != '\n');
4475 to
4476 while ((c = *(++ptr)) != 0 && c != '\n') ;
4477
4478 Totally extraordinary, but if that's what it takes...
4479
4480 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
4481 available. Added HAVE_BCOPY and an autoconf test for it; if neither
4482 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
4483 assumes the way PCRE uses memmove() (always moving upwards).
4484
4485 6. PCRE is being used in one environment where strchr() is not available. There
4486 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
4487 faster code anyway.
4488
4489
4490 Version 3.1 09-Feb-00
4491 ---------------------
4492
4493 The only change in this release is the fixing of some bugs in Makefile.in for
4494 the "install" target:
4495
4496 (1) It was failing to install pcreposix.h.
4497
4498 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
4499
4500
4501 Version 3.0 01-Feb-00
4502 ---------------------
4503
4504 1. Add support for the /+ modifier to perltest (to output $` like it does in
4505 pcretest).
4506
4507 2. Add support for the /g modifier to perltest.
4508
4509 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
4510 matches null strings.
4511
4512 4. Fix perltest so that it doesn't do unwanted things when fed an empty
4513 pattern. Perl treats empty patterns specially - it reuses the most recent
4514 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
4515 effect.
4516
4517 5. The POSIX interface was broken in that it was just handing over the POSIX
4518 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
4519 required a bigger vector, with some working space on the end. This means that
4520 the POSIX wrapper now has to get and free some memory, and copy the results.
4521
4522 6. Added some simple autoconf support, placing the test data and the
4523 documentation in separate directories, re-organizing some of the
4524 information files, and making it build pcre-config (a GNU standard). Also added
4525 libtool support for building PCRE as a shared library, which is now the
4526 default.
4527
4528 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
4529 09 are not valid octal constants. Single digits will be used for minor values
4530 less than 10.
4531
4532 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
4533 existing programs that set these in the POSIX interface can use PCRE without
4534 modification.
4535
4536 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
4537 return all that pcre_info() returns, plus additional data. The pcre_info()
4538 function is retained for compatibility, but is considered to be obsolete.
4539
4540 10. Added experimental recursion feature (?R) to handle one common case that
4541 Perl 5.6 will be able to do with (?p{...}).
4542
4543 11. Added support for POSIX character classes like [:alpha:], which Perl is
4544 adopting.
4545
4546
4547 Version 2.08 31-Aug-99
4548 ----------------------
4549
4550 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
4551 trying to match at the startoffset position, but instead was moving forward to
4552 the next newline as if a previous match had failed.
4553
4554 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
4555 and could get into a loop if a null string was matched other than at the start
4556 of the subject.
4557
4558 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
4559 be distinguished at compile time, and for completeness also added PCRE_DATE.
4560
4561 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
4562 in GnuWin32 environments.
4563
4564
4565 Version 2.07 29-Jul-99
4566 ----------------------
4567
4568 1. The documentation is now supplied in plain text form and HTML as well as in
4569 the form of man page sources.
4570
4571 2. C++ compilers don't like assigning (void *) values to other pointer types.
4572 In particular this affects malloc(). Although there is no problem in Standard
4573 C, I've put in casts to keep C++ compilers happy.
4574
4575 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
4576 should be (const char *).
4577
4578 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
4579 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
4580 However, I haven't made this a standard facility. The documentation doesn't
4581 mention it, and the Makefile doesn't support it.
4582
4583 5. The Makefile now contains an "install" target, with editable destinations at
4584 the top of the file. The pcretest program is not installed.
4585
4586 6. pgrep -V now gives the PCRE version number and date.
4587
4588 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
4589 causing the entire string to be ignored, instead of just the last character.
4590
4591 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
4592 non-matching string, it can take a very, very long time, even for strings of
4593 quite modest length, because of the nested recursion. PCRE now does better in
4594 some of these cases. It does this by remembering the last required literal
4595 character in the pattern, and pre-searching the subject to ensure it is present
4596 before running the real match. In other words, it applies a heuristic to detect
4597 some types of certain failure quickly, and in the above example, if presented
4598 with a string that has no trailing " it gives "no match" very quickly.
4599
4600 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
4601 other alternatives are tried instead.
4602
4603
4604 Version 2.06 09-Jun-99
4605 ----------------------
4606
4607 1. Change pcretest's output for amount of store used to show just the code
4608 space, because the remainder (the data block) varies in size between 32-bit and
4609 64-bit systems.
4610
4611 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
4612 start matching at. This allows lookbehinds to work when searching for multiple
4613 occurrences in a string.
4614
4615 3. Added additional options to pcretest for testing multiple occurrences:
4616
4617 /+ outputs the rest of the string that follows a match
4618 /g loops for multiple occurrences, using the new startoffset argument
4619 /G loops for multiple occurrences by passing an incremented pointer
4620
4621 4. PCRE wasn't doing the "first character" optimization for patterns starting
4622 with \b or \B, though it was doing it for other lookbehind assertions. That is,
4623 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
4624 the letter 'x'. On long subject strings, this gives a significant speed-up.
4625
4626
4627 Version 2.05 21-Apr-99
4628 ----------------------
4629
4630 1. Changed the type of magic_number from int to long int so that it works
4631 properly on 16-bit systems.
4632
4633 2. Fixed a bug which caused patterns starting with .* not to work correctly
4634 when the subject string contained newline characters. PCRE was assuming
4635 anchoring for such patterns in all cases, which is not correct because .* will
4636 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
4637 DOTALL is set at top level; otherwise it knows that patterns starting with .*
4638 must be retried after every newline in the subject.
4639
4640
4641 Version 2.04 18-Feb-99
4642 ----------------------
4643
4644 1. For parenthesized subpatterns with repeats whose minimum was zero, the
4645 computation of the store needed to hold the pattern was incorrect (too large).
4646 If such patterns were nested a few deep, this could multiply and become a real
4647 problem.
4648
4649 2. Added /M option to pcretest to show the memory requirement of a specific
4650 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
4651
4652 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
4653 compiled in such a way that the backtracking after subsequent failure was
4654 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
4655 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
4656
4657
4658 Version 2.03 02-Feb-99
4659 ----------------------
4660
4661 1. Fixed typo and small mistake in man page.
4662
4663 2. Added 4th condition (GPL supersedes if conflict) and created separate
4664 LICENCE file containing the conditions.
4665
4666 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
4667 Perl, that is the internal \ allows the delimiter to be included in the
4668 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
4669 the final delimiter, add \ to the end of the pattern (to test the error).
4670
4671 4. Added the convenience functions for extracting substrings after a successful
4672 match. Updated pcretest to make it able to test these functions.
4673
4674
4675 Version 2.02 14-Jan-99
4676 ----------------------
4677
4678 1. Initialized the working variables associated with each extraction so that
4679 their saving and restoring doesn't refer to uninitialized store.
4680
4681 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
4682 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
4683 fix the problem.
4684
4685 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
4686 calls, and wasn't printing the correct value for compiling calls. Increased the
4687 default value of LOOPREPEAT, and the number of significant figures in the
4688 times.
4689
4690 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
4691
4692 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
4693 a building problem on Windows NT with a FAT file system.
4694
4695
4696 Version 2.01 21-Oct-98
4697 ----------------------
4698
4699 1. Changed the API for pcre_compile() to allow for the provision of a pointer
4700 to character tables built by pcre_maketables() in the current locale. If NULL
4701 is passed, the default tables are used.
4702
4703
4704 Version 2.00 24-Sep-98
4705 ----------------------
4706
4707 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
4708 it any more.
4709
4710 2. Allow quantification of (?>) groups, and make it work correctly.
4711
4712 3. The first character computation wasn't working for (?>) groups.
4713
4714 4. Correct the implementation of \Z (it is permitted to match on the \n at the
4715 end of the subject) and add 5.005's \z, which really does match only at the
4716 very end of the subject.
4717
4718 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
4719
4720 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
4721 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
4722 localized options. All options to pcre_study() were also removed.
4723
4724 7. Add other new features from 5.005:
4725
4726 $(?<= positive lookbehind
4727 $(?<! negative lookbehind
4728 (?imsx-imsx) added the unsetting capability
4729 such a setting is global if at outer level; local otherwise
4730 (?imsx-imsx:) non-capturing groups with option setting
4731 (?(cond)re|re) conditional pattern matching
4732
4733 A backreference to itself in a repeated group matches the previous
4734 captured string.
4735
4736 8. General tidying up of studying (both automatic and via "study")
4737 consequential on the addition of new assertions.
4738
4739 9. As in 5.005, unlimited repeated groups that could match an empty substring
4740 are no longer faulted at compile time. Instead, the loop is forcibly broken at
4741 runtime if any iteration does actually match an empty substring.
4742
4743 10. Include the RunTest script in the distribution.
4744
4745 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
4746 discrepancies, some of which were old and were also with respect to 5.004. They
4747 have now been fixed.
4748
4749
4750 Version 1.09 28-Apr-98
4751 ----------------------
4752
4753 1. A negated single character class followed by a quantifier with a minimum
4754 value of one (e.g. [^x]{1,6} ) was not compiled correctly. This could lead to
4755 program crashes, or just wrong answers. This did not apply to negated classes
4756 containing more than one character, or to minima other than one.
4757
4758
4759 Version 1.08 27-Mar-98
4760 ----------------------
4761
4762 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
4763
4764 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
4765 latter must appear before anything that relies on it in the pattern.
4766
4767
4768 Version 1.07 16-Feb-98
4769 ----------------------
4770
4771 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
4772 repeat of a potentially empty string).
4773
4774
4775 Version 1.06 23-Jan-98
4776 ----------------------
4777
4778 1. Added Markus Oberhumer's little patches for C++.
4779
4780 2. Literal strings longer than 255 characters were broken.
4781
4782
4783 Version 1.05 23-Dec-97
4784 ----------------------
4785
4786 1. Negated character classes containing more than one character were failing if
4787 PCRE_CASELESS was set at run time.
4788
4789
4790 Version 1.04 19-Dec-97
4791 ----------------------
4792
4793 1. Corrected the man page, where some "const" qualifiers had been omitted.
4794
4795 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
4796 input syntax.
4797
4798 3. Fixed memory leak which occurred when a regex with back references was
4799 matched with an offsets vector that wasn't big enough. The temporary memory
4800 that is used in this case wasn't being freed if the match failed.
4801
4802 4. Tidied pcretest to ensure it frees memory that it gets.
4803
4804 5. Temporary memory was being obtained in the case where the passed offsets
4805 vector was exactly big enough.
4806
4807 6. Corrected definition of offsetof() from change 5 below.
4808
4809 7. I had screwed up change 6 below and broken the rules for the use of
4810 setjmp(). Now fixed.
4811
4812
4813 Version 1.03 18-Dec-97
4814 ----------------------
4815
4816 1. A erroneous regex with a missing opening parenthesis was correctly
4817 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
4818 on some systems.
4819
4820 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
4821 it was reported that one broken compiler failed on the former because "code" is
4822 also an independent variable.
4823
4824 3. The erroneous regex a[]b caused an array overrun reference.
4825
4826 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
4827 fail on data ending with that character. (It was going on too far, and checking
4828 the next character, typically a binary zero.) This was specific to the
4829 optimized code for single-character negative classes.
4830
4831 5. Added a contributed patch from the TIN world which does the following:
4832
4833 + Add an undef for memmove, in case the the system defines a macro for it.
4834
4835 + Add a definition of offsetof(), in case there isn't one. (I don't know
4836 the reason behind this - offsetof() is part of the ANSI standard - but
4837 it does no harm).
4838
4839 + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
4840 most of the places where whitespace preceded '#'. I have given up and
4841 allowed the remaining 2 cases to be at the margin.
4842
4843 + Rename some variables in pcre to eliminate shadowing. This seems very
4844 pedantic, but does no harm, of course.
4845
4846 6. Moved the call to setjmp() into its own function, to get rid of warnings
4847 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
4848
4849 7. Constructs such as \d{8,} were compiling into the equivalent of
4850 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
4851 outcome, but in this particular case used more store than had been allocated,
4852 which caused the bug to be discovered because it threw up an internal error.
4853
4854 8. The debugging code in both pcre and pcretest for outputting the compiled
4855 form of a regex was going wrong in the case of back references followed by
4856 curly-bracketed repeats.
4857
4858
4859 Version 1.02 12-Dec-97
4860 ----------------------
4861
4862 1. Typos in pcre.3 and comments in the source fixed.
4863
4864 2. Applied a contributed patch to get rid of places where it used to remove
4865 'const' from variables, and fixed some signed/unsigned and uninitialized
4866 variable warnings.
4867
4868 3. Added the "runtest" target to Makefile.
4869
4870 4. Set default compiler flag to -O2 rather than just -O.
4871
4872
4873 Version 1.01 19-Nov-97
4874 ----------------------
4875
4876 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
4877 like /([ab]*)*/, that is, for classes with more than one character in them.
4878
4879 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
4880 as /((?>a*))*/ (a PCRE_EXTRA facility).
4881
4882
4883 Version 1.00 18-Nov-97
4884 ----------------------
4885
4886 1. Added compile-time macros to support systems such as SunOS4 which don't have
4887 memmove() or strerror() but have other things that can be used instead.
4888
4889 2. Arranged that "make clean" removes the executables.
4890
4891
4892 Version 0.99 27-Oct-97
4893 ----------------------
4894
4895 1. Fixed bug in code for optimizing classes with only one character. It was
4896 initializing a 32-byte map regardless, which could cause it to run off the end
4897 of the memory it had got.
4898
4899 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
4900
4901
4902 Version 0.98 22-Oct-97
4903 ----------------------
4904
4905 1. Fixed bug in code for handling temporary memory usage when there are more
4906 back references than supplied space in the ovector. This could cause segfaults.
4907
4908
4909 Version 0.97 21-Oct-97
4910 ----------------------
4911
4912 1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
4913
4914 2. Optimized negated single characters not to use a bit map.
4915
4916 3. Brought error texts together as macro definitions; clarified some of them;
4917 fixed one that was wrong - it said "range out of order" when it meant "invalid
4918 escape sequence".
4919
4920 4. Changed some char * arguments to const char *.
4921
4922 5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
4923
4924 6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
4925 pcretest.
4926
4927
4928 Version 0.96 16-Oct-97
4929 ----------------------
4930
4931 1. Added a simple "pgrep" utility to the distribution.
4932
4933 2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
4934 unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
4935 where "ddd" means "one or more decimal digits".
4936
4937 3. Fixed serious bug. If a pattern had a back reference, but the call to
4938 pcre_exec() didn't supply a large enough ovector to record the related
4939 identifying subpattern, the match always failed. PCRE now remembers the number
4940 of the largest back reference, and gets some temporary memory in which to save
4941 the offsets during matching if necessary, in order to ensure that
4942 backreferences always work.
4943
4944 4. Increased the compatibility with Perl in a number of ways:
4945
4946 (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
4947 to request this handling. The option can be set at compile or exec time.
4948
4949 (b) $ matches before a terminating newline by default; an option
4950 PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
4951 mode). The option can be set at compile or exec time.
4952
4953 (c) The handling of \ followed by a digit other than 0 is now supposed to be
4954 the same as Perl's. If the decimal number it represents is less than 10
4955 or there aren't that many previous left capturing parentheses, an octal
4956 escape is read. Inside a character class, it's always an octal escape,
4957 even if it is a single digit.
4958
4959 (d) An escaped but undefined alphabetic character is taken as a literal,
4960 unless PCRE_EXTRA is set. Currently this just reserves the remaining
4961 escapes.
4962
4963 (e) {0} is now permitted. (The previous item is removed from the compiled
4964 pattern).
4965
4966 5. Changed all the names of code files so that the basic parts are no longer
4967 than 10 characters, and abolished the teeny "globals.c" file.
4968
4969 6. Changed the handling of character classes; they are now done with a 32-byte
4970 bit map always.
4971
4972 7. Added the -d and /D options to pcretest to make it possible to look at the
4973 internals of compilation without having to recompile pcre.
4974
4975
4976 Version 0.95 23-Sep-97
4977 ----------------------
4978
4979 1. Fixed bug in pre-pass concerning escaped "normal" characters such as \x5c or
4980 \x20 at the start of a run of normal characters. These were being treated as
4981 real characters, instead of the source characters being re-checked.
4982
4983
4984 Version 0.94 18-Sep-97
4985 ----------------------
4986
4987 1. The functions are now thread-safe, with the caveat that the global variables
4988 containing pointers to malloc() and free() or alternative functions are the
4989 same for all threads.
4990
4991 2. Get pcre_study() to generate a bitmap of initial characters for non-
4992 anchored patterns when this is possible, and use it if passed to pcre_exec().
4993
4994
4995 Version 0.93 15-Sep-97
4996 ----------------------
4997
4998 1. /(b)|(:+)/ was computing an incorrect first character.
4999
5000 2. Add pcre_study() to the API and the passing of pcre_extra to pcre_exec(),
5001 but not actually doing anything yet.
5002
5003 3. Treat "-" characters in classes that cannot be part of ranges as literals,
5004 as Perl does (e.g. [-az] or [az-]).
5005
5006 4. Set the anchored flag if a branch starts with .* or .*? because that tests
5007 all possible positions.
5008
5009 5. Split up into different modules to avoid including unneeded functions in a
5010 compiled binary. However, compile and exec are still in one module. The "study"
5011 function is split off.
5012
5013 6. The character tables are now in a separate module whose source is generated
5014 by an auxiliary program - but can then be edited by hand if required. There are
5015 now no calls to isalnum(), isspace(), isdigit(), isxdigit(), tolower() or
5016 toupper() in the code.
5017
5018 7. Turn the malloc/free funtions variables into pcre_malloc and pcre_free and
5019 make them global. Abolish the function for setting them, as the caller can now
5020 set them directly.
5021
5022
5023 Version 0.92 11-Sep-97
5024 ----------------------
5025
5026 1. A repeat with a fixed maximum and a minimum of 1 for an ordinary character
5027 (e.g. /a{1,3}/) was broken (I mis-optimized it).
5028
5029 2. Caseless matching was not working in character classes if the characters in
5030 the pattern were in upper case.
5031
5032 3. Make ranges like [W-c] work in the same way as Perl for caseless matching.
5033
5034 4. Make PCRE_ANCHORED public and accept as a compile option.
5035
5036 5. Add an options word to pcre_exec() and accept PCRE_ANCHORED and
5037 PCRE_CASELESS at run time. Add escapes \A and \I to pcretest to cause it to
5038 pass them.
5039
5040 6. Give an error if bad option bits passed at compile or run time.
5041
5042 7. Add PCRE_MULTILINE at compile and exec time, and (?m) as well. Add \M to
5043 pcretest to cause it to pass that flag.
5044
5045 8. Add pcre_info(), to get the number of identifying subpatterns, the stored
5046 options, and the first character, if set.
5047
5048 9. Recognize C+ or C{n,m} where n >= 1 as providing a fixed starting character.
5049
5050
5051 Version 0.91 10-Sep-97
5052 ----------------------
5053
5054 1. PCRE was failing to diagnose unlimited repeats of subpatterns that could
5055 match the empty string as in /(a*)*/. It was looping and ultimately crashing.
5056
5057 2. PCRE was looping on encountering an indefinitely repeated back reference to
5058 a subpattern that had matched an empty string, e.g. /(a|)\1*/. It now does what
5059 Perl does - treats the match as successful.
5060
5061 ****

Properties

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

  ViewVC Help
Powered by ViewVC 1.1.5