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

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1275 - (show annotations)
Sun Mar 10 05:32:10 2013 UTC (6 years, 8 months ago) by zherczeg
File size: 244666 byte(s)
Error occurred while calculating annotation data.
Experimental support of (*SKIP) backtracking verb in the JIT compiler.
1 ChangeLog for PCRE
2 ------------------
3
4 Version 8.33 xx-xxxx-201x
5 -------------------------
6
7 1. Added 'U' to some constants that are compared to unsigned integers, to
8 avoid compiler signed/unsigned warnings. Added (int) casts to unsigned
9 variables that are added to signed variables, to ensure the result is
10 signed and can be negated.
11
12 2. Applied patch by Daniel Richard G for quashing MSVC warnings to the
13 CMake config files.
14
15 3. Revise the creation of config.h.generic so that all boolean macros are
16 #undefined, whereas non-boolean macros are #ifndef/#endif-ed. This makes
17 overriding via -D on the command line possible.
18
19 4. Changing the definition of the variable "op" in pcre_exec.c from pcre_uchar
20 to unsigned int is reported to make a quite noticeable speed difference in
21 a specific Windows environment. Testing on Linux did also appear to show
22 some benefit (and it is clearly not harmful). Also fixed the definition of
23 Xop which should be unsigned.
24
25 5. Related to (4), changing the definition of the intermediate variable cc
26 in repeated character loops from pcre_uchar to pcre_uint32 also gave speed
27 improvements.
28
29 6. Fix forward search in JIT when link size is 3 or greater. Also removed some
30 unnecessary spaces.
31
32 7. Adjust autogen.sh and configure.ac to lose warnings given by automake 1.12
33 and later.
34
35 8. Fix two buffer over read issues in 16 and 32 bit modes. Affects JIT only.
36
37 9. Optimizing fast_forward_start_bits in JIT.
38
39 10. Adding experimental support for callouts in JIT, and fixing some
40 issues revealed during this work. Namely:
41
42 (a) Unoptimized capturing brackets incorrectly reset on backtrack.
43
44 (b) Minimum length was not checked before the matching is started.
45
46 11. The value of capture_last that is passed to callouts was incorrect in some
47 cases when there was a capture on one path that was subsequently abandoned
48 after a backtrack. Also, the capture_last value is now reset after a
49 recursion, since all captures are also reset in this case.
50
51 12. The interpreter no longer returns the "too many substrings" error in the
52 case when an overflowing capture is in a branch that is subsequently
53 abandoned after a backtrack.
54
55 13. In the pathological case when an offset vector of size 2 is used, pcretest
56 now prints out the matched string after a yield of 0 or 1.
57
58 14. Inlining subpatterns in recursions, when certain conditions are fulfilled.
59 Only supported by the JIT compiler at the moment.
60
61 15. JIT compiler now supports 32 bit Macs thanks to Lawrence Velazquez.
62
63 16. Partial matches now set offsets[2] to the "bumpalong" value, that is, the
64 offset of the starting point of the matching process, provided the offsets
65 vector is large enough.
66
67 17. The \A escape now records a lookbehind value of 1, though its execution
68 does not actually inspect the previous character. This is to ensure that,
69 in partial multi-segment matching, at least one character from the old
70 segment is retained when a new segment is processed. Otherwise, if there
71 are no lookbehinds in the pattern, \A might match incorrectly at the start
72 of a new segment.
73
74 18. Added some #ifdef __VMS code into pcretest.c to help VMS implementations.
75
76 19. Redefined some pcre_uchar variables in pcre_exec.c as pcre_uint32; this
77 gives some modest performance improvement in 8-bit mode.
78
79 20. Added the PCRE-specific property \p{Xuc} for matching characters that can
80 be expressed in certain programming languages using Universal Character
81 Names.
82
83 21. Unicode validation has been updated in the light of Unicode Corrigendum #9,
84 which points out that "non characters" are not "characters that may not
85 appear in Unicode strings" but rather "characters that are reserved for
86 internal use and have only local meaning".
87
88 22. When a pattern was compiled with automatic callouts (PCRE_AUTO_CALLOUT) and
89 there was a conditional group that depended on an assertion, if the
90 assertion was false, the callout that immediately followed the alternation
91 in the condition was skipped when pcre_exec() was used for matching.
92
93 23. Allow an explicit callout to be inserted before an assertion that is the
94 condition for a conditional group, for compatibility with automatic
95 callouts, which always insert a callout at this point.
96
97 24. In 8.31, (*COMMIT) was confined to within a recursive subpattern. Perl also
98 confines (*SKIP) and (*PRUNE) in the same way, and this has now been done.
99
100 25. (*PRUNE) is now supported by the JIT compiler.
101
102 26. Fix infinite loop when /(?<=(*SKIP)ac)a/ is matched against aa.
103
104 27. Fix the case where there are two or more SKIPs with arguments that may be
105 ignored.
106
107 28. Experimental support of (*SKIP) backtracking verb in the JIT compiler.
108
109
110 Version 8.32 30-November-2012
111 -----------------------------
112
113 1. Improved JIT compiler optimizations for first character search and single
114 character iterators.
115
116 2. Supporting IBM XL C compilers for PPC architectures in the JIT compiler.
117 Patch by Daniel Richard G.
118
119 3. Single character iterator optimizations in the JIT compiler.
120
121 4. Improved JIT compiler optimizations for character ranges.
122
123 5. Rename the "leave" variable names to "quit" to improve WinCE compatibility.
124 Reported by Giuseppe D'Angelo.
125
126 6. The PCRE_STARTLINE bit, indicating that a match can occur only at the start
127 of a line, was being set incorrectly in cases where .* appeared inside
128 atomic brackets at the start of a pattern, or where there was a subsequent
129 *PRUNE or *SKIP.
130
131 7. Improved instruction cache flush for POWER/PowerPC.
132 Patch by Daniel Richard G.
133
134 8. Fixed a number of issues in pcregrep, making it more compatible with GNU
135 grep:
136
137 (a) There is now no limit to the number of patterns to be matched.
138
139 (b) An error is given if a pattern is too long.
140
141 (c) Multiple uses of --exclude, --exclude-dir, --include, and --include-dir
142 are now supported.
143
144 (d) --exclude-from and --include-from (multiple use) have been added.
145
146 (e) Exclusions and inclusions now apply to all files and directories, not
147 just to those obtained from scanning a directory recursively.
148
149 (f) Multiple uses of -f and --file-list are now supported.
150
151 (g) In a Windows environment, the default for -d has been changed from
152 "read" (the GNU grep default) to "skip", because otherwise the presence
153 of a directory in the file list provokes an error.
154
155 (h) The documentation has been revised and clarified in places.
156
157 9. Improve the matching speed of capturing brackets.
158
159 10. Changed the meaning of \X so that it now matches a Unicode extended
160 grapheme cluster.
161
162 11. Patch by Daniel Richard G to the autoconf files to add a macro for sorting
163 out POSIX threads when JIT support is configured.
164
165 12. Added support for PCRE_STUDY_EXTRA_NEEDED.
166
167 13. In the POSIX wrapper regcomp() function, setting re_nsub field in the preg
168 structure could go wrong in environments where size_t is not the same size
169 as int.
170
171 14. Applied user-supplied patch to pcrecpp.cc to allow PCRE_NO_UTF8_CHECK to be
172 set.
173
174 15. The EBCDIC support had decayed; later updates to the code had included
175 explicit references to (e.g.) \x0a instead of CHAR_LF. There has been a
176 general tidy up of EBCDIC-related issues, and the documentation was also
177 not quite right. There is now a test that can be run on ASCII systems to
178 check some of the EBCDIC-related things (but is it not a full test).
179
180 16. The new PCRE_STUDY_EXTRA_NEEDED option is now used by pcregrep, resulting
181 in a small tidy to the code.
182
183 17. Fix JIT tests when UTF is disabled and both 8 and 16 bit mode are enabled.
184
185 18. If the --only-matching (-o) option in pcregrep is specified multiple
186 times, each one causes appropriate output. For example, -o1 -o2 outputs the
187 substrings matched by the 1st and 2nd capturing parentheses. A separating
188 string can be specified by --om-separator (default empty).
189
190 19. Improving the first n character searches.
191
192 20. Turn case lists for horizontal and vertical white space into macros so that
193 they are defined only once.
194
195 21. This set of changes together give more compatible Unicode case-folding
196 behaviour for characters that have more than one other case when UCP
197 support is available.
198
199 (a) The Unicode property table now has offsets into a new table of sets of
200 three or more characters that are case-equivalent. The MultiStage2.py
201 script that generates these tables (the pcre_ucd.c file) now scans
202 CaseFolding.txt instead of UnicodeData.txt for character case
203 information.
204
205 (b) The code for adding characters or ranges of characters to a character
206 class has been abstracted into a generalized function that also handles
207 case-independence. In UTF-mode with UCP support, this uses the new data
208 to handle characters with more than one other case.
209
210 (c) A bug that is fixed as a result of (b) is that codepoints less than 256
211 whose other case is greater than 256 are now correctly matched
212 caselessly. Previously, the high codepoint matched the low one, but not
213 vice versa.
214
215 (d) The processing of \h, \H, \v, and \ in character classes now makes use
216 of the new class addition function, using character lists defined as
217 macros alongside the case definitions of 20 above.
218
219 (e) Caseless back references now work with characters that have more than
220 one other case.
221
222 (f) General caseless matching of characters with more than one other case
223 is supported.
224
225 22. Unicode character properties were updated from Unicode 6.2.0
226
227 23. Improved CMake support under Windows. Patch by Daniel Richard G.
228
229 24. Add support for 32-bit character strings, and UTF-32
230
231 25. Major JIT compiler update (code refactoring and bugfixing).
232 Experimental Sparc 32 support is added.
233
234 26. Applied a modified version of Daniel Richard G's patch to create
235 pcre.h.generic and config.h.generic by "make" instead of in the
236 PrepareRelease script.
237
238 27. Added a definition for CHAR_NULL (helpful for the z/OS port), and use it in
239 pcre_compile.c when checking for a zero character.
240
241 28. Introducing a native interface for JIT. Through this interface, the compiled
242 machine code can be directly executed. The purpose of this interface is to
243 provide fast pattern matching, so several sanity checks are not performed.
244 However, feature tests are still performed. The new interface provides
245 1.4x speedup compared to the old one.
246
247 29. If pcre_exec() or pcre_dfa_exec() was called with a negative value for
248 the subject string length, the error given was PCRE_ERROR_BADOFFSET, which
249 was confusing. There is now a new error PCRE_ERROR_BADLENGTH for this case.
250
251 30. In 8-bit UTF-8 mode, pcretest failed to give an error for data codepoints
252 greater than 0x7fffffff (which cannot be represented in UTF-8, even under
253 the "old" RFC 2279). Instead, it ended up passing a negative length to
254 pcre_exec().
255
256 31. Add support for GCC's visibility feature to hide internal functions.
257
258 32. Running "pcretest -C pcre8" or "pcretest -C pcre16" gave a spurious error
259 "unknown -C option" after outputting 0 or 1.
260
261 33. There is now support for generating a code coverage report for the test
262 suite in environments where gcc is the compiler and lcov is installed. This
263 is mainly for the benefit of the developers.
264
265 34. If PCRE is built with --enable-valgrind, certain memory regions are marked
266 unaddressable using valgrind annotations, allowing valgrind to detect
267 invalid memory accesses. This is mainly for the benefit of the developers.
268
269 25. (*UTF) can now be used to start a pattern in any of the three libraries.
270
271 26. Give configure error if --enable-cpp but no C++ compiler found.
272
273
274 Version 8.31 06-July-2012
275 -------------------------
276
277 1. Fixing a wrong JIT test case and some compiler warnings.
278
279 2. Removed a bashism from the RunTest script.
280
281 3. Add a cast to pcre_exec.c to fix the warning "unary minus operator applied
282 to unsigned type, result still unsigned" that was given by an MS compiler
283 on encountering the code "-sizeof(xxx)".
284
285 4. Partial matching support is added to the JIT compiler.
286
287 5. Fixed several bugs concerned with partial matching of items that consist
288 of more than one character:
289
290 (a) /^(..)\1/ did not partially match "aba" because checking references was
291 done on an "all or nothing" basis. This also applied to repeated
292 references.
293
294 (b) \R did not give a hard partial match if \r was found at the end of the
295 subject.
296
297 (c) \X did not give a hard partial match after matching one or more
298 characters at the end of the subject.
299
300 (d) When newline was set to CRLF, a pattern such as /a$/ did not recognize
301 a partial match for the string "\r".
302
303 (e) When newline was set to CRLF, the metacharacter "." did not recognize
304 a partial match for a CR character at the end of the subject string.
305
306 6. If JIT is requested using /S++ or -s++ (instead of just /S+ or -s+) when
307 running pcretest, the text "(JIT)" added to the output whenever JIT is
308 actually used to run the match.
309
310 7. Individual JIT compile options can be set in pcretest by following -s+[+]
311 or /S+[+] with a digit between 1 and 7.
312
313 8. OP_NOT now supports any UTF character not just single-byte ones.
314
315 9. (*MARK) control verb is now supported by the JIT compiler.
316
317 10. The command "./RunTest list" lists the available tests without actually
318 running any of them. (Because I keep forgetting what they all are.)
319
320 11. Add PCRE_INFO_MAXLOOKBEHIND.
321
322 12. Applied a (slightly modified) user-supplied patch that improves performance
323 when the heap is used for recursion (compiled with --disable-stack-for-
324 recursion). Instead of malloc and free for each heap frame each time a
325 logical recursion happens, frames are retained on a chain and re-used where
326 possible. This sometimes gives as much as 30% improvement.
327
328 13. As documented, (*COMMIT) is now confined to within a recursive subpattern
329 call.
330
331 14. As documented, (*COMMIT) is now confined to within a positive assertion.
332
333 15. It is now possible to link pcretest with libedit as an alternative to
334 libreadline.
335
336 16. (*COMMIT) control verb is now supported by the JIT compiler.
337
338 17. The Unicode data tables have been updated to Unicode 6.1.0.
339
340 18. Added --file-list option to pcregrep.
341
342 19. Added binary file support to pcregrep, including the -a, --binary-files,
343 -I, and --text options.
344
345 20. The madvise function is renamed for posix_madvise for QNX compatibility
346 reasons. Fixed by Giuseppe D'Angelo.
347
348 21. Fixed a bug for backward assertions with REVERSE 0 in the JIT compiler.
349
350 22. Changed the option for creating symbolic links for 16-bit man pages from
351 -s to -sf so that re-installing does not cause issues.
352
353 23. Support PCRE_NO_START_OPTIMIZE in JIT as (*MARK) support requires it.
354
355 24. Fixed a very old bug in pcretest that caused errors with restarted DFA
356 matches in certain environments (the workspace was not being correctly
357 retained). Also added to pcre_dfa_exec() a simple plausibility check on
358 some of the workspace data at the beginning of a restart.
359
360 25. \s*\R was auto-possessifying the \s* when it should not, whereas \S*\R
361 was not doing so when it should - probably a typo introduced by SVN 528
362 (change 8.10/14).
363
364 26. When PCRE_UCP was not set, \w+\x{c4} was incorrectly auto-possessifying the
365 \w+ when the character tables indicated that \x{c4} was a word character.
366 There were several related cases, all because the tests for doing a table
367 lookup were testing for characters less than 127 instead of 255.
368
369 27. If a pattern contains capturing parentheses that are not used in a match,
370 their slots in the ovector are set to -1. For those that are higher than
371 any matched groups, this happens at the end of processing. In the case when
372 there were back references that the ovector was too small to contain
373 (causing temporary malloc'd memory to be used during matching), and the
374 highest capturing number was not used, memory off the end of the ovector
375 was incorrectly being set to -1. (It was using the size of the temporary
376 memory instead of the true size.)
377
378 28. To catch bugs like 27 using valgrind, when pcretest is asked to specify an
379 ovector size, it uses memory at the end of the block that it has got.
380
381 29. Check for an overlong MARK name and give an error at compile time. The
382 limit is 255 for the 8-bit library and 65535 for the 16-bit library.
383
384 30. JIT compiler update.
385
386 31. JIT is now supported on jailbroken iOS devices. Thanks for Ruiger
387 Rill for the patch.
388
389 32. Put spaces around SLJIT_PRINT_D in the JIT compiler. Required by CXX11.
390
391 33. Variable renamings in the PCRE-JIT compiler. No functionality change.
392
393 34. Fixed typos in pcregrep: in two places there was SUPPORT_LIBZ2 instead of
394 SUPPORT_LIBBZ2. This caused a build problem when bzip2 but not gzip (zlib)
395 was enabled.
396
397 35. Improve JIT code generation for greedy plus quantifier.
398
399 36. When /((?:a?)*)*c/ or /((?>a?)*)*c/ was matched against "aac", it set group
400 1 to "aa" instead of to an empty string. The bug affected repeated groups
401 that could potentially match an empty string.
402
403 37. Optimizing single character iterators in JIT.
404
405 38. Wide characters specified with \uxxxx in JavaScript mode are now subject to
406 the same checks as \x{...} characters in non-JavaScript mode. Specifically,
407 codepoints that are too big for the mode are faulted, and in a UTF mode,
408 disallowed codepoints are also faulted.
409
410 39. If PCRE was compiled with UTF support, in three places in the DFA
411 matcher there was code that should only have been obeyed in UTF mode, but
412 was being obeyed unconditionally. In 8-bit mode this could cause incorrect
413 processing when bytes with values greater than 127 were present. In 16-bit
414 mode the bug would be provoked by values in the range 0xfc00 to 0xdc00. In
415 both cases the values are those that cannot be the first data item in a UTF
416 character. The three items that might have provoked this were recursions,
417 possessively repeated groups, and atomic groups.
418
419 40. Ensure that libpcre is explicitly listed in the link commands for pcretest
420 and pcregrep, because some OS require shared objects to be explicitly
421 passed to ld, causing the link step to fail if they are not.
422
423 41. There were two incorrect #ifdefs in pcre_study.c, meaning that, in 16-bit
424 mode, patterns that started with \h* or \R* might be incorrectly matched.
425
426
427 Version 8.30 04-February-2012
428 -----------------------------
429
430 1. Renamed "isnumber" as "is_a_number" because in some Mac environments this
431 name is defined in ctype.h.
432
433 2. Fixed a bug in fixed-length calculation for lookbehinds that would show up
434 only in quite long subpatterns.
435
436 3. Removed the function pcre_info(), which has been obsolete and deprecated
437 since it was replaced by pcre_fullinfo() in February 2000.
438
439 4. For a non-anchored pattern, if (*SKIP) was given with a name that did not
440 match a (*MARK), and the match failed at the start of the subject, a
441 reference to memory before the start of the subject could occur. This bug
442 was introduced by fix 17 of release 8.21.
443
444 5. A reference to an unset group with zero minimum repetition was giving
445 totally wrong answers (in non-JavaScript-compatibility mode). For example,
446 /(another)?(\1?)test/ matched against "hello world test". This bug was
447 introduced in release 8.13.
448
449 6. Add support for 16-bit character strings (a large amount of work involving
450 many changes and refactorings).
451
452 7. RunGrepTest failed on msys because \r\n was replaced by whitespace when the
453 command "pattern=`printf 'xxx\r\njkl'`" was run. The pattern is now taken
454 from a file.
455
456 8. Ovector size of 2 is also supported by JIT based pcre_exec (the ovector size
457 rounding is not applied in this particular case).
458
459 9. The invalid Unicode surrogate codepoints U+D800 to U+DFFF are now rejected
460 if they appear, or are escaped, in patterns.
461
462 10. Get rid of a number of -Wunused-but-set-variable warnings.
463
464 11. The pattern /(?=(*:x))(q|)/ matches an empty string, and returns the mark
465 "x". The similar pattern /(?=(*:x))((*:y)q|)/ did not return a mark at all.
466 Oddly, Perl behaves the same way. PCRE has been fixed so that this pattern
467 also returns the mark "x". This bug applied to capturing parentheses,
468 non-capturing parentheses, and atomic parentheses. It also applied to some
469 assertions.
470
471 12. Stephen Kelly's patch to CMakeLists.txt allows it to parse the version
472 information out of configure.ac instead of relying on pcre.h.generic, which
473 is not stored in the repository.
474
475 13. Applied Dmitry V. Levin's patch for a more portable method for linking with
476 -lreadline.
477
478 14. ZH added PCRE_CONFIG_JITTARGET; added its output to pcretest -C.
479
480 15. Applied Graycode's patch to put the top-level frame on the stack rather
481 than the heap when not using the stack for recursion. This gives a
482 performance improvement in many cases when recursion is not deep.
483
484 16. Experimental code added to "pcretest -C" to output the stack frame size.
485
486
487 Version 8.21 12-Dec-2011
488 ------------------------
489
490 1. Updating the JIT compiler.
491
492 2. JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
493 are added as well.
494
495 3. Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
496 PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
497 calling _pcre_jit_exec. Some extra comments are added.
498
499 4. (*MARK) settings inside atomic groups that do not contain any capturing
500 parentheses, for example, (?>a(*:m)), were not being passed out. This bug
501 was introduced by change 18 for 8.20.
502
503 5. Supporting of \x, \U and \u in JavaScript compatibility mode based on the
504 ECMA-262 standard.
505
506 6. Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
507 erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
508 This bug was probably introduced by change 9 of 8.13.
509
510 7. While fixing 6 above, I noticed that a number of other items were being
511 incorrectly rejected as "not fixed length". This arose partly because newer
512 opcodes had not been added to the fixed-length checking code. I have (a)
513 corrected the bug and added tests for these items, and (b) arranged for an
514 error to occur if an unknown opcode is encountered while checking for fixed
515 length instead of just assuming "not fixed length". The items that were
516 rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
517 (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
518 repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
519
520 8. A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
521 being incorrectly compiled and would have given unpredicatble results.
522
523 9. A possessively repeated subpattern with minimum repeat count greater than
524 one behaved incorrectly. For example, (A){2,}+ behaved as if it was
525 (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
526 the first (A) could occur when it should not.
527
528 10. Add a cast and remove a redundant test from the code.
529
530 11. JIT should use pcre_malloc/pcre_free for allocation.
531
532 12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
533 best practice nowadays, and helps with cross-compiling. (If the exec_prefix
534 is anything other than /usr, -L is still shown).
535
536 13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
537
538 14. Perl does not support \N without a following name in a [] class; PCRE now
539 also gives an error.
540
541 15. If a forward reference was repeated with an upper limit of around 2000,
542 it caused the error "internal error: overran compiling workspace". The
543 maximum number of forward references (including repeats) was limited by the
544 internal workspace, and dependent on the LINK_SIZE. The code has been
545 rewritten so that the workspace expands (via pcre_malloc) if necessary, and
546 the default depends on LINK_SIZE. There is a new upper limit (for safety)
547 of around 200,000 forward references. While doing this, I also speeded up
548 the filling in of repeated forward references.
549
550 16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
551 incorrectly expecting the subject to contain another "a" after the start.
552
553 17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
554 in the match, the SKIP should be ignored. This was not happening; instead
555 the SKIP was being treated as NOMATCH. For patterns such as
556 /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
557 tested.
558
559 18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
560 now much more compatible with Perl, in particular in cases where the result
561 is a non-match for a non-anchored pattern. For example, if
562 /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
563 "m", where previously it did not return a name. A side effect of this
564 change is that for partial matches, the last encountered mark name is
565 returned, as for non matches. A number of tests that were previously not
566 Perl-compatible have been moved into the Perl-compatible test files. The
567 refactoring has had the pleasing side effect of removing one argument from
568 the match() function, thus reducing its stack requirements.
569
570 19. If the /S+ option was used in pcretest to study a pattern using JIT,
571 subsequent uses of /S (without +) incorrectly behaved like /S+.
572
573 21. Retrieve executable code size support for the JIT compiler and fixing
574 some warnings.
575
576 22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
577 not work when the shorter character appeared right at the end of the
578 subject string.
579
580 23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
581 systems.
582
583 24. Added PCRE_INFO_JITSIZE to pass on the value from (21) above, and also
584 output it when the /M option is used in pcretest.
585
586 25. The CheckMan script was not being included in the distribution. Also, added
587 an explicit "perl" to run Perl scripts from the PrepareRelease script
588 because this is reportedly needed in Windows.
589
590 26. If study data was being save in a file and studying had not found a set of
591 "starts with" bytes for the pattern, the data written to the file (though
592 never used) was taken from uninitialized memory and so caused valgrind to
593 complain.
594
595 27. Updated RunTest.bat as provided by Sheri Pierce.
596
597 28. Fixed a possible uninitialized memory bug in pcre_jit_compile.c.
598
599 29. Computation of memory usage for the table of capturing group names was
600 giving an unnecessarily large value.
601
602
603 Version 8.20 21-Oct-2011
604 ------------------------
605
606 1. Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
607 a POSIX class. After further experiments with Perl, which convinced me that
608 Perl has bugs and confusions, a closing square bracket is no longer allowed
609 in a POSIX name. This bug also affected patterns with classes that started
610 with full stops.
611
612 2. If a pattern such as /(a)b|ac/ is matched against "ac", there is no
613 captured substring, but while checking the failing first alternative,
614 substring 1 is temporarily captured. If the output vector supplied to
615 pcre_exec() was not big enough for this capture, the yield of the function
616 was still zero ("insufficient space for captured substrings"). This cannot
617 be totally fixed without adding another stack variable, which seems a lot
618 of expense for a edge case. However, I have improved the situation in cases
619 such as /(a)(b)x|abc/ matched against "abc", where the return code
620 indicates that fewer than the maximum number of slots in the ovector have
621 been set.
622
623 3. Related to (2) above: when there are more back references in a pattern than
624 slots in the output vector, pcre_exec() uses temporary memory during
625 matching, and copies in the captures as far as possible afterwards. It was
626 using the entire output vector, but this conflicts with the specification
627 that only 2/3 is used for passing back captured substrings. Now it uses
628 only the first 2/3, for compatibility. This is, of course, another edge
629 case.
630
631 4. Zoltan Herczeg's just-in-time compiler support has been integrated into the
632 main code base, and can be used by building with --enable-jit. When this is
633 done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
634 runtime --no-jit option is given.
635
636 5. When the number of matches in a pcre_dfa_exec() run exactly filled the
637 ovector, the return from the function was zero, implying that there were
638 other matches that did not fit. The correct "exactly full" value is now
639 returned.
640
641 6. If a subpattern that was called recursively or as a subroutine contained
642 (*PRUNE) or any other control that caused it to give a non-standard return,
643 invalid errors such as "Error -26 (nested recursion at the same subject
644 position)" or even infinite loops could occur.
645
646 7. If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
647 computing the minimum length on reaching *ACCEPT, and so ended up with the
648 wrong value of 1 rather than 0. Further investigation indicates that
649 computing a minimum subject length in the presence of *ACCEPT is difficult
650 (think back references, subroutine calls), and so I have changed the code
651 so that no minimum is registered for a pattern that contains *ACCEPT.
652
653 8. If (*THEN) was present in the first (true) branch of a conditional group,
654 it was not handled as intended. [But see 16 below.]
655
656 9. Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
657 Sheri Pierce.
658
659 10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
660 the first byte in a match must be "a".
661
662 11. Change 17 for 8.13 increased the recursion depth for patterns like
663 /a(?:.)*?a/ drastically. I've improved things by remembering whether a
664 pattern contains any instances of (*THEN). If it does not, the old
665 optimizations are restored. It would be nice to do this on a per-group
666 basis, but at the moment that is not feasible.
667
668 12. In some environments, the output of pcretest -C is CRLF terminated. This
669 broke RunTest's code that checks for the link size. A single white space
670 character after the value is now allowed for.
671
672 13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
673 For "fr", it uses the Windows-specific input and output files.
674
675 14. If (*THEN) appeared in a group that was called recursively or as a
676 subroutine, it did not work as intended. [But see next item.]
677
678 15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
679 pattern fragments (but not containing any | characters). If A and B are
680 matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
681 was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
682 D. In other words, Perl considers parentheses that do not contain any |
683 characters to be part of a surrounding alternative, whereas PCRE was
684 treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
685 differently. PCRE now behaves in the same way as Perl, except in the case
686 of subroutine/recursion calls such as (?1) which have in any case always
687 been different (but PCRE had them first :-).
688
689 16. Related to 15 above: Perl does not treat the | in a conditional group as
690 creating alternatives. Such a group is treated in the same way as an
691 ordinary group without any | characters when processing (*THEN). PCRE has
692 been changed to match Perl's behaviour.
693
694 17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
695 RunGrepTest script failed.
696
697 18. Change 22 for version 13 caused atomic groups to use more stack. This is
698 inevitable for groups that contain captures, but it can lead to a lot of
699 stack use in large patterns. The old behaviour has been restored for atomic
700 groups that do not contain any capturing parentheses.
701
702 19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
703 suppress the check for a minimum subject length at run time. (If it was
704 given to pcre_exec() or pcre_dfa_exec() it did work.)
705
706 20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
707 fail to work when decoding hex characters in data strings in EBCDIC
708 environments.
709
710 21. It appears that in at least one Mac OS environment, the isxdigit() function
711 is implemented as a macro that evaluates to its argument more than once,
712 contravening the C 90 Standard (I haven't checked a later standard). There
713 was an instance in pcretest which caused it to go wrong when processing
714 \x{...} escapes in subject strings. The has been rewritten to avoid using
715 things like p++ in the argument of isxdigit().
716
717
718 Version 8.13 16-Aug-2011
719 ------------------------
720
721 1. The Unicode data tables have been updated to Unicode 6.0.0.
722
723 2. Two minor typos in pcre_internal.h have been fixed.
724
725 3. Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
726 pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
727 in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
728
729 4. There were a number of related bugs in the code for matching backrefences
730 caselessly in UTF-8 mode when codes for the characters concerned were
731 different numbers of bytes. For example, U+023A and U+2C65 are an upper
732 and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
733 (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
734 code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
735 2-byte code at the end of the subject (it thought there wasn't enough data
736 left).
737
738 5. Comprehensive information about what went wrong is now returned by
739 pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
740 as the output vector has at least 2 elements. The offset of the start of
741 the failing character and a reason code are placed in the vector.
742
743 6. When the UTF-8 string check fails for pcre_compile(), the offset that is
744 now returned is for the first byte of the failing character, instead of the
745 last byte inspected. This is an incompatible change, but I hope it is small
746 enough not to be a problem. It makes the returned offset consistent with
747 pcre_exec() and pcre_dfa_exec().
748
749 7. pcretest now gives a text phrase as well as the error number when
750 pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
751 failure, the offset and reason code are output.
752
753 8. When \R was used with a maximizing quantifier it failed to skip backwards
754 over a \r\n pair if the subsequent match failed. Instead, it just skipped
755 back over a single character (\n). This seems wrong (because it treated the
756 two characters as a single entity when going forwards), conflicts with the
757 documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
758 behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
759 has been changed.
760
761 9. Some internal refactoring has changed the processing so that the handling
762 of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
763 time (the PCRE_DOTALL option was changed this way some time ago: version
764 7.7 change 16). This has made it possible to abolish the OP_OPT op code,
765 which was always a bit of a fudge. It also means that there is one less
766 argument for the match() function, which reduces its stack requirements
767 slightly. This change also fixes an incompatibility with Perl: the pattern
768 (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
769
770 10. More internal refactoring has drastically reduced the number of recursive
771 calls to match() for possessively repeated groups such as (abc)++ when
772 using pcre_exec().
773
774 11. While implementing 10, a number of bugs in the handling of groups were
775 discovered and fixed:
776
777 (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
778 (a|)*(?1) gave a compile-time internal error.
779 ((a|)+)+ did not notice that the outer group could match an empty string.
780 (^a|^)+ was not marked as anchored.
781 (.*a|.*)+ was not marked as matching at start or after a newline.
782
783 12. Yet more internal refactoring has removed another argument from the match()
784 function. Special calls to this function are now indicated by setting a
785 value in a variable in the "match data" data block.
786
787 13. Be more explicit in pcre_study() instead of relying on "default" for
788 opcodes that mean there is no starting character; this means that when new
789 ones are added and accidentally left out of pcre_study(), testing should
790 pick them up.
791
792 14. The -s option of pcretest has been documented for ages as being an old
793 synonym of -m (show memory usage). I have changed it to mean "force study
794 for every regex", that is, assume /S for every regex. This is similar to -i
795 and -d etc. It's slightly incompatible, but I'm hoping nobody is still
796 using it. It makes it easier to run collections of tests with and without
797 study enabled, and thereby test pcre_study() more easily. All the standard
798 tests are now run with and without -s (but some patterns can be marked as
799 "never study" - see 20 below).
800
801 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
802 restoration of the capturing data to the outer values was not happening
803 correctly.
804
805 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
806 empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
807 pattern had matched an empty string, and so incorrectly returned a no
808 match.
809
810 17. There was optimizing code for the last branch of non-capturing parentheses,
811 and also for the obeyed branch of a conditional subexpression, which used
812 tail recursion to cut down on stack usage. Unfortunately, now that there is
813 the possibility of (*THEN) occurring in these branches, tail recursion is
814 no longer possible because the return has to be checked for (*THEN). These
815 two optimizations have therefore been removed. [But see 8.20/11 above.]
816
817 18. If a pattern containing \R was studied, it was assumed that \R always
818 matched two bytes, thus causing the minimum subject length to be
819 incorrectly computed because \R can also match just one byte.
820
821 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
822 was incorrectly computed.
823
824 20. If /S is present twice on a test pattern in pcretest input, it now
825 *disables* studying, thereby overriding the use of -s on the command line
826 (see 14 above). This is necessary for one or two tests to keep the output
827 identical in both cases.
828
829 21. When (*ACCEPT) was used in an assertion that matched an empty string and
830 PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
831
832 22. When an atomic group that contained a capturing parenthesis was
833 successfully matched, but the branch in which it appeared failed, the
834 capturing was not being forgotten if a higher numbered group was later
835 captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
836 group 1 to "a", when in fact it should be unset. This applied to multi-
837 branched capturing and non-capturing groups, repeated or not, and also to
838 positive assertions (capturing in negative assertions does not happen
839 in PCRE) and also to nested atomic groups.
840
841 23. Add the ++ qualifier feature to pcretest, to show the remainder of the
842 subject after a captured substring, to make it easier to tell which of a
843 number of identical substrings has been captured.
844
845 24. The way atomic groups are processed by pcre_exec() has been changed so that
846 if they are repeated, backtracking one repetition now resets captured
847 values correctly. For example, if ((?>(a+)b)+aabab) is matched against
848 "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
849 "aaa". Previously, it would have been "a". As part of this code
850 refactoring, the way recursive calls are handled has also been changed.
851
852 25. If an assertion condition captured any substrings, they were not passed
853 back unless some other capturing happened later. For example, if
854 (?(?=(a))a) was matched against "a", no capturing was returned.
855
856 26. When studying a pattern that contained subroutine calls or assertions,
857 the code for finding the minimum length of a possible match was handling
858 direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
859 group 1 called group 2 while simultaneously a separate group 2 called group
860 1). A stack overflow occurred in this case. I have fixed this by limiting
861 the recursion depth to 10.
862
863 27. Updated RunTest.bat in the distribution to the version supplied by Tom
864 Fortmann. This supports explicit test numbers on the command line, and has
865 argument validation and error reporting.
866
867 28. An instance of \X with an unlimited repeat could fail if at any point the
868 first character it looked at was a mark character.
869
870 29. Some minor code refactoring concerning Unicode properties and scripts
871 should reduce the stack requirement of match() slightly.
872
873 30. Added the '=' option to pcretest to check the setting of unused capturing
874 slots at the end of the pattern, which are documented as being -1, but are
875 not included in the return count.
876
877 31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
878 compiled something random. Now it gives a compile-time error (as does
879 Perl).
880
881 32. A *MARK encountered during the processing of a positive assertion is now
882 recorded and passed back (compatible with Perl).
883
884 33. If --only-matching or --colour was set on a pcregrep call whose pattern
885 had alternative anchored branches, the search for a second match in a line
886 was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
887 matched the line "0102" twice. The same bug affected patterns that started
888 with a backwards assertion. For example /\b01|\b02/ also matched "0102"
889 twice.
890
891 34. Previously, PCRE did not allow quantification of assertions. However, Perl
892 does, and because of capturing effects, quantifying parenthesized
893 assertions may at times be useful. Quantifiers are now allowed for
894 parenthesized assertions.
895
896 35. A minor code tidy in pcre_compile() when checking options for \R usage.
897
898 36. \g was being checked for fancy things in a character class, when it should
899 just be a literal "g".
900
901 37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
902 appearance of a nested POSIX class supersedes an apparent external class.
903 For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
904 unescaped square brackets may also appear as part of class names. For
905 example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
906 more like Perl. (But see 8.20/1 above.)
907
908 38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
909 was because it thought it was \N{name}, which is not supported).
910
911 39. Add minix to OS list not supporting the -S option in pcretest.
912
913 40. PCRE tries to detect cases of infinite recursion at compile time, but it
914 cannot analyze patterns in sufficient detail to catch mutual recursions
915 such as ((?1))((?2)). There is now a runtime test that gives an error if a
916 subgroup is called recursively as a subpattern for a second time at the
917 same position in the subject string. In previous releases this might have
918 been caught by the recursion limit, or it might have run out of stack.
919
920 41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
921 happen only once. PCRE was, however incorrectly giving a compile time error
922 "recursive call could loop indefinitely" because it cannot analyze the
923 pattern in sufficient detail. The compile time test no longer happens when
924 PCRE is compiling a conditional subpattern, but actual runaway loops are
925 now caught at runtime (see 40 above).
926
927 42. It seems that Perl allows any characters other than a closing parenthesis
928 to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
929 has been changed to be the same.
930
931 43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
932 as not to get warnings when autogen.sh is called. Also changed
933 AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
934
935 44. To help people who use pcregrep to scan files containing exceedingly long
936 lines, the following changes have been made:
937
938 (a) The default value of the buffer size parameter has been increased from
939 8K to 20K. (The actual buffer used is three times this size.)
940
941 (b) The default can be changed by ./configure --with-pcregrep-bufsize when
942 PCRE is built.
943
944 (c) A --buffer-size=n option has been added to pcregrep, to allow the size
945 to be set at run time.
946
947 (d) Numerical values in pcregrep options can be followed by K or M, for
948 example --buffer-size=50K.
949
950 (e) If a line being scanned overflows pcregrep's buffer, an error is now
951 given and the return code is set to 2.
952
953 45. Add a pointer to the latest mark to the callout data block.
954
955 46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
956 partial match of an empty string instead of no match. This was specific to
957 the use of ".".
958
959 47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
960 complete match instead of a partial match. This bug was dependent on both
961 the PCRE_UTF8 and PCRE_DOTALL options being set.
962
963 48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
964 starting byte set, because \b was not being ignored.
965
966
967 Version 8.12 15-Jan-2011
968 ------------------------
969
970 1. Fixed some typos in the markup of the man pages, and wrote a script that
971 checks for such things as part of the documentation building process.
972
973 2. On a big-endian 64-bit system, pcregrep did not correctly process the
974 --match-limit and --recursion-limit options (added for 8.11). In
975 particular, this made one of the standard tests fail. (The integer value
976 went into the wrong half of a long int.)
977
978 3. If the --colour option was given to pcregrep with -v (invert match), it
979 did strange things, either producing crazy output, or crashing. It should,
980 of course, ignore a request for colour when reporting lines that do not
981 match.
982
983 4. Another pcregrep bug caused similar problems if --colour was specified with
984 -M (multiline) and the pattern match finished with a line ending.
985
986 5. In pcregrep, when a pattern that ended with a literal newline sequence was
987 matched in multiline mode, the following line was shown as part of the
988 match. This seems wrong, so I have changed it.
989
990 6. Another pcregrep bug in multiline mode, when --colour was specified, caused
991 the check for further matches in the same line (so they could be coloured)
992 to overrun the end of the current line. If another match was found, it was
993 incorrectly shown (and then shown again when found in the next line).
994
995 7. If pcregrep was compiled under Windows, there was a reference to the
996 function pcregrep_exit() before it was defined. I am assuming this was
997 the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
998 reported by a user. I've moved the definition above the reference.
999
1000
1001 Version 8.11 10-Dec-2010
1002 ------------------------
1003
1004 1. (*THEN) was not working properly if there were untried alternatives prior
1005 to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
1006 backtracked to try for "b" instead of moving to the next alternative branch
1007 at the same level (in this case, to look for "c"). The Perl documentation
1008 is clear that when (*THEN) is backtracked onto, it goes to the "next
1009 alternative in the innermost enclosing group".
1010
1011 2. (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
1012 such as (A(*COMMIT)B(*THEN)C|D) any failure after matching A should
1013 result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
1014 (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
1015 (*THEN).
1016
1017 3. If \s appeared in a character class, it removed the VT character from
1018 the class, even if it had been included by some previous item, for example
1019 in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
1020 of \s, but is part of the POSIX "space" class.)
1021
1022 4. A partial match never returns an empty string (because you can always
1023 match an empty string at the end of the subject); however the checking for
1024 an empty string was starting at the "start of match" point. This has been
1025 changed to the "earliest inspected character" point, because the returned
1026 data for a partial match starts at this character. This means that, for
1027 example, /(?<=abc)def/ gives a partial match for the subject "abc"
1028 (previously it gave "no match").
1029
1030 5. Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
1031 of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
1032 previously a full match would be given. However, setting PCRE_PARTIAL_HARD
1033 has an implication that the given string is incomplete (because a partial
1034 match is preferred over a full match). For this reason, these items now
1035 give a partial match in this situation. [Aside: previously, the one case
1036 /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
1037 match rather than a full match, which was wrong by the old rules, but is
1038 now correct.]
1039
1040 6. There was a bug in the handling of #-introduced comments, recognized when
1041 PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
1042 If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
1043 UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
1044 scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
1045 but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
1046 places in pcre_compile().
1047
1048 7. Related to (6) above, when pcre_compile() was skipping #-introduced
1049 comments when looking ahead for named forward references to subpatterns,
1050 the only newline sequence it recognized was NL. It now handles newlines
1051 according to the set newline convention.
1052
1053 8. SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
1054 former, but used strtoul(), whereas pcretest avoided strtoul() but did not
1055 cater for a lack of strerror(). These oversights have been fixed.
1056
1057 9. Added --match-limit and --recursion-limit to pcregrep.
1058
1059 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
1060
1061 11. When the -o option was used, pcregrep was setting a return code of 1, even
1062 when matches were found, and --line-buffered was not being honoured.
1063
1064 12. Added an optional parentheses number to the -o and --only-matching options
1065 of pcregrep.
1066
1067 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
1068 can match an empty string. The code to do it in pcretest and pcredemo
1069 needed fixing:
1070
1071 (a) When the newline convention was "crlf", pcretest got it wrong, skipping
1072 only one byte after an empty string match just before CRLF (this case
1073 just got forgotten; "any" and "anycrlf" were OK).
1074
1075 (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
1076 mode when an empty string match preceded an ASCII character followed by
1077 a non-ASCII character. (The code for advancing by one character rather
1078 than one byte was nonsense.)
1079
1080 (c) The pcredemo.c sample program did not have any code at all to handle
1081 the cases when CRLF is a valid newline sequence.
1082
1083 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
1084 as a starting offset was within the subject string. There is now a new
1085 error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
1086 negative or greater than the length of the string. In order to test this,
1087 pcretest is extended to allow the setting of negative starting offsets.
1088
1089 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
1090 starting offset points to the beginning of a UTF-8 character was
1091 unnecessarily clumsy. I tidied it up.
1092
1093 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
1094 bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
1095
1096 17. Nobody had reported that the --include_dir option, which was added in
1097 release 7.7 should have been called --include-dir (hyphen, not underscore)
1098 for compatibility with GNU grep. I have changed it to --include-dir, but
1099 left --include_dir as an undocumented synonym, and the same for
1100 --exclude-dir, though that is not available in GNU grep, at least as of
1101 release 2.5.4.
1102
1103 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
1104 characters from a string of bytes) have been redefined so as not to use
1105 loops, in order to improve performance in some environments. At the same
1106 time, I abstracted some of the common code into auxiliary macros to save
1107 repetition (this should not affect the compiled code).
1108
1109 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
1110 compile-time error is now given if \c is not followed by an ASCII
1111 character, that is, a byte less than 128. (In EBCDIC mode, the code is
1112 different, and any byte value is allowed.)
1113
1114 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
1115 START_OPTIMIZE option, which is now allowed at compile time - but just
1116 passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
1117 to pcregrep and other applications that have no direct access to PCRE
1118 options. The new /Y option in pcretest sets this option when calling
1119 pcre_compile().
1120
1121 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
1122 back references. Groups containing recursive back references were forced to
1123 be atomic by that change, but in the case of named groups, the amount of
1124 memory required was incorrectly computed, leading to "Failed: internal
1125 error: code overflow". This has been fixed.
1126
1127 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
1128 pcretest.c, to avoid build problems in some Borland environments.
1129
1130
1131 Version 8.10 25-Jun-2010
1132 ------------------------
1133
1134 1. Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
1135 THEN.
1136
1137 2. (*ACCEPT) was not working when inside an atomic group.
1138
1139 3. Inside a character class, \B is treated as a literal by default, but
1140 faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
1141 causes the error). The code is unchanged, but I tidied the documentation.
1142
1143 4. Inside a character class, PCRE always treated \R and \X as literals,
1144 whereas Perl faults them if its -w option is set. I have changed PCRE so
1145 that it faults them when PCRE_EXTRA is set.
1146
1147 5. Added support for \N, which always matches any character other than
1148 newline. (It is the same as "." when PCRE_DOTALL is not set.)
1149
1150 6. When compiling pcregrep with newer versions of gcc which may have
1151 FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
1152 declared with attribute warn_unused_result" were given. Just casting the
1153 result to (void) does not stop the warnings; a more elaborate fudge is
1154 needed. I've used a macro to implement this.
1155
1156 7. Minor change to pcretest.c to avoid a compiler warning.
1157
1158 8. Added four artifical Unicode properties to help with an option to make
1159 \s etc use properties (see next item). The new properties are: Xan
1160 (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
1161
1162 9. Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
1163 use Unicode properties. (*UCP) at the start of a pattern can be used to set
1164 this option. Modified pcretest to add /W to test this facility. Added
1165 REG_UCP to make it available via the POSIX interface.
1166
1167 10. Added --line-buffered to pcregrep.
1168
1169 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
1170 studied, and the match started with a letter with a code point greater than
1171 127 whose first byte was different to the first byte of the other case of
1172 the letter, the other case of this starting letter was not recognized
1173 (#976).
1174
1175 12. If a pattern that was studied started with a repeated Unicode property
1176 test, for example, \p{Nd}+, there was the theoretical possibility of
1177 setting up an incorrect bitmap of starting bytes, but fortunately it could
1178 not have actually happened in practice until change 8 above was made (it
1179 added property types that matched character-matching opcodes).
1180
1181 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
1182 possible starting bytes for non-anchored patterns.
1183
1184 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
1185 \R, and also a number of cases that involve Unicode properties, both
1186 explicit and implicit when PCRE_UCP is set.
1187
1188 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
1189 input, it could crash or give wrong results if characters with values
1190 greater than 0xc0 were present in the subject string. (Detail: it assumed
1191 UTF-8 input when processing these items.)
1192
1193 16. Added a lot of (int) casts to avoid compiler warnings in systems where
1194 size_t is 64-bit (#991).
1195
1196 17. Added a check for running out of memory when PCRE is compiled with
1197 --disable-stack-for-recursion (#990).
1198
1199 18. If the last data line in a file for pcretest does not have a newline on
1200 the end, a newline was missing in the output.
1201
1202 19. The default pcre_chartables.c file recognizes only ASCII characters (values
1203 less than 128) in its various bitmaps. However, there is a facility for
1204 generating tables according to the current locale when PCRE is compiled. It
1205 turns out that in some environments, 0x85 and 0xa0, which are Unicode space
1206 characters, are recognized by isspace() and therefore were getting set in
1207 these tables, and indeed these tables seem to approximate to ISO 8859. This
1208 caused a problem in UTF-8 mode when pcre_study() was used to create a list
1209 of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
1210 which of course cannot start UTF-8 characters. I have changed the code so
1211 that only real ASCII characters (less than 128) and the correct starting
1212 bytes for UTF-8 encodings are set for characters greater than 127 when in
1213 UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
1214 altogether.)
1215
1216 20. Added the /T option to pcretest so as to be able to run tests with non-
1217 standard character tables, thus making it possible to include the tests
1218 used for 19 above in the standard set of tests.
1219
1220 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
1221 reference to a subpattern the other side of a comment that contains an
1222 opening parenthesis caused either an internal compiling error, or a
1223 reference to the wrong subpattern.
1224
1225
1226 Version 8.02 19-Mar-2010
1227 ------------------------
1228
1229 1. The Unicode data tables have been updated to Unicode 5.2.0.
1230
1231 2. Added the option --libs-cpp to pcre-config, but only when C++ support is
1232 configured.
1233
1234 3. Updated the licensing terms in the pcregexp.pas file, as agreed with the
1235 original author of that file, following a query about its status.
1236
1237 4. On systems that do not have stdint.h (e.g. Solaris), check for and include
1238 inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
1239
1240 5. A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
1241 quantifier applied to a forward-referencing subroutine call, could compile
1242 incorrect code or give the error "internal error: previously-checked
1243 referenced subpattern not found".
1244
1245 6. Both MS Visual Studio and Symbian OS have problems with initializing
1246 variables to point to external functions. For these systems, therefore,
1247 pcre_malloc etc. are now initialized to local functions that call the
1248 relevant global functions.
1249
1250 7. There were two entries missing in the vectors called coptable and poptable
1251 in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
1252 I've fixed the data, and added a kludgy way of testing at compile time that
1253 the lengths are correct (equal to the number of opcodes).
1254
1255 8. Following on from 7, I added a similar kludge to check the length of the
1256 eint vector in pcreposix.c.
1257
1258 9. Error texts for pcre_compile() are held as one long string to avoid too
1259 much relocation at load time. To find a text, the string is searched,
1260 counting zeros. There was no check for running off the end of the string,
1261 which could happen if a new error number was added without updating the
1262 string.
1263
1264 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
1265
1266 11. \K was not working if it appeared in an atomic group or in a group that
1267 was called as a "subroutine", or in an assertion. Perl 5.11 documents that
1268 \K is "not well defined" if used in an assertion. PCRE now accepts it if
1269 the assertion is positive, but not if it is negative.
1270
1271 12. Change 11 fortuitously reduced the size of the stack frame used in the
1272 "match()" function of pcre_exec.c by one pointer. Forthcoming
1273 implementation of support for (*MARK) will need an extra pointer on the
1274 stack; I have reserved it now, so that the stack frame size does not
1275 decrease.
1276
1277 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
1278 item in branch that calls a recursion is a subroutine call - as in the
1279 second branch in the above example - was incorrectly given the compile-
1280 time error "recursive call could loop indefinitely" because pcre_compile()
1281 was not correctly checking the subroutine for matching a non-empty string.
1282
1283 14. The checks for overrunning compiling workspace could trigger after an
1284 overrun had occurred. This is a "should never occur" error, but it can be
1285 triggered by pathological patterns such as hundreds of nested parentheses.
1286 The checks now trigger 100 bytes before the end of the workspace.
1287
1288 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
1289
1290
1291 Version 8.01 19-Jan-2010
1292 ------------------------
1293
1294 1. If a pattern contained a conditional subpattern with only one branch (in
1295 particular, this includes all (*DEFINE) patterns), a call to pcre_study()
1296 computed the wrong minimum data length (which is of course zero for such
1297 subpatterns). This could cause incorrect "no match" results.
1298
1299 2. For patterns such as (?i)a(?-i)b|c where an option setting at the start of
1300 the pattern is reset in the first branch, pcre_compile() failed with
1301 "internal error: code overflow at offset...". This happened only when
1302 the reset was to the original external option setting. (An optimization
1303 abstracts leading options settings into an external setting, which was the
1304 cause of this.)
1305
1306 3. A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
1307 of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
1308 assertion pattern did not match (meaning that the assertion was true), it
1309 was incorrectly treated as false if the SKIP had been reached during the
1310 matching. This also applied to assertions used as conditions.
1311
1312 4. If an item that is not supported by pcre_dfa_exec() was encountered in an
1313 assertion subpattern, including such a pattern used as a condition,
1314 unpredictable results occurred, instead of the error return
1315 PCRE_ERROR_DFA_UITEM.
1316
1317 5. The C++ GlobalReplace function was not working like Perl for the special
1318 situation when an empty string is matched. It now does the fancy magic
1319 stuff that is necessary.
1320
1321 6. In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
1322 removed. (These were left over from very, very early versions of PCRE.)
1323
1324 7. Some cosmetic changes to the code to make life easier when compiling it
1325 as part of something else:
1326
1327 (a) Change DEBUG to PCRE_DEBUG.
1328
1329 (b) In pcre_compile(), rename the member of the "branch_chain" structure
1330 called "current" as "current_branch", to prevent a collision with the
1331 Linux macro when compiled as a kernel module.
1332
1333 (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
1334 prevent a collision with the Linux macro when compiled as a kernel
1335 module.
1336
1337 8. In pcre_compile() there are some checks for integer overflows that used to
1338 cast potentially large values to (double). This has been changed to that
1339 when building, a check for int64_t is made, and if it is found, it is used
1340 instead, thus avoiding the use of floating point arithmetic. (There is no
1341 other use of FP in PCRE.) If int64_t is not found, the fallback is to
1342 double.
1343
1344 9. Added two casts to avoid signed/unsigned warnings from VS Studio Express
1345 2005 (difference between two addresses compared to an unsigned value).
1346
1347 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
1348 custom one, because of the following reported problem in Windows:
1349
1350 - libbz2 uses the Pascal calling convention (WINAPI) for the functions
1351 under Win32.
1352 - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
1353 therefore missing the function definition.
1354 - The compiler thus generates a "C" signature for the test function.
1355 - The linker fails to find the "C" function.
1356 - PCRE fails to configure if asked to do so against libbz2.
1357
1358 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
1359 messages were output:
1360
1361 Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
1362 rerunning libtoolize, to keep the correct libtool macros in-tree.
1363 Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
1364
1365 I have done both of these things.
1366
1367 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
1368 most of the time, it *can* run out if it is given a pattern that contains a
1369 runaway infinite recursion. I updated the discussion in the pcrestack man
1370 page.
1371
1372 13. Now that we have gone to the x.xx style of version numbers, the minor
1373 version may start with zero. Using 08 or 09 is a bad idea because users
1374 might check the value of PCRE_MINOR in their code, and 08 or 09 may be
1375 interpreted as invalid octal numbers. I've updated the previous comment in
1376 configure.ac, and also added a check that gives an error if 08 or 09 are
1377 used.
1378
1379 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
1380 causing partial matching to fail when the end of the subject matched \W
1381 in a UTF-8 pattern where \W was quantified with a minimum of 3.
1382
1383 15. There were some discrepancies between the declarations in pcre_internal.h
1384 of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
1385 their definitions. The declarations used "const uschar *" and the
1386 definitions used USPTR. Even though USPTR is normally defined as "const
1387 unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
1388 reported that: "This difference in casting confuses some C++ compilers, for
1389 example, SunCC recognizes above declarations as different functions and
1390 generates broken code for hbpcre." I have changed the declarations to use
1391 USPTR.
1392
1393 16. GNU libtool is named differently on some systems. The autogen.sh script now
1394 tries several variants such as glibtoolize (MacOSX) and libtoolize1x
1395 (FreeBSD).
1396
1397 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
1398 (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
1399 comment: "Figure out how to create a longlong from a string: strtoll and
1400 equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
1401 instance, but it only takes 2 args instead of 3!"
1402
1403 18. A subtle bug concerned with back references has been fixed by a change of
1404 specification, with a corresponding code fix. A pattern such as
1405 ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
1406 refers, was giving matches when it shouldn't. For example, xa=xaaa would
1407 match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
1408 same bug. Such groups have to be quantified to be useful, or contained
1409 inside another quantified group. (If there's no repetition, the reference
1410 can never match.) The problem arises because, having left the group and
1411 moved on to the rest of the pattern, a later failure that backtracks into
1412 the group uses the captured value from the final iteration of the group
1413 rather than the correct earlier one. I have fixed this in PCRE by forcing
1414 any group that contains a reference to itself to be an atomic group; that
1415 is, there cannot be any backtracking into it once it has completed. This is
1416 similar to recursive and subroutine calls.
1417
1418
1419 Version 8.00 19-Oct-09
1420 ----------------------
1421
1422 1. The table for translating pcre_compile() error codes into POSIX error codes
1423 was out-of-date, and there was no check on the pcre_compile() error code
1424 being within the table. This could lead to an OK return being given in
1425 error.
1426
1427 2. Changed the call to open a subject file in pcregrep from fopen(pathname,
1428 "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
1429 in a Windows environment.
1430
1431 3. The pcregrep --count option prints the count for each file even when it is
1432 zero, as does GNU grep. However, pcregrep was also printing all files when
1433 --files-with-matches was added. Now, when both options are given, it prints
1434 counts only for those files that have at least one match. (GNU grep just
1435 prints the file name in this circumstance, but including the count seems
1436 more useful - otherwise, why use --count?) Also ensured that the
1437 combination -clh just lists non-zero counts, with no names.
1438
1439 4. The long form of the pcregrep -F option was incorrectly implemented as
1440 --fixed_strings instead of --fixed-strings. This is an incompatible change,
1441 but it seems right to fix it, and I didn't think it was worth preserving
1442 the old behaviour.
1443
1444 5. The command line items --regex=pattern and --regexp=pattern were not
1445 recognized by pcregrep, which required --regex pattern or --regexp pattern
1446 (with a space rather than an '='). The man page documented the '=' forms,
1447 which are compatible with GNU grep; these now work.
1448
1449 6. No libpcreposix.pc file was created for pkg-config; there was just
1450 libpcre.pc and libpcrecpp.pc. The omission has been rectified.
1451
1452 7. Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
1453 when UCP support is not needed, by modifying the Python script that
1454 generates it from Unicode data files. This should not matter if the module
1455 is correctly used as a library, but I received one complaint about 50K of
1456 unwanted data. My guess is that the person linked everything into his
1457 program rather than using a library. Anyway, it does no harm.
1458
1459 8. A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
1460 was a minimum greater than 1 for a wide character in a possessive
1461 repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
1462 which had an unlimited repeat of a nested, fixed maximum repeat of a wide
1463 character. Chaos in the form of incorrect output or a compiling loop could
1464 result.
1465
1466 9. The restrictions on what a pattern can contain when partial matching is
1467 requested for pcre_exec() have been removed. All patterns can now be
1468 partially matched by this function. In addition, if there are at least two
1469 slots in the offset vector, the offset of the earliest inspected character
1470 for the match and the offset of the end of the subject are set in them when
1471 PCRE_ERROR_PARTIAL is returned.
1472
1473 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
1474 synonymous with PCRE_PARTIAL, for backwards compatibility, and
1475 PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
1476 and may be more useful for multi-segment matching.
1477
1478 11. Partial matching with pcre_exec() is now more intuitive. A partial match
1479 used to be given if ever the end of the subject was reached; now it is
1480 given only if matching could not proceed because another character was
1481 needed. This makes a difference in some odd cases such as Z(*FAIL) with the
1482 string "Z", which now yields "no match" instead of "partial match". In the
1483 case of pcre_dfa_exec(), "no match" is given if every matching path for the
1484 final character ended with (*FAIL).
1485
1486 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
1487 if the pattern had a "must contain" character that was already found in the
1488 earlier partial match, unless partial matching was again requested. For
1489 example, with the pattern /dog.(body)?/, the "must contain" character is
1490 "g". If the first part-match was for the string "dog", restarting with
1491 "sbody" failed. This bug has been fixed.
1492
1493 13. The string returned by pcre_dfa_exec() after a partial match has been
1494 changed so that it starts at the first inspected character rather than the
1495 first character of the match. This makes a difference only if the pattern
1496 starts with a lookbehind assertion or \b or \B (\K is not supported by
1497 pcre_dfa_exec()). It's an incompatible change, but it makes the two
1498 matching functions compatible, and I think it's the right thing to do.
1499
1500 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
1501 so that the demonstration program is easily available in environments where
1502 PCRE has not been installed from source.
1503
1504 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
1505 libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
1506 library.
1507
1508 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
1509 It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
1510 is not the first non-POSIX option to be added. Clearly some people find
1511 these options useful.
1512
1513 17. If a caller to the POSIX matching function regexec() passes a non-zero
1514 value for nmatch with a NULL value for pmatch, the value of
1515 nmatch is forced to zero.
1516
1517 18. RunGrepTest did not have a test for the availability of the -u option of
1518 the diff command, as RunTest does. It now checks in the same way as
1519 RunTest, and also checks for the -b option.
1520
1521 19. If an odd number of negated classes containing just a single character
1522 interposed, within parentheses, between a forward reference to a named
1523 subpattern and the definition of the subpattern, compilation crashed with
1524 an internal error, complaining that it could not find the referenced
1525 subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
1526 [The bug was that it was starting one character too far in when skipping
1527 over the character class, thus treating the ] as data rather than
1528 terminating the class. This meant it could skip too much.]
1529
1530 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
1531 /g option in pcretest when the pattern contains \K, which makes it possible
1532 to have an empty string match not at the start, even when the pattern is
1533 anchored. Updated pcretest and pcredemo to use this option.
1534
1535 21. If the maximum number of capturing subpatterns in a recursion was greater
1536 than the maximum at the outer level, the higher number was returned, but
1537 with unset values at the outer level. The correct (outer level) value is
1538 now given.
1539
1540 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
1541 PCRE did not set those parentheses (unlike Perl). I have now found a way to
1542 make it do so. The string so far is captured, making this feature
1543 compatible with Perl.
1544
1545 23. The tests have been re-organized, adding tests 11 and 12, to make it
1546 possible to check the Perl 5.10 features against Perl 5.10.
1547
1548 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
1549 pattern matches a fixed length string. PCRE did not allow this; now it
1550 does. Neither allows recursion.
1551
1552 25. I finally figured out how to implement a request to provide the minimum
1553 length of subject string that was needed in order to match a given pattern.
1554 (It was back references and recursion that I had previously got hung up
1555 on.) This code has now been added to pcre_study(); it finds a lower bound
1556 to the length of subject needed. It is not necessarily the greatest lower
1557 bound, but using it to avoid searching strings that are too short does give
1558 some useful speed-ups. The value is available to calling programs via
1559 pcre_fullinfo().
1560
1561 26. While implementing 25, I discovered to my embarrassment that pcretest had
1562 not been passing the result of pcre_study() to pcre_dfa_exec(), so the
1563 study optimizations had never been tested with that matching function.
1564 Oops. What is worse, even when it was passed study data, there was a bug in
1565 pcre_dfa_exec() that meant it never actually used it. Double oops. There
1566 were also very few tests of studied patterns with pcre_dfa_exec().
1567
1568 27. If (?| is used to create subpatterns with duplicate numbers, they are now
1569 allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
1570 on the other side of the coin, they are no longer allowed to have different
1571 names, because these cannot be distinguished in PCRE, and this has caused
1572 confusion. (This is a difference from Perl.)
1573
1574 28. When duplicate subpattern names are present (necessarily with different
1575 numbers, as required by 27 above), and a test is made by name in a
1576 conditional pattern, either for a subpattern having been matched, or for
1577 recursion in such a pattern, all the associated numbered subpatterns are
1578 tested, and the overall condition is true if the condition is true for any
1579 one of them. This is the way Perl works, and is also more like the way
1580 testing by number works.
1581
1582
1583 Version 7.9 11-Apr-09
1584 ---------------------
1585
1586 1. When building with support for bzlib/zlib (pcregrep) and/or readline
1587 (pcretest), all targets were linked against these libraries. This included
1588 libpcre, libpcreposix, and libpcrecpp, even though they do not use these
1589 libraries. This caused unwanted dependencies to be created. This problem
1590 has been fixed, and now only pcregrep is linked with bzlib/zlib and only
1591 pcretest is linked with readline.
1592
1593 2. The "typedef int BOOL" in pcre_internal.h that was included inside the
1594 "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
1595 moved outside it again, because FALSE and TRUE are already defined in AIX,
1596 but BOOL is not.
1597
1598 3. The pcre_config() function was treating the PCRE_MATCH_LIMIT and
1599 PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
1600
1601 4. The pcregrep documentation said spaces were inserted as well as colons (or
1602 hyphens) following file names and line numbers when outputting matching
1603 lines. This is not true; no spaces are inserted. I have also clarified the
1604 wording for the --colour (or --color) option.
1605
1606 5. In pcregrep, when --colour was used with -o, the list of matching strings
1607 was not coloured; this is different to GNU grep, so I have changed it to be
1608 the same.
1609
1610 6. When --colo(u)r was used in pcregrep, only the first matching substring in
1611 each matching line was coloured. Now it goes on to look for further matches
1612 of any of the test patterns, which is the same behaviour as GNU grep.
1613
1614 7. A pattern that could match an empty string could cause pcregrep to loop; it
1615 doesn't make sense to accept an empty string match in pcregrep, so I have
1616 locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
1617 seems to be how GNU grep behaves.
1618
1619 8. The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
1620 start or after a newline", because the conditional assertion was not being
1621 correctly handled. The rule now is that both the assertion and what follows
1622 in the first alternative must satisfy the test.
1623
1624 9. If auto-callout was enabled in a pattern with a conditional group whose
1625 condition was an assertion, PCRE could crash during matching, both with
1626 pcre_exec() and pcre_dfa_exec().
1627
1628 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
1629 used for matching.
1630
1631 11. Unicode property support in character classes was not working for
1632 characters (bytes) greater than 127 when not in UTF-8 mode.
1633
1634 12. Added the -M command line option to pcretest.
1635
1636 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
1637
1638 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
1639
1640 16. Added comments and documentation about mis-use of no_arg in the C++
1641 wrapper.
1642
1643 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
1644 from Martin Jerabek that uses macro names for all relevant character and
1645 string constants.
1646
1647 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
1648 SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
1649 SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
1650 these, but not everybody uses configure.
1651
1652 19. A conditional group that had only one branch was not being correctly
1653 recognized as an item that could match an empty string. This meant that an
1654 enclosing group might also not be so recognized, causing infinite looping
1655 (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
1656 with the subject "ab", where knowledge that the repeated group can match
1657 nothing is needed in order to break the loop.
1658
1659 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
1660 exec(), but without supplying a callout function, matching went wrong.
1661
1662 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
1663 leak if the size of the offset vector was greater than 30. When the vector
1664 is smaller, the saved offsets during recursion go onto a local stack
1665 vector, but for larger vectors malloc() is used. It was failing to free
1666 when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
1667 error, in fact).
1668
1669 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
1670 heapframe that is used only when UTF-8 support is enabled. This caused no
1671 problem, but was untidy.
1672
1673 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
1674 CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
1675 included within another project.
1676
1677 24. Steven Van Ingelgem's patches to add more options to the CMake support,
1678 slightly modified by me:
1679
1680 (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
1681 not building pcregrep.
1682
1683 (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
1684 if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
1685
1686 25. Forward references, both numeric and by name, in patterns that made use of
1687 duplicate group numbers, could behave incorrectly or give incorrect errors,
1688 because when scanning forward to find the reference group, PCRE was not
1689 taking into account the duplicate group numbers. A pattern such as
1690 ^X(?3)(a)(?|(b)|(q))(Y) is an example.
1691
1692 26. Changed a few more instances of "const unsigned char *" to USPTR, making
1693 the feature of a custom pointer more persuasive (as requested by a user).
1694
1695 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
1696 pcretest.c, inside #ifndefs, because it seems they are sometimes already
1697 pre-defined.
1698
1699 28. Added support for (*UTF8) at the start of a pattern.
1700
1701 29. Arrange for flags added by the "release type" setting in CMake to be shown
1702 in the configuration summary.
1703
1704
1705 Version 7.8 05-Sep-08
1706 ---------------------
1707
1708 1. Replaced UCP searching code with optimized version as implemented for Ad
1709 Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
1710 stage table and inline lookup instead of a function, giving speed ups of 2
1711 to 5 times on some simple patterns that I tested. Permission was given to
1712 distribute the MultiStage2.py script that generates the tables (it's not in
1713 the tarball, but is in the Subversion repository).
1714
1715 2. Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
1716 scripts.
1717
1718 3. Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
1719 a group with a zero qualifier. The result of the study could be incorrect,
1720 or the function might crash, depending on the pattern.
1721
1722 4. Caseless matching was not working for non-ASCII characters in back
1723 references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
1724 It now works when Unicode Property Support is available.
1725
1726 5. In pcretest, an escape such as \x{de} in the data was always generating
1727 a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
1728 non-UTF-8 mode. If the value is greater than 255, it gives a warning about
1729 truncation.
1730
1731 6. Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
1732
1733 7. Added two (int) casts to pcregrep when printing the difference of two
1734 pointers, in case they are 64-bit values.
1735
1736 8. Added comments about Mac OS X stack usage to the pcrestack man page and to
1737 test 2 if it fails.
1738
1739 9. Added PCRE_CALL_CONVENTION just before the names of all exported functions,
1740 and a #define of that name to empty if it is not externally set. This is to
1741 allow users of MSVC to set it if necessary.
1742
1743 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
1744 the convenience functions in the pcre_get.c source file.
1745
1746 11. An option change at the start of a pattern that had top-level alternatives
1747 could cause overwriting and/or a crash. This command provoked a crash in
1748 some environments:
1749
1750 printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
1751
1752 This potential security problem was recorded as CVE-2008-2371.
1753
1754 12. For a pattern where the match had to start at the beginning or immediately
1755 after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
1756 pcre_dfa_exec() could read past the end of the passed subject if there was
1757 no match. To help with detecting such bugs (e.g. with valgrind), I modified
1758 pcretest so that it places the subject at the end of its malloc-ed buffer.
1759
1760 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
1761 exec() might read past the end of the data buffer in UTF-8 mode.
1762
1763 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
1764 the data contained the byte 0x85 as part of a UTF-8 character within its
1765 first line. This applied both to normal and DFA matching.
1766
1767 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
1768 /^[^d]*?$/8 failed to match "abc".
1769
1770 16. Added a missing copyright notice to pcrecpp_internal.h.
1771
1772 17. Make it more clear in the documentation that values returned from
1773 pcre_exec() in ovector are byte offsets, not character counts.
1774
1775 18. Tidied a few places to stop certain compilers from issuing warnings.
1776
1777 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
1778 supplied by Stefan Weber. I made a further small update for 7.8 because
1779 there is a change of source arrangements: the pcre_searchfuncs.c module is
1780 replaced by pcre_ucd.c.
1781
1782
1783 Version 7.7 07-May-08
1784 ---------------------
1785
1786 1. Applied Craig's patch to sort out a long long problem: "If we can't convert
1787 a string to a long long, pretend we don't even have a long long." This is
1788 done by checking for the strtoq, strtoll, and _strtoi64 functions.
1789
1790 2. Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
1791 pre-7.6 versions, which defined a global no_arg variable instead of putting
1792 it in the RE class. (See also #8 below.)
1793
1794 3. Remove a line of dead code, identified by coverity and reported by Nuno
1795 Lopes.
1796
1797 4. Fixed two related pcregrep bugs involving -r with --include or --exclude:
1798
1799 (1) The include/exclude patterns were being applied to the whole pathnames
1800 of files, instead of just to the final components.
1801
1802 (2) If there was more than one level of directory, the subdirectories were
1803 skipped unless they satisfied the include/exclude conditions. This is
1804 inconsistent with GNU grep (and could even be seen as contrary to the
1805 pcregrep specification - which I improved to make it absolutely clear).
1806 The action now is always to scan all levels of directory, and just
1807 apply the include/exclude patterns to regular files.
1808
1809 5. Added the --include_dir and --exclude_dir patterns to pcregrep, and used
1810 --exclude_dir in the tests to avoid scanning .svn directories.
1811
1812 6. Applied Craig's patch to the QuoteMeta function so that it escapes the
1813 NUL character as backslash + 0 rather than backslash + NUL, because PCRE
1814 doesn't support NULs in patterns.
1815
1816 7. Added some missing "const"s to declarations of static tables in
1817 pcre_compile.c and pcre_dfa_exec.c.
1818
1819 8. Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
1820 caused by fix #2 above. (Subsequently also a second patch to fix the
1821 first patch. And a third patch - this was a messy problem.)
1822
1823 9. Applied Craig's patch to remove the use of push_back().
1824
1825 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
1826 matching function regexec().
1827
1828 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
1829 which, however, unlike Perl's \g{...}, are subroutine calls, not back
1830 references. PCRE supports relative numbers with this syntax (I don't think
1831 Oniguruma does).
1832
1833 12. Previously, a group with a zero repeat such as (...){0} was completely
1834 omitted from the compiled regex. However, this means that if the group
1835 was called as a subroutine from elsewhere in the pattern, things went wrong
1836 (an internal error was given). Such groups are now left in the compiled
1837 pattern, with a new opcode that causes them to be skipped at execution
1838 time.
1839
1840 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
1841 to the way PCRE behaves:
1842
1843 (a) A lone ] character is dis-allowed (Perl treats it as data).
1844
1845 (b) A back reference to an unmatched subpattern matches an empty string
1846 (Perl fails the current match path).
1847
1848 (c) A data ] in a character class must be notated as \] because if the
1849 first data character in a class is ], it defines an empty class. (In
1850 Perl it is not possible to have an empty class.) The empty class []
1851 never matches; it forces failure and is equivalent to (*FAIL) or (?!).
1852 The negative empty class [^] matches any one character, independently
1853 of the DOTALL setting.
1854
1855 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
1856 non-existent subpattern following a character class starting with ']' and
1857 containing () gave an internal compiling error instead of "reference to
1858 non-existent subpattern". Fortunately, when the pattern did exist, the
1859 compiled code was correct. (When scanning forwards to check for the
1860 existencd of the subpattern, it was treating the data ']' as terminating
1861 the class, so got the count wrong. When actually compiling, the reference
1862 was subsequently set up correctly.)
1863
1864 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
1865 it was being rejected as not supported by pcre_dfa_exec(), even though
1866 other assertions are supported. I have made pcre_dfa_exec() support
1867 (*FAIL).
1868
1869 16. The implementation of 13c above involved the invention of a new opcode,
1870 OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
1871 cannot be changed at match time, I realized I could make a small
1872 improvement to matching performance by compiling OP_ALLANY instead of
1873 OP_ANY for "." when DOTALL was set, and then removing the runtime tests
1874 on the OP_ANY path.
1875
1876 17. Compiling pcretest on Windows with readline support failed without the
1877 following two fixes: (1) Make the unistd.h include conditional on
1878 HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
1879
1880 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
1881 ncurses library to be included for pcretest when ReadLine support is
1882 requested, but also to allow for it to be overridden. This patch came from
1883 Daniel Bergström.
1884
1885 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
1886 as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
1887 any errors with the current Unicode tables. Thanks to Peter Kankowski for
1888 spotting this.
1889
1890
1891 Version 7.6 28-Jan-08
1892 ---------------------
1893
1894 1. A character class containing a very large number of characters with
1895 codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
1896 overflow.
1897
1898 2. Patch to cut out the "long long" test in pcrecpp_unittest when
1899 HAVE_LONG_LONG is not defined.
1900
1901 3. Applied Christian Ehrlicher's patch to update the CMake build files to
1902 bring them up to date and include new features. This patch includes:
1903
1904 - Fixed PH's badly added libz and libbz2 support.
1905 - Fixed a problem with static linking.
1906 - Added pcredemo. [But later removed - see 7 below.]
1907 - Fixed dftables problem and added an option.
1908 - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
1909 HAVE_LONG_LONG.
1910 - Added readline support for pcretest.
1911 - Added an listing of the option settings after cmake has run.
1912
1913 4. A user submitted a patch to Makefile that makes it easy to create
1914 "pcre.dll" under mingw when using Configure/Make. I added stuff to
1915 Makefile.am that cause it to include this special target, without
1916 affecting anything else. Note that the same mingw target plus all
1917 the other distribution libraries and programs are now supported
1918 when configuring with CMake (see 6 below) instead of with
1919 Configure/Make.
1920
1921 5. Applied Craig's patch that moves no_arg into the RE class in the C++ code.
1922 This is an attempt to solve the reported problem "pcrecpp::no_arg is not
1923 exported in the Windows port". It has not yet been confirmed that the patch
1924 solves the problem, but it does no harm.
1925
1926 6. Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
1927 NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
1928 with CMake, and also correct the comment about stack recursion.
1929
1930 7. Remove the automatic building of pcredemo from the ./configure system and
1931 from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
1932 of a program that users should build themselves after PCRE is installed, so
1933 building it automatically is not really right. What is more, it gave
1934 trouble in some build environments.
1935
1936 8. Further tidies to CMakeLists.txt from Sheri and Christian.
1937
1938
1939 Version 7.5 10-Jan-08
1940 ---------------------
1941
1942 1. Applied a patch from Craig: "This patch makes it possible to 'ignore'
1943 values in parens when parsing an RE using the C++ wrapper."
1944
1945 2. Negative specials like \S did not work in character classes in UTF-8 mode.
1946 Characters greater than 255 were excluded from the class instead of being
1947 included.
1948
1949 3. The same bug as (2) above applied to negated POSIX classes such as
1950 [:^space:].
1951
1952 4. PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
1953 defined or documented. It seems to have been a typo for PCRE_STATIC, so
1954 I have changed it.
1955
1956 5. The construct (?&) was not diagnosed as a syntax error (it referenced the
1957 first named subpattern) and a construct such as (?&a) would reference the
1958 first named subpattern whose name started with "a" (in other words, the
1959 length check was missing). Both these problems are fixed. "Subpattern name
1960 expected" is now given for (?&) (a zero-length name), and this patch also
1961 makes it give the same error for \k'' (previously it complained that that
1962 was a reference to a non-existent subpattern).
1963
1964 6. The erroneous patterns (?+-a) and (?-+a) give different error messages;
1965 this is right because (?- can be followed by option settings as well as by
1966 digits. I have, however, made the messages clearer.
1967
1968 7. Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
1969 than the number used in the conditional) now cause a compile-time error.
1970 This is actually not compatible with Perl, which accepts such patterns, but
1971 treats the conditional as always being FALSE (as PCRE used to), but it
1972 seems to me that giving a diagnostic is better.
1973
1974 8. Change "alphameric" to the more common word "alphanumeric" in comments
1975 and messages.
1976
1977 9. Fix two occurrences of "backslash" in comments that should have been
1978 "backspace".
1979
1980 10. Remove two redundant lines of code that can never be obeyed (their function
1981 was moved elsewhere).
1982
1983 11. The program that makes PCRE's Unicode character property table had a bug
1984 which caused it to generate incorrect table entries for sequences of
1985 characters that have the same character type, but are in different scripts.
1986 It amalgamated them into a single range, with the script of the first of
1987 them. In other words, some characters were in the wrong script. There were
1988 thirteen such cases, affecting characters in the following ranges:
1989
1990 U+002b0 - U+002c1
1991 U+0060c - U+0060d
1992 U+0061e - U+00612
1993 U+0064b - U+0065e
1994 U+0074d - U+0076d
1995 U+01800 - U+01805
1996 U+01d00 - U+01d77
1997 U+01d9b - U+01dbf
1998 U+0200b - U+0200f
1999 U+030fc - U+030fe
2000 U+03260 - U+0327f
2001 U+0fb46 - U+0fbb1
2002 U+10450 - U+1049d
2003
2004 12. The -o option (show only the matching part of a line) for pcregrep was not
2005 compatible with GNU grep in that, if there was more than one match in a
2006 line, it showed only the first of them. It now behaves in the same way as
2007 GNU grep.
2008
2009 13. If the -o and -v options were combined for pcregrep, it printed a blank
2010 line for every non-matching line. GNU grep prints nothing, and pcregrep now
2011 does the same. The return code can be used to tell if there were any
2012 non-matching lines.
2013
2014 14. Added --file-offsets and --line-offsets to pcregrep.
2015
2016 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
2017 infinitely looping recursion. The bug was that positive lookaheads were not
2018 being skipped when checking for a possible empty match (negative lookaheads
2019 and both kinds of lookbehind were skipped).
2020
2021 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
2022 inclusion of <windows.h> to before rather than after the definition of
2023 INVALID_FILE_ATTRIBUTES (patch from David Byron).
2024
2025 17. Specifying a possessive quantifier with a specific limit for a Unicode
2026 character property caused pcre_compile() to compile bad code, which led at
2027 runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
2028 are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
2029 caused the error; without that there was no problem.
2030
2031 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
2032
2033 19. Added --enable-pcretest-libreadline.
2034
2035 20. In pcrecpp.cc, the variable 'count' was incremented twice in
2036 RE::GlobalReplace(). As a result, the number of replacements returned was
2037 double what it should be. I removed one of the increments, but Craig sent a
2038 later patch that removed the other one (the right fix) and added unit tests
2039 that check the return values (which was not done before).
2040
2041 21. Several CMake things:
2042
2043 (1) Arranged that, when cmake is used on Unix, the libraries end up with
2044 the names libpcre and libpcreposix, not just pcre and pcreposix.
2045
2046 (2) The above change means that pcretest and pcregrep are now correctly
2047 linked with the newly-built libraries, not previously installed ones.
2048
2049 (3) Added PCRE_SUPPORT_LIBREADLINE, PCRE_SUPPORT_LIBZ, PCRE_SUPPORT_LIBBZ2.
2050
2051 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
2052 crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
2053 UTF-8 newline character). The key issue is that the pattern starts .*;
2054 this means that the match must be either at the beginning, or after a
2055 newline. The bug was in the code for advancing after a failed match and
2056 checking that the new position followed a newline. It was not taking
2057 account of UTF-8 characters correctly.
2058
2059 23. PCRE was behaving differently from Perl in the way it recognized POSIX
2060 character classes. PCRE was not treating the sequence [:...:] as a
2061 character class unless the ... were all letters. Perl, however, seems to
2062 allow any characters between [: and :], though of course it rejects as
2063 unknown any "names" that contain non-letters, because all the known class
2064 names consist only of letters. Thus, Perl gives an error for [[:1234:]],
2065 for example, whereas PCRE did not - it did not recognize a POSIX character
2066 class. This seemed a bit dangerous, so the code has been changed to be
2067 closer to Perl. The behaviour is not identical to Perl, because PCRE will
2068 diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
2069 treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
2070 Perl does, and where it didn't before.
2071
2072 24. Rewrite so as to remove the single use of %n from pcregrep because in some
2073 Windows environments %n is disabled by default.
2074
2075
2076 Version 7.4 21-Sep-07
2077 ---------------------
2078
2079 1. Change 7.3/28 was implemented for classes by looking at the bitmap. This
2080 means that a class such as [\s] counted as "explicit reference to CR or
2081 LF". That isn't really right - the whole point of the change was to try to
2082 help when there was an actual mention of one of the two characters. So now
2083 the change happens only if \r or \n (or a literal CR or LF) character is
2084 encountered.
2085
2086 2. The 32-bit options word was also used for 6 internal flags, but the numbers
2087 of both had grown to the point where there were only 3 bits left.
2088 Fortunately, there was spare space in the data structure, and so I have
2089 moved the internal flags into a new 16-bit field to free up more option
2090 bits.
2091
2092 3. The appearance of (?J) at the start of a pattern set the DUPNAMES option,
2093 but did not set the internal JCHANGED flag - either of these is enough to
2094 control the way the "get" function works - but the PCRE_INFO_JCHANGED
2095 facility is supposed to tell if (?J) was ever used, so now (?J) at the
2096 start sets both bits.
2097
2098 4. Added options (at build time, compile time, exec time) to change \R from
2099 matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
2100
2101 5. doc/pcresyntax.html was missing from the distribution.
2102
2103 6. Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
2104 compatibility, even though it is no longer used.
2105
2106 7. Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
2107 strtoull to pcrecpp.cc to select the available functions in WIN32 when the
2108 windows.h file is present (where different names are used). [This was
2109 reversed later after testing - see 16 below.]
2110
2111 8. Changed all #include <config.h> to #include "config.h". There were also
2112 some further <pcre.h> cases that I changed to "pcre.h".
2113
2114 9. When pcregrep was used with the --colour option, it missed the line ending
2115 sequence off the lines that it output.
2116
2117 10. It was pointed out to me that arrays of string pointers cause lots of
2118 relocations when a shared library is dynamically loaded. A technique of
2119 using a single long string with a table of offsets can drastically reduce
2120 these. I have refactored PCRE in four places to do this. The result is
2121 dramatic:
2122
2123 Originally: 290
2124 After changing UCP table: 187
2125 After changing error message table: 43
2126 After changing table of "verbs" 36
2127 After changing table of Posix names 22
2128
2129 Thanks to the folks working on Gregex for glib for this insight.
2130
2131 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
2132 unicode-properties was also set.
2133
2134 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
2135
2136 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
2137 checked only for CRLF.
2138
2139 14. Added casts to pcretest.c to avoid compiler warnings.
2140
2141 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
2142
2143 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
2144 and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
2145 entirely. This removes changes made in 7 above.
2146
2147 17. The CMake files have been updated, and there is now more information about
2148 building with CMake in the NON-UNIX-USE document.
2149
2150
2151 Version 7.3 28-Aug-07
2152 ---------------------
2153
2154 1. In the rejigging of the build system that eventually resulted in 7.1, the
2155 line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
2156 brackets there is not right, since it causes compilers to look for an
2157 installed pcre.h, not the version that is in the source that is being
2158 compiled (which of course may be different). I have changed it back to:
2159
2160 #include "pcre.h"
2161
2162 I have a vague recollection that the change was concerned with compiling in
2163 different directories, but in the new build system, that is taken care of
2164 by the VPATH setting the Makefile.
2165
2166 2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
2167 when the subject happened to end in the byte 0x85 (e.g. if the last
2168 character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
2169 characters but of course it shouldn't be taken as a newline when it is part
2170 of another character. The bug was that, for an unlimited repeat of . in
2171 not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
2172 characters when looking for a newline.
2173
2174 3. A small performance improvement in the DOTALL UTF-8 mode .* case.
2175
2176 4. Debugging: adjusted the names of opcodes for different kinds of parentheses
2177 in debug output.
2178
2179 5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
2180 long printing in the pcrecpp unittest when running under MinGW.
2181
2182 6. ESC_K was left out of the EBCDIC table.
2183
2184 7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
2185 parentheses; I made it 1000, which seemed large enough. Unfortunately, the
2186 limit also applies to "virtual nesting" when a pattern is recursive, and in
2187 this case 1000 isn't so big. I have been able to remove this limit at the
2188 expense of backing off one optimization in certain circumstances. Normally,
2189 when pcre_exec() would call its internal match() function recursively and
2190 immediately return the result unconditionally, it uses a "tail recursion"
2191 feature to save stack. However, when a subpattern that can match an empty
2192 string has an unlimited repetition quantifier, it no longer makes this
2193 optimization. That gives it a stack frame in which to save the data for
2194 checking that an empty string has been matched. Previously this was taken
2195 from the 1000-entry workspace that had been reserved. So now there is no
2196 explicit limit, but more stack is used.
2197
2198 8. Applied Daniel's patches to solve problems with the import/export magic
2199 syntax that is required for Windows, and which was going wrong for the
2200 pcreposix and pcrecpp parts of the library. These were overlooked when this
2201 problem was solved for the main library.
2202
2203 9. There were some crude static tests to avoid integer overflow when computing
2204 the size of patterns that contain repeated groups with explicit upper
2205 limits. As the maximum quantifier is 65535, the maximum group length was
2206 set at 30,000 so that the product of these two numbers did not overflow a
2207 32-bit integer. However, it turns out that people want to use groups that
2208 are longer than 30,000 bytes (though not repeat them that many times).
2209 Change 7.0/17 (the refactoring of the way the pattern size is computed) has
2210 made it possible to implement the integer overflow checks in a much more
2211 dynamic way, which I have now done. The artificial limitation on group
2212 length has been removed - we now have only the limit on the total length of
2213 the compiled pattern, which depends on the LINK_SIZE setting.
2214
2215 10. Fixed a bug in the documentation for get/copy named substring when
2216 duplicate names are permitted. If none of the named substrings are set, the
2217 functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
2218 empty string.
2219
2220 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
2221 instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
2222 because the ] is interpreted as the first data character and the
2223 terminating ] is not found. PCRE has been made compatible with Perl in this
2224 regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
2225 cause memory overwriting.
2226
2227 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
2228 string has been matched (to stop an infinite loop). It was not recognizing
2229 a conditional subpattern that could match an empty string if that
2230 subpattern was within another subpattern. For example, it looped when
2231 trying to match (((?(1)X|))*) but it was OK with ((?(1)X|)*) where the
2232 condition was not nested. This bug has been fixed.
2233
2234 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
2235 past the start of the subject in the presence of bytes with the top bit
2236 set, for example "\x8aBCD".
2237
2238 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
2239 (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
2240
2241 14. Optimized (?!) to (*FAIL).
2242
2243 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
2244 This restricts code points to be within the range 0 to 0x10FFFF, excluding
2245 the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
2246 full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
2247 does: it's just the validity check that is more restrictive.
2248
2249 16. Inserted checks for integer overflows during escape sequence (backslash)
2250 processing, and also fixed erroneous offset values for syntax errors during
2251 backslash processing.
2252
2253 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
2254 for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
2255
2256 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
2257 caused an overrun.
2258
2259 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
2260 something other than just ASCII characters) inside a group that had an
2261 unlimited repeat caused a loop at compile time (while checking to see
2262 whether the group could match an empty string).
2263
2264 20. Debugging a pattern containing \p or \P could cause a crash. For example,
2265 [\P{Any}] did so. (Error in the code for printing property names.)
2266
2267 21. An orphan \E inside a character class could cause a crash.
2268
2269 22. A repeated capturing bracket such as (A)? could cause a wild memory
2270 reference during compilation.
2271
2272 23. There are several functions in pcre_compile() that scan along a compiled
2273 expression for various reasons (e.g. to see if it's fixed length for look
2274 behind). There were bugs in these functions when a repeated \p or \P was
2275 present in the pattern. These operators have additional parameters compared
2276 with \d, etc, and these were not being taken into account when moving along
2277 the compiled data. Specifically:
2278
2279 (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
2280 length.
2281
2282 (b) An item such as \pL+ within a repeated group could cause crashes or
2283 loops.
2284
2285 (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
2286 "reference to non-existent subpattern" error.
2287
2288 (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
2289
2290 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
2291 characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
2292
2293 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
2294
2295 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
2296 character were causing crashes (broken optimization).
2297
2298 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
2299 \p or \P) caused a compile-time loop.
2300
2301 28. More problems have arisen in unanchored patterns when CRLF is a valid line
2302 break. For example, the unstudied pattern [\r\n]A does not match the string
2303 "\r\nA" because change 7.0/46 below moves the current point on by two
2304 characters after failing to match at the start. However, the pattern \nA
2305 *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
2306 the same is true. There doesn't seem any very clean way out of this, but
2307 what I have chosen to do makes the common cases work: PCRE now takes note
2308 of whether there can be an explicit match for \r or \n anywhere in the
2309 pattern, and if so, 7.0/46 no longer applies. As part of this change,
2310 there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
2311 pattern has explicit CR or LF references.
2312
2313 29. Added (*CR) etc for changing newline setting at start of pattern.
2314
2315
2316 Version 7.2 19-Jun-07
2317 ---------------------
2318
2319 1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
2320 which is apparently normally available under Windows.
2321
2322 2. Re-jig the pcregrep tests with different newline settings in an attempt
2323 to make them independent of the local environment's newline setting.
2324
2325 3. Add code to configure.ac to remove -g from the CFLAGS default settings.
2326
2327 4. Some of the "internals" tests were previously cut out when the link size
2328 was not 2, because the output contained actual offsets. The recent new
2329 "Z" feature of pcretest means that these can be cut out, making the tests
2330 usable with all link sizes.
2331
2332 5. Implemented Stan Switzer's goto replacement for longjmp() when not using
2333 stack recursion. This gives a massive performance boost under BSD, but just
2334 a small improvement under Linux. However, it saves one field in the frame
2335 in all cases.
2336
2337 6. Added more features from the forthcoming Perl 5.10:
2338
2339 (a) (?-n) (where n is a string of digits) is a relative subroutine or
2340 recursion call. It refers to the nth most recently opened parentheses.
2341
2342 (b) (?+n) is also a relative subroutine call; it refers to the nth next
2343 to be opened parentheses.
2344
2345 (c) Conditions that refer to capturing parentheses can be specified
2346 relatively, for example, (?(-2)... or (?(+3)...
2347
2348 (d) \K resets the start of the current match so that everything before
2349 is not part of it.
2350
2351 (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
2352
2353 (f) \g{name} is another synonym - part of Perl 5.10's unification of
2354 reference syntax.
2355
2356 (g) (?| introduces a group in which the numbering of parentheses in each
2357 alternative starts with the same number.
2358
2359 (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
2360
2361 7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
2362 PCRE_INFO_JCHANGED.
2363
2364 8. A pattern such as (.*(.)?)* caused pcre_exec() to fail by either not
2365 terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
2366 for detecting groups that can match an empty string.
2367
2368 9. A pattern with a very large number of alternatives (more than several
2369 hundred) was running out of internal workspace during the pre-compile
2370 phase, where pcre_compile() figures out how much memory will be needed. A
2371 bit of new cunning has reduced the workspace needed for groups with
2372 alternatives. The 1000-alternative test pattern now uses 12 bytes of
2373 workspace instead of running out of the 4096 that are available.
2374
2375 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
2376
2377 11. Applied patch from Google to remove an optimization that didn't quite work.
2378 The report of the bug said:
2379
2380 pcrecpp::RE("a*").FullMatch("aaa") matches, while
2381 pcrecpp::RE("a*?").FullMatch("aaa") does not, and
2382 pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
2383
2384 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
2385 it matched the wrong number of bytes.
2386
2387
2388 Version 7.1 24-Apr-07
2389 ---------------------
2390
2391 1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
2392 that is more "standard", making use of automake and other Autotools. There
2393 is some re-arrangement of the files and adjustment of comments consequent
2394 on this.
2395
2396 2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
2397 for recursive directory scanning broke on some systems because the files
2398 are not scanned in any specific order and on different systems the order
2399 was different. A call to "sort" has been inserted into RunGrepTest for the
2400 approprate test as a short-term fix. In the longer term there may be an
2401 alternative.
2402
2403 3. I had an email from Eric Raymond about problems translating some of PCRE's
2404 man pages to HTML (despite the fact that I distribute HTML pages, some
2405 people do their own conversions for various reasons). The problems
2406 concerned the use of low-level troff macros .br and .in. I have therefore
2407 removed all such uses from the man pages (some were redundant, some could
2408 be replaced by .nf/.fi pairs). The 132html script that I use to generate
2409 HTML has been updated to handle .nf/.fi and to complain if it encounters
2410 .br or .in.
2411
2412 4. Updated comments in configure.ac that get placed in config.h.in and also
2413 arranged for config.h to be included in the distribution, with the name
2414 config.h.generic, for the benefit of those who have to compile without
2415 Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
2416
2417 5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
2418 Weber: (1) pcre_internal.h was missing some function renames; (2) updated
2419 makevp.bat for the current PCRE, using the additional files
2420 makevp_c.txt, makevp_l.txt, and pcregexp.pas.
2421
2422 6. A Windows user reported a minor discrepancy with test 2, which turned out
2423 to be caused by a trailing space on an input line that had got lost in his
2424 copy. The trailing space was an accident, so I've just removed it.
2425
2426 7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
2427 that is needed.
2428
2429 8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
2430 as "const" (a) because they are and (b) because it helps the PHP
2431 maintainers who have recently made a script to detect big data structures
2432 in the php code that should be moved to the .rodata section. I remembered
2433 to update Builducptable as well, so it won't revert if ucptable.h is ever
2434 re-created.
2435
2436 9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
2437 pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
2438 order to be able to cut out the UTF-8 tables in the latter when UTF-8
2439 support is not required. This saves 1.5-2K of code, which is important in
2440 some applications.
2441
2442 Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
2443 so as not to refer to the tables, even though these functions will never be
2444 called when UTF-8 support is disabled. Otherwise there are problems with a
2445 shared library.
2446
2447 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
2448
2449 (a) It was defining its arguments as char * instead of void *.
2450
2451 (b) It was assuming that all moves were upwards in memory; this was true
2452 a long time ago when I wrote it, but is no longer the case.
2453
2454 The emulated memove() is provided for those environments that have neither
2455 memmove() nor bcopy(). I didn't think anyone used it these days, but that
2456 is clearly not the case, as these two bugs were recently reported.
2457
2458 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
2459 and Detrail to create the HTML documentation, the .txt form of the man
2460 pages, and it removes trailing spaces from listed files. It also creates
2461 pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
2462 case, it wraps all the #defines with #ifndefs. This script should be run
2463 before "make dist".
2464
2465 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
2466 with Unicode property support.
2467
2468 (a) For a maximizing quantifier, if the two different cases of the
2469 character were of different lengths in their UTF-8 codings (there are
2470 some cases like this - I found 11), and the matching function had to
2471 back up over a mixture of the two cases, it incorrectly assumed they
2472 were both the same length.
2473
2474 (b) When PCRE was configured to use the heap rather than the stack for
2475 recursion during matching, it was not correctly preserving the data for
2476 the other case of a UTF-8 character when checking ahead for a match
2477 while processing a minimizing repeat. If the check also involved
2478 matching a wide character, but failed, corruption could cause an
2479 erroneous result when trying to check for a repeat of the original
2480 character.
2481
2482 13. Some tidying changes to the testing mechanism:
2483
2484 (a) The RunTest script now detects the internal link size and whether there
2485 is UTF-8 and UCP support by running ./pcretest -C instead of relying on
2486 values substituted by "configure". (The RunGrepTest script already did
2487 this for UTF-8.) The configure.ac script no longer substitutes the
2488 relevant variables.
2489
2490 (b) The debugging options /B and /D in pcretest show the compiled bytecode
2491 with length and offset values. This means that the output is different
2492 for different internal link sizes. Test 2 is skipped for link sizes
2493 other than 2 because of this, bypassing the problem. Unfortunately,
2494 there was also a test in test 3 (the locale tests) that used /B and
2495 failed for link sizes other than 2. Rather than cut the whole test out,
2496 I have added a new /Z option to pcretest that replaces the length and
2497 offset values with spaces. This is now used to make test 3 independent
2498 of link size. (Test 2 will be tidied up later.)
2499
2500 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
2501 segmentation fault instead of returning the appropriate error message.
2502
2503 15. In multiline mode when the newline sequence was set to "any", the pattern
2504 ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
2505 This doesn't seem right; it now treats the CRLF combination as the line
2506 ending, and so does not match in that case. It's only a pattern such as ^$
2507 that would hit this one: something like ^ABC$ would have failed after \r
2508 and then tried again after \r\n.
2509
2510 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
2511 in an attempt to make files that differ only in their line terminators
2512 compare equal. This works on Linux.
2513
2514 17. Under certain error circumstances pcregrep might try to free random memory
2515 as it exited. This is now fixed, thanks to valgrind.
2516
2517 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
2518 "abc\r\n\r\n", it found an unwanted second match after the second \r. This
2519 was because its rules for how to advance for /g after matching an empty
2520 string at the end of a line did not allow for this case. They now check for
2521 it specially.
2522
2523 20. pcretest is supposed to handle patterns and data of any length, by
2524 extending its buffers when necessary. It was getting this wrong when the
2525 buffer for a data line had to be extended.
2526
2527 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
2528 CRLF as a newline sequence.
2529
2530 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
2531 out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
2532 I have nevertheless tidied it up.
2533
2534 23. Added some casts to kill warnings from HP-UX ia64 compiler.
2535
2536 24. Added a man page for pcre-config.
2537
2538
2539 Version 7.0 19-Dec-06
2540 ---------------------
2541
2542 1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
2543 moving to gcc 4.1.1.
2544
2545 2. The -S option for pcretest uses setrlimit(); I had omitted to #include
2546 sys/time.h, which is documented as needed for this function. It doesn't
2547 seem to matter on Linux, but it showed up on some releases of OS X.
2548
2549 3. It seems that there are systems where bytes whose values are greater than
2550 127 match isprint() in the "C" locale. The "C" locale should be the
2551 default when a C program starts up. In most systems, only ASCII printing
2552 characters match isprint(). This difference caused the output from pcretest
2553 to vary, making some of the tests fail. I have changed pcretest so that:
2554
2555 (a) When it is outputting text in the compiled version of a pattern, bytes
2556 other than 32-126 are always shown as hex escapes.
2557
2558 (b) When it is outputting text that is a matched part of a subject string,
2559 it does the same, unless a different locale has been set for the match
2560 (using the /L modifier). In this case, it uses isprint() to decide.
2561
2562 4. Fixed a major bug that caused incorrect computation of the amount of memory
2563 required for a compiled pattern when options that changed within the
2564 pattern affected the logic of the preliminary scan that determines the
2565 length. The relevant options are -x, and -i in UTF-8 mode. The result was
2566 that the computed length was too small. The symptoms of this bug were
2567 either the PCRE error "internal error: code overflow" from pcre_compile(),
2568 or a glibc crash with a message such as "pcretest: free(): invalid next
2569 size (fast)". Examples of patterns that provoked this bug (shown in
2570 pcretest format) are:
2571
2572 /(?-x: )/x
2573 /(?x)(?-x: \s*#\s*)/
2574 /((?i)[\x{c0}])/8
2575 /(?i:[\x{c0}])/8
2576
2577 HOWEVER: Change 17 below makes this fix obsolete as the memory computation
2578 is now done differently.
2579
2580 5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
2581 wrapper classes; (b) implement a new function in the C++ scanner that is
2582 more efficient than the old way of doing things because it avoids levels of
2583 recursion in the regex matching; (c) add a paragraph to the documentation
2584 for the FullMatch() function.
2585
2586 6. The escape sequence \n was being treated as whatever was defined as
2587 "newline". Not only was this contrary to the documentation, which states
2588 that \n is character 10 (hex 0A), but it also went horribly wrong when
2589 "newline" was defined as CRLF. This has been fixed.
2590
2591 7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
2592 was being set to -1 for the "end of line" case (supposedly a value that no
2593 character can have). Though this value is never used (the check for end of
2594 line is "zero bytes in current character"), it caused compiler complaints.
2595 I've changed it to 0xffffffff.
2596
2597 8. In pcre_version.c, the version string was being built by a sequence of
2598 C macros that, in the event of PCRE_PRERELEASE being defined as an empty
2599 string (as it is for production releases) called a macro with an empty
2600 argument. The C standard says the result of this is undefined. The gcc
2601 compiler treats it as an empty string (which was what was wanted) but it is
2602 reported that Visual C gives an error. The source has been hacked around to
2603 avoid this problem.
2604
2605 9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
2606 builds of pcretest, and changed the call to _setmode() to use _O_BINARY
2607 instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
2608 of them did).
2609
2610 10. Originally, pcretest opened its input and output without "b"; then I was
2611 told that "b" was needed in some environments, so it was added for release
2612 5.0 to both the input and output. (It makes no difference on Unix-like
2613 systems.) Later I was told that it is wrong for the input on Windows. I've
2614 now abstracted the modes into two macros, to make it easier to fiddle with
2615 them, and removed "b" from the input mode under Windows.
2616
2617 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
2618
2619 12. Added -help and --help to pcretest as an official way of being reminded
2620 of the options.
2621
2622 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
2623 and pcrecpp.cc because they annoy compilers at high warning levels.
2624
2625 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
2626
2627 15. Fixed an occurrence of == in configure.ac that should have been = (shell
2628 scripts are not C programs :-) and which was not noticed because it works
2629 on Linux.
2630
2631 16. pcretest is supposed to handle any length of pattern and data line (as one
2632 line or as a continued sequence of lines) by extending its input buffer if
2633 necessary. This feature was broken for very long pattern lines, leading to
2634 a string of junk being passed to pcre_compile() if the pattern was longer
2635 than about 50K.
2636
2637 17. I have done a major re-factoring of the way pcre_compile() computes the
2638 amount of memory needed for a compiled pattern. Previously, there was code
2639 that made a preliminary scan of the pattern in order to do this. That was
2640 OK when PCRE was new, but as the facilities have expanded, it has become
2641 harder and harder to keep it in step with the real compile phase, and there
2642 have been a number of bugs (see for example, 4 above). I have now found a
2643 cunning way of running the real compile function in a "fake" mode that
2644 enables it to compute how much memory it would need, while actually only
2645 ever using a few hundred bytes of working memory and without too many
2646 tests of the mode. This should make future maintenance and development
2647 easier. A side effect of this work is that the limit of 200 on the nesting
2648 depth of parentheses has been removed (though this was never a serious
2649 limitation, I suspect). However, there is a downside: pcre_compile() now
2650 runs more slowly than before (30% or more, depending on the pattern). I
2651 hope this isn't a big issue. There is no effect on runtime performance.
2652
2653 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
2654 newline (only possible for the last line of a file) and it was a
2655 pattern that set a locale (followed by /Lsomething), pcretest crashed.
2656
2657 19. Added additional timing features to pcretest. (1) The -tm option now times
2658 matching only, not compiling. (2) Both -t and -tm can be followed, as a
2659 separate command line item, by a number that specifies the number of
2660 repeats to use when timing. The default is 50000; this gives better
2661 precision, but takes uncomfortably long for very large patterns.
2662
2663 20. Extended pcre_study() to be more clever in cases where a branch of a
2664 subpattern has no definite first character. For example, (a*|b*)[cd] would
2665 previously give no result from pcre_study(). Now it recognizes that the
2666 first character must be a, b, c, or d.
2667
2668 21. There was an incorrect error "recursive call could loop indefinitely" if
2669 a subpattern (or the entire pattern) that was being tested for matching an
2670 empty string contained only one non-empty item after a nested subpattern.
2671 For example, the pattern (?>\x{100}*)\d(?R) provoked this error
2672 incorrectly, because the \d was being skipped in the check.
2673
2674 22. The pcretest program now has a new pattern option /B and a command line
2675 option -b, which is equivalent to adding /B to every pattern. This causes
2676 it to show the compiled bytecode, without the additional information that
2677 -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
2678 is the same as /B/I).
2679
2680 23. A new optimization is now able automatically to treat some sequences such
2681 as a*b as a*+b. More specifically, if something simple (such as a character
2682 or a simple class like \d) has an unlimited quantifier, and is followed by
2683 something that cannot possibly match the quantified thing, the quantifier
2684 is automatically "possessified".
2685
2686 24. A recursive reference to a subpattern whose number was greater than 39
2687 went wrong under certain circumstances in UTF-8 mode. This bug could also
2688 have affected the operation of pcre_study().
2689
2690 25. Realized that a little bit of performance could be had by replacing
2691 (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
2692
2693 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
2694
2695 27. Possessive quantifiers such as a++ were previously implemented by turning
2696 them into atomic groups such as ($>a+). Now they have their own opcodes,
2697 which improves performance. This includes the automatically created ones
2698 from 23 above.
2699
2700 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
2701 lookahead was broken if it was not anchored. PCRE was mistakenly expecting
2702 the first matched character to be a colon. This applied both to named and
2703 numbered groups.
2704
2705 29. The ucpinternal.h header file was missing its idempotency #ifdef.
2706
2707 30. I was sent a "project" file called libpcre.a.dev which I understand makes
2708 building PCRE on Windows easier, so I have included it in the distribution.
2709
2710 31. There is now a check in pcretest against a ridiculously large number being
2711 returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
2712 loop, the loop is abandoned.
2713
2714 32. Forward references to subpatterns in conditions such as (?(2)...) where
2715 subpattern 2 is defined later cause pcre_compile() to search forwards in
2716 the pattern for the relevant set of parentheses. This search went wrong
2717 when there were unescaped parentheses in a character class, parentheses
2718 escaped with \Q...\E, or parentheses in a #-comment in /x mode.
2719
2720 33. "Subroutine" calls and backreferences were previously restricted to
2721 referencing subpatterns earlier in the regex. This restriction has now
2722 been removed.
2723
2724 34. Added a number of extra features that are going to be in Perl 5.10. On the
2725 whole, these are just syntactic alternatives for features that PCRE had
2726 previously implemented using the Python syntax or my own invention. The
2727 other formats are all retained for compatibility.
2728
2729 (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
2730 as (?P<name>...). The new forms, as well as being in Perl 5.10, are
2731 also .NET compatible.
2732
2733 (b) A recursion or subroutine call to a named group can now be defined as
2734 (?&name) as well as (?P>name).
2735
2736 (c) A backreference to a named group can now be defined as \k<name> or
2737 \k'name' as well as (?P=name). The new forms, as well as being in Perl
2738 5.10, are also .NET compatible.
2739
2740 (d) A conditional reference to a named group can now use the syntax
2741 (?(<name>) or (?('name') as well as (?(name).
2742
2743 (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
2744 groups (named and numbered) that are never evaluated inline, but can be
2745 called as "subroutines" from elsewhere. In effect, the DEFINE condition
2746 is always false. There may be only one alternative in such a group.
2747
2748 (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
2749 as the simple (?(R). The condition is true only if the most recent
2750 recursion is that of the given number or name. It does not search out
2751 through the entire recursion stack.
2752
2753 (g) The escape \gN or \g{N} has been added, where N is a positive or
2754 negative number, specifying an absolute or relative reference.
2755
2756 35. Tidied to get rid of some further signed/unsigned compiler warnings and
2757 some "unreachable code" warnings.
2758
2759 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
2760 things, this adds five new scripts.
2761
2762 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
2763 There were also incompatibilities regarding the handling of \Q..\E inside
2764 character classes, for example with patterns like [\Qa\E-\Qz\E] where the
2765 hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
2766
2767 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
2768 matches an empty string, and forcibly breaks the loop. There were bugs in
2769 this code in non-simple cases. For a pattern such as ^(a()*)* matched
2770 against aaaa the result was just "a" rather than "aaaa", for example. Two
2771 separate and independent bugs (that affected different cases) have been
2772 fixed.
2773
2774 39. Refactored the code to abolish the use of different opcodes for small
2775 capturing bracket numbers. This is a tidy that I avoided doing when I
2776 removed the limit on the number of capturing brackets for 3.5 back in 2001.
2777 The new approach is not only tidier, it makes it possible to reduce the
2778 memory needed to fix the previous bug (38).
2779
2780 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
2781 sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
2782 processing dot, circumflex, or dollar metacharacters, or #-comments in /x
2783 mode.
2784
2785 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
2786 report.
2787
2788 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
2789 copy construction and assignment in the C++ wrapper.
2790
2791 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
2792 couple of bugs that could have given wrong results in the "--newline=crlf"
2793 case.
2794
2795 44. Added a number of casts and did some reorganization of signed/unsigned int
2796 variables following suggestions from Dair Grant. Also renamed the variable
2797 "this" as "item" because it is a C++ keyword.
2798
2799 45. Arranged for dftables to add
2800
2801 #include "pcre_internal.h"
2802
2803 to pcre_chartables.c because without it, gcc 4.x may remove the array
2804 definition from the final binary if PCRE is built into a static library and
2805 dead code stripping is activated.
2806
2807 46. For an unanchored pattern, if a match attempt fails at the start of a
2808 newline sequence, and the newline setting is CRLF or ANY, and the next two
2809 characters are CRLF, advance by two characters instead of one.
2810
2811
2812 Version 6.7 04-Jul-06
2813 ---------------------
2814
2815 1. In order to handle tests when input lines are enormously long, pcretest has
2816 been re-factored so that it automatically extends its buffers when
2817 necessary. The code is crude, but this _is_ just a test program. The
2818 default size has been increased from 32K to 50K.
2819
2820 2. The code in pcre_study() was using the value of the re argument before
2821 testing it for NULL. (Of course, in any sensible call of the function, it
2822 won't be NULL.)
2823
2824 3. The memmove() emulation function in pcre_internal.h, which is used on
2825 systems that lack both memmove() and bcopy() - that is, hardly ever -
2826 was missing a "static" storage class specifier.
2827
2828 4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
2829 containing an extended class (one that cannot be represented by a bitmap
2830 because it contains high-valued characters or Unicode property items, e.g.
2831 [\pZ]). Almost always one would set UTF-8 mode when processing such a
2832 pattern, but PCRE should not loop if you do not (it no longer does).
2833 [Detail: two cases were found: (a) a repeated subpattern containing an
2834 extended class; (b) a recursive reference to a subpattern that followed a
2835 previous extended class. It wasn't skipping over the extended class
2836 correctly when UTF-8 mode was not set.]
2837
2838 5. A negated single-character class was not being recognized as fixed-length
2839 in lookbehind assertions such as (?<=[^f]), leading to an incorrect
2840 compile error "lookbehind assertion is not fixed length".
2841
2842 6. The RunPerlTest auxiliary script was showing an unexpected difference
2843 between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
2844 write a Perl script that can interpret lines of an input file either as
2845 byte characters or as UTF-8, which is what "perltest" was being required to
2846 do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
2847 can't do is switch easily at run time between having the "use utf8;" pragma
2848 or not. In the end, I fudged it by using the RunPerlTest script to insert
2849 "use utf8;" explicitly for the UTF-8 tests.
2850
2851 7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
2852 the end of the subject string, contrary to the documentation and to what
2853 Perl does. This was true of both matching functions. Now it matches only at
2854 the start of the subject and immediately after *internal* newlines.
2855
2856 8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
2857 a pointer to an int instead of a pointer to an unsigned long int. This
2858 caused problems on 64-bit systems.
2859
2860 9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
2861 instance of the 'standard' template library not being so standard".
2862
2863 10. There was no check on the number of named subpatterns nor the maximum
2864 length of a subpattern name. The product of these values is used to compute
2865 the size of the memory block for a compiled pattern. By supplying a very
2866 long subpattern name and a large number of named subpatterns, the size
2867 computation could be caused to overflow. This is now prevented by limiting
2868 the length of names to 32 characters, and the number of named subpatterns
2869 to 10,000.
2870
2871 11. Subpatterns that are repeated with specific counts have to be replicated in
2872 the compiled pattern. The size of memory for this was computed from the
2873 length of the subpattern and the repeat count. The latter is limited to
2874 65535, but there was no limit on the former, meaning that integer overflow
2875 could in principle occur. The compiled length of a repeated subpattern is
2876 now limited to 30,000 bytes in order to prevent this.
2877
2878 12. Added the optional facility to have named substrings with the same name.
2879
2880 13. Added the ability to use a named substring as a condition, using the
2881 Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
2882 are numbers (not recommended). Forward references are permitted.
2883
2884 14. Added forward references in named backreferences (if you see what I mean).
2885
2886 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
2887 pattern could run off the end of the subject. For example, the pattern
2888 "(?s)(.{1,5})"8 did this with the subject "ab".
2889
2890 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
2891 PCRE_CASELESS was set when matching characters that were quantified with ?
2892 or *.
2893
2894 17. A character class other than a single negated character that had a minimum
2895 but no maximum quantifier - for example [ab]{6,} - was not handled
2896 correctly by pce_dfa_exec(). It would match only one character.
2897
2898 18. A valid (though odd) pattern that looked like a POSIX character
2899 class but used an invalid character after [ (for example [[,abc,]]) caused
2900 pcre_compile() to give the error "Failed: internal error: code overflow" or
2901 in some cases to crash with a glibc free() error. This could even happen if
2902 the pattern terminated after [[ but there just happened to be a sequence of
2903 letters, a binary zero, and a closing ] in the memory that followed.
2904
2905 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
2906 over the years. Originally (before any Unicode support), just the bottom 8
2907 bits were taken. Thus, for example, \500 really meant \100. Nowadays the
2908 output from "man perlunicode" includes this:
2909
2910 The regular expression compiler produces polymorphic opcodes. That
2911 is, the pattern adapts to the data and automatically switches to
2912 the Unicode character scheme when presented with Unicode data--or
2913 instead uses a traditional byte scheme when presented with byte
2914 data.
2915
2916 Sadly, a wide octal escape does not cause a switch, and in a string with
2917 no other multibyte characters, these octal escapes are treated as before.
2918 Thus, in Perl, the pattern /\500/ actually matches \100 but the pattern
2919 /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
2920 Unicode string.
2921
2922 I have not perpetrated such confusion in PCRE. Up till now, it took just
2923 the bottom 8 bits, as in old Perl. I have now made octal escapes with
2924 values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
2925 translate to the appropriate multibyte character.
2926
2927 29. Applied some refactoring to reduce the number of warnings from Microsoft
2928 and Borland compilers. This has included removing the fudge introduced
2929 seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
2930 a warning about an unused variable.
2931
2932 21. PCRE has not included VT (character 0x0b) in the set of whitespace
2933 characters since release 4.0, because Perl (from release 5.004) does not.
2934 [Or at least, is documented not to: some releases seem to be in conflict
2935 with the documentation.] However, when a pattern was studied with
2936 pcre_study() and all its branches started with \s, PCRE still included VT
2937 as a possible starting character. Of course, this did no harm; it just
2938 caused an unnecessary match attempt.
2939
2940 22. Removed a now-redundant internal flag bit that recorded the fact that case
2941 dependency changed within the pattern. This was once needed for "required
2942 byte" processing, but is no longer used. This recovers a now-scarce options
2943 bit. Also moved the least significant internal flag bit to the most-
2944 significant bit of the word, which was not previously used (hangover from
2945 the days when it was an int rather than a uint) to free up another bit for
2946 the future.
2947
2948 23. Added support for CRLF line endings as well as CR and LF. As well as the
2949 default being selectable at build time, it can now be changed at runtime
2950 via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
2951 specify that it is scanning data with non-default line endings.
2952
2953 24. Changed the definition of CXXLINK to make it agree with the definition of
2954 LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
2955
2956 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
2957 recursions. This makes a big different to stack usage for some patterns.
2958
2959 26. If a subpattern containing a named recursion or subroutine reference such
2960 as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
2961 the space required for the compiled pattern went wrong and gave too small a
2962 value. Depending on the environment, this could lead to "Failed: internal
2963 error: code overflow at offset 49" or "glibc detected double free or
2964 corruption" errors.
2965
2966 27. Applied patches from Google (a) to support the new newline modes and (b) to
2967 advance over multibyte UTF-8 characters in GlobalReplace.
2968
2969 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
2970 difference for some implementation of PCRE in some Windows version.
2971
2972 29. Added some extra testing facilities to pcretest:
2973
2974 \q<number> in a data line sets the "match limit" value
2975 \Q<number> in a data line sets the "match recursion limt" value
2976 -S <number> sets the stack size, where <number> is in megabytes
2977
2978 The -S option isn't available for Windows.
2979
2980
2981 Version 6.6 06-Feb-06
2982 ---------------------
2983
2984 1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
2985 in pcreposix.h. I have copied the definition from pcre.h.
2986
2987 2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
2988 because pcre.h is no longer a built file.
2989
2990 3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
2991 not normally included in the compiled code.
2992
2993
2994 Version 6.5 01-Feb-06
2995 ---------------------
2996
2997 1. When using the partial match feature with pcre_dfa_exec(), it was not
2998 anchoring the second and subsequent partial matches at the new starting
2999 point. This could lead to incorrect results. For example, with the pattern
3000 /1234/, partially matching against "123" and then "a4" gave a match.
3001
3002 2. Changes to pcregrep:
3003
3004 (a) All non-match returns from pcre_exec() were being treated as failures
3005 to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
3006 error message is output. Some extra information is given for the
3007 PCRE_ERROR_MATCHLIMIT and PCRE_ERROR_RECURSIONLIMIT errors, which are
3008 probably the only errors that are likely to be caused by users (by
3009 specifying a regex that has nested indefinite repeats, for instance).
3010 If there are more than 20 of these errors, pcregrep is abandoned.
3011
3012 (b) A binary zero was treated as data while matching, but terminated the
3013 output line if it was written out. This has been fixed: binary zeroes
3014 are now no different to any other data bytes.
3015
3016 (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
3017 used to set a locale for matching. The --locale=xxxx long option has
3018 been added (no short equivalent) to specify a locale explicitly on the
3019 pcregrep command, overriding the environment variables.
3020
3021 (d) When -B was used with -n, some line numbers in the output were one less
3022 than they should have been.
3023
3024 (e) Added the -o (--only-matching) option.
3025
3026 (f) If -A or -C was used with -c (count only), some lines of context were
3027 accidentally printed for the final match.
3028
3029 (g) Added the -H (--with-filename) option.
3030
3031 (h) The combination of options -rh failed to suppress file names for files
3032 that were found from directory arguments.
3033
3034 (i) Added the -D (--devices) and -d (--directories) options.
3035
3036 (j) Added the -F (--fixed-strings) option.
3037
3038 (k) Allow "-" to be used as a file name for -f as well as for a data file.
3039
3040 (l) Added the --colo(u)r option.
3041
3042 (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
3043 is not present by default.
3044
3045 3. A nasty bug was discovered in the handling of recursive patterns, that is,
3046 items such as (?R) or (?1), when the recursion could match a number of
3047 alternatives. If it matched one of the alternatives, but subsequently,
3048 outside the recursion, there was a failure, the code tried to back up into
3049 the recursion. However, because of the way PCRE is implemented, this is not
3050 possible, and the result was an incorrect result from the match.
3051
3052 In order to prevent this happening, the specification of recursion has
3053 been changed so that all such subpatterns are automatically treated as
3054 atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
3055
3056 4. I had overlooked the fact that, in some locales, there are characters for
3057 which isalpha() is true but neither isupper() nor islower() are true. In
3058 the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
3059 and ordfeminine) are like this. This affected the treatment of \w and \W
3060 when they appeared in character classes, but not when they appeared outside
3061 a character class. The bit map for "word" characters is now created
3062 separately from the results of isalnum() instead of just taking it from the
3063 upper, lower, and digit maps. (Plus the underscore character, of course.)
3064
3065 5. The above bug also affected the handling of POSIX character classes such as
3066 [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
3067 permanent tables. Instead, the bit maps for such a class were previously
3068 created as the appropriate unions of the upper, lower, and digit bitmaps.
3069 Now they are created by subtraction from the [[:word:]] class, which has
3070 its own bitmap.
3071
3072 6. The [[:blank:]] character class matches horizontal, but not vertical space.
3073 It is created by subtracting the vertical space characters (\x09, \x0a,
3074 \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
3075 subtraction was done in the overall bitmap for a character class, meaning
3076 that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
3077 be recognized. This bug has been fixed.
3078
3079 7. Patches from the folks at Google:
3080
3081 (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
3082 real life, but is still worth protecting against".
3083
3084 (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
3085 regular expressions".
3086
3087 (c) pcre_scanner.cc: avoid use of std::count() because not all systems
3088 have it.
3089
3090 (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
3091 "configure" and the latter not, in order to fix a problem somebody had
3092 with compiling the Arg class on HP-UX.
3093
3094 (e) Improve the error-handling of the C++ wrapper a little bit.
3095
3096 (f) New tests for checking recursion limiting.
3097
3098 8. The pcre_memmove() function, which is used only if the environment does not
3099 have a standard memmove() function (and is therefore rarely compiled),
3100 contained two bugs: (a) use of int instead of size_t, and (b) it was not
3101 returning a result (though PCRE never actually uses the result).
3102
3103 9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
3104 large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
3105 returned instead of calling malloc() with an overflowing number that would
3106 most likely cause subsequent chaos.
3107
3108 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
3109
3110 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
3111 with this option is matched, the nmatch and pmatch options of regexec() are
3112 ignored.
3113
3114 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
3115 provided in case anyone wants to the the POSIX interface with UTF-8
3116 strings.
3117
3118 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
3119 C++ linking (needed for some HP-UX environments).
3120
3121 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
3122 (unused parameter) and in the pcre_printint() function (omitted "default"
3123 switch label when the default is to do nothing).
3124
3125 15. Added some code to make it possible, when PCRE is compiled as a C++
3126 library, to replace subject pointers for pcre_exec() with a smart pointer
3127 class, thus making it possible to process discontinuous strings.
3128
3129 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
3130 much the same function. They were added by different people who were trying
3131 to make PCRE easy to compile on non-Unix systems. It has been suggested
3132 that PCRE_EXPORT be abolished now that there is more automatic apparatus
3133 for compiling on Windows systems. I have therefore replaced it with
3134 PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
3135 defaults to "extern" for C or "extern C" for C++, which works fine on
3136 Unix-like systems. It is now possible to override the value of PCRE_DATA_
3137 SCOPE with something explicit in config.h. In addition:
3138
3139 (a) pcreposix.h still had just "extern" instead of either of these macros;
3140 I have replaced it with PCRE_DATA_SCOPE.
3141
3142 (b) Functions such as _pcre_xclass(), which are internal to the library,
3143 but external in the C sense, all had PCRE_EXPORT in their definitions.
3144 This is apparently wrong for the Windows case, so I have removed it.
3145 (It makes no difference on Unix-like systems.)
3146
3147 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
3148 of recursive calls to match(). This is different to MATCH_LIMIT because
3149 that limits the total number of calls to match(), not all of which increase
3150 the depth of recursion. Limiting the recursion depth limits the amount of
3151 stack (or heap if NO_RECURSE is set) that is used. The default can be set
3152 when PCRE is compiled, and changed at run time. A patch from Google adds
3153 this functionality to the C++ interface.
3154
3155 18. Changes to the handling of Unicode character properties:
3156
3157 (a) Updated the table to Unicode 4.1.0.
3158
3159 (b) Recognize characters that are not in the table as "Cn" (undefined).
3160
3161 (c) I revised the way the table is implemented to a much improved format
3162 which includes recognition of ranges. It now supports the ranges that
3163 are defined in UnicodeData.txt, and it also amalgamates other
3164 characters into ranges. This has reduced the number of entries in the
3165 table from around 16,000 to around 3,000, thus reducing its size
3166 considerably. I realized I did not need to use a tree structure after
3167 all - a binary chop search is just as efficient. Having reduced the
3168 number of entries, I extended their size from 6 bytes to 8 bytes to
3169 allow for more data.
3170
3171 (d) Added support for Unicode script names via properties such as \p{Han}.
3172
3173 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
3174 matching that character.
3175
3176 20. When matching a repeated Unicode property with a minimum greater than zero,
3177 (for example \pL{2,}), PCRE could look past the end of the subject if it
3178 reached it while seeking the minimum number of characters. This could
3179 happen only if some of the characters were more than one byte long, because
3180 there is a check for at least the minimum number of bytes.
3181
3182 21. Refactored the implementation of \p and \P so as to be more general, to
3183 allow for more different types of property in future. This has changed the
3184 compiled form incompatibly. Anybody with saved compiled patterns that use
3185 \p or \P will have to recompile them.
3186
3187 22. Added "Any" and "L&" to the supported property types.
3188
3189 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
3190 but give a compile time error if the value is greater than 0xff.
3191
3192 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
3193 accidentally not being installed or uninstalled.
3194
3195 25. The pcre.h file was built from pcre.h.in, but the only changes that were
3196 made were to insert the current release number. This seemed silly, because
3197 it made things harder for people building PCRE on systems that don't run
3198 "configure". I have turned pcre.h into a distributed file, no longer built
3199 by "configure", with the version identification directly included. There is
3200 no longer a pcre.h.in file.
3201
3202 However, this change necessitated a change to the pcre-config script as
3203 well. It is built from pcre-config.in, and one of the substitutions was the
3204 release number. I have updated configure.ac so that ./configure now finds
3205 the release number by grepping pcre.h.
3206
3207 26. Added the ability to run the tests under valgrind.
3208
3209
3210 Version 6.4 05-Sep-05
3211 ---------------------
3212
3213 1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
3214 "--" to be printed when multiple files were scanned, even when none of the
3215 -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
3216 consider it to be a bug, and have restored the previous behaviour.
3217
3218 2. A couple of code tidies to get rid of compiler warnings.
3219
3220 3. The pcretest program used to cheat by referring to symbols in the library
3221 whose names begin with _pcre_. These are internal symbols that are not
3222 really supposed to be visible externally, and in some environments it is
3223 possible to suppress them. The cheating is now confined to including
3224 certain files from the library's source, which is a bit cleaner.
3225
3226 4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
3227 file's purpose clearer.
3228
3229 5. Reorganized pcre_ucp_findchar().
3230
3231
3232 Version 6.3 15-Aug-05
3233 ---------------------
3234
3235 1. The file libpcre.pc.in did not have general read permission in the tarball.
3236
3237 2. There were some problems when building without C++ support:
3238
3239 (a) If C++ support was not built, "make install" and "make test" still
3240 tried to test it.
3241
3242 (b) There were problems when the value of CXX was explicitly set. Some
3243 changes have been made to try to fix these, and ...
3244
3245 (c) --disable-cpp can now be used to explicitly disable C++ support.
3246
3247 (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
3248 backslash in a target when C++ was disabled. This confuses some
3249 versions of "make", apparently. Using an intermediate variable solves
3250 this. (Same for CPP_LOBJ.)
3251
3252 3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
3253 (non-Windows) now includes $(CFLAGS) because these flags are sometimes
3254 necessary on certain architectures.
3255
3256 4. Added a setting of -export-symbols-regex to the link command to remove
3257 those symbols that are exported in the C sense, but actually are local
3258 within the library, and not documented. Their names all begin with
3259 "_pcre_". This is not a perfect job, because (a) we have to except some
3260 symbols that pcretest ("illegally") uses, and (b) the facility isn't always
3261 available (and never for static libraries). I have made a note to try to
3262 find a way round (a) in the future.
3263
3264
3265 Version 6.2 01-Aug-05
3266 ---------------------
3267
3268 1. There was no test for integer overflow of quantifier values. A construction
3269 such as {1111111111111111} would give undefined results. What is worse, if
3270 a minimum quantifier for a parenthesized subpattern overflowed and became
3271 negative, the calculation of the memory size went wrong. This could have
3272 led to memory overwriting.
3273
3274 2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
3275
3276 3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
3277 operating environments where this matters.
3278
3279 4. Applied Giuseppe Maxia's patch to add additional features for controlling
3280 PCRE options from within the C++ wrapper.
3281
3282 5. Named capturing subpatterns were not being correctly counted when a pattern
3283 was compiled. This caused two problems: (a) If there were more than 100
3284 such subpatterns, the calculation of the memory needed for the whole
3285 compiled pattern went wrong, leading to an overflow error. (b) Numerical
3286 back references of the form \12, where the number was greater than 9, were
3287 not recognized as back references, even though there were sufficient
3288 previous subpatterns.
3289
3290 6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
3291 versions of gcc, e.g. 2.95.4.
3292
3293
3294 Version 6.1 21-Jun-05
3295 ---------------------
3296
3297 1. There was one reference to the variable "posix" in pcretest.c that was not
3298 surrounded by "#if !defined NOPOSIX".
3299
3300 2. Make it possible to compile pcretest without DFA support, UTF8 support, or
3301 the cross-check on the old pcre_info() function, for the benefit of the
3302 cut-down version of PCRE that is currently imported into Exim.
3303
3304 3. A (silly) pattern starting with (?i)(?-i) caused an internal space
3305 allocation error. I've done the easy fix, which wastes 2 bytes for sensible
3306 patterns that start (?i) but I don't think that matters. The use of (?i) is
3307 just an example; this all applies to the other options as well.
3308
3309 4. Since libtool seems to echo the compile commands it is issuing, the output
3310 from "make" can be reduced a bit by putting "@" in front of each libtool
3311 compile command.
3312
3313 5. Patch from the folks at Google for configure.in to be a bit more thorough
3314 in checking for a suitable C++ installation before trying to compile the
3315 C++ stuff. This should fix a reported problem when a compiler was present,
3316 but no suitable headers.
3317
3318 6. The man pages all had just "PCRE" as their title. I have changed them to
3319 be the relevant file name. I have also arranged that these names are
3320 retained in the file doc/pcre.txt, which is a concatenation in text format
3321 of all the man pages except the little individual ones for each function.
3322
3323 7. The NON-UNIX-USE file had not been updated for the different set of source
3324 files that come with release 6. I also added a few comments about the C++
3325 wrapper.
3326
3327
3328 Version 6.0 07-Jun-05
3329 ---------------------
3330
3331 1. Some minor internal re-organization to help with my DFA experiments.
3332
3333 2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
3334 didn't matter for the library itself when fully configured, but did matter
3335 when compiling without UCP support, or within Exim, where the ucp files are
3336 not imported.
3337
3338 3. Refactoring of the library code to split up the various functions into
3339 different source modules. The addition of the new DFA matching code (see
3340 below) to a single monolithic source would have made it really too
3341 unwieldy, quite apart from causing all the code to be include in a
3342 statically linked application, when only some functions are used. This is
3343 relevant even without the DFA addition now that patterns can be compiled in
3344 one application and matched in another.
3345
3346 The downside of splitting up is that there have to be some external
3347 functions and data tables that are used internally in different modules of
3348 the library but which are not part of the API. These have all had their
3349 names changed to start with "_pcre_" so that they are unlikely to clash
3350 with other external names.
3351
3352 4. Added an alternate matching function, pcre_dfa_exec(), which matches using
3353 a different (DFA) algorithm. Although it is slower than the original
3354 function, it does have some advantages for certain types of matching
3355 problem.
3356
3357 5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
3358 including restarting after a partial match.
3359
3360 6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
3361 defined when compiling for Windows was sent to me. I have put it into the
3362 code, though I have no means of testing or verifying it.
3363
3364 7. Added the pcre_refcount() auxiliary function.
3365
3366 8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
3367 match before or at the first newline in the subject string. In pcretest,
3368 the /f option on a pattern can be used to set this.
3369
3370 9. A repeated \w when used in UTF-8 mode with characters greater than 256
3371 would behave wrongly. This has been present in PCRE since release 4.0.
3372
3373 10. A number of changes to the pcregrep command:
3374
3375 (a) Refactored how -x works; insert ^(...)$ instead of setting
3376 PCRE_ANCHORED and checking the length, in preparation for adding
3377 something similar for -w.
3378
3379 (b) Added the -w (match as a word) option.
3380
3381 (c) Refactored the way lines are read and buffered so as to have more
3382 than one at a time available.
3383
3384 (d) Implemented a pcregrep test script.
3385
3386 (e) Added the -M (multiline match) option. This allows patterns to match
3387 over several lines of the subject. The buffering ensures that at least
3388 8K, or the rest of the document (whichever is the shorter) is available
3389 for matching (and similarly the previous 8K for lookbehind assertions).
3390
3391 (f) Changed the --help output so that it now says
3392
3393 -w, --word-regex(p)
3394
3395 instead of two lines, one with "regex" and the other with "regexp"
3396 because that confused at least one person since the short forms are the
3397 same. (This required a bit of code, as the output is generated
3398 automatically from a table. It wasn't just a text change.)
3399
3400 (g) -- can be used to terminate pcregrep options if the next thing isn't an
3401 option but starts with a hyphen. Could be a pattern or a path name
3402 starting with a hyphen, for instance.
3403
3404 (h) "-" can be given as a file name to represent stdin.
3405
3406 (i) When file names are being printed, "(standard input)" is used for
3407 the standard input, for compatibility with GNU grep. Previously
3408 "<stdin>" was used.
3409
3410 (j) The option --label=xxx can be used to supply a name to be used for
3411 stdin when file names are being printed. There is no short form.
3412
3413 (k) Re-factored the options decoding logic because we are going to add
3414 two more options that take data. Such options can now be given in four
3415 different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
3416
3417 (l) Added the -A, -B, and -C options for requesting that lines of context
3418 around matches be printed.
3419
3420 (m) Added the -L option to print the names of files that do not contain
3421 any matching lines, that is, the complement of -l.
3422
3423 (n) The return code is 2 if any file cannot be opened, but pcregrep does
3424 continue to scan other files.
3425
3426 (o) The -s option was incorrectly implemented. For compatibility with other
3427 greps, it now suppresses the error message for a non-existent or non-
3428 accessible file (but not the return code). There is a new option called
3429 -q that suppresses the output of matching lines, which was what -s was
3430 previously doing.
3431
3432 (p) Added --include and --exclude options to specify files for inclusion
3433 and exclusion when recursing.
3434
3435 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
3436 Hopefully, it now does.
3437
3438 12. Missing cast in pcre_study().
3439
3440 13. Added an "uninstall" target to the makefile.
3441
3442 14. Replaced "extern" in the function prototypes in Makefile.in with
3443 "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
3444 world, but is set differently for Windows.
3445
3446 15. Added a second compiling function called pcre_compile2(). The only
3447 difference is that it has an extra argument, which is a pointer to an
3448 integer error code. When there is a compile-time failure, this is set
3449 non-zero, in addition to the error test pointer being set to point to an
3450 error message. The new argument may be NULL if no error number is required
3451 (but then you may as well call pcre_compile(), which is now just a
3452 wrapper). This facility is provided because some applications need a
3453 numeric error indication, but it has also enabled me to tidy up the way
3454 compile-time errors are handled in the POSIX wrapper.
3455
3456 16. Added VPATH=.libs to the makefile; this should help when building with one
3457 prefix path and installing with another. (Or so I'm told by someone who
3458 knows more about this stuff than I do.)
3459
3460 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
3461 passes PCRE_DOTALL to the pcre_compile() function, making the "." character
3462 match everything, including newlines. This is not POSIX-compatible, but
3463 somebody wanted the feature. From pcretest it can be activated by using
3464 both the P and the s flags.
3465
3466 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
3467
3468 19. libpcre.pc was being incorrectly installed as executable.
3469
3470 20. A couple of places in pcretest check for end-of-line by looking for '\n';
3471 it now also looks for '\r' so that it will work unmodified on Windows.
3472
3473 21. Added Google's contributed C++ wrapper to the distribution.
3474
3475 22. Added some untidy missing memory free() calls in pcretest, to keep
3476 Electric Fence happy when testing.
3477
3478
3479
3480 Version 5.0 13-Sep-04
3481 ---------------------
3482
3483 1. Internal change: literal characters are no longer packed up into items
3484 containing multiple characters in a single byte-string. Each character
3485 is now matched using a separate opcode. However, there may be more than one
3486 byte in the character in UTF-8 mode.
3487
3488 2. The pcre_callout_block structure has two new fields: pattern_position and
3489 next_item_length. These contain the offset in the pattern to the next match
3490 item, and its length, respectively.
3491
3492 3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
3493 insertion of callouts before each pattern item. Added the /C option to
3494 pcretest to make use of this.
3495
3496 4. On the advice of a Windows user, the lines
3497
3498 #if defined(_WIN32) || defined(WIN32)
3499 _setmode( _fileno( stdout ), 0x8000 );
3500 #endif /* defined(_WIN32) || defined(WIN32) */
3501
3502 have been added to the source of pcretest. This apparently does useful
3503 magic in relation to line terminators.
3504
3505 5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
3506 for the benefit of those environments where the "b" makes a difference.
3507
3508 6. The icc compiler has the same options as gcc, but "configure" doesn't seem
3509 to know about it. I have put a hack into configure.in that adds in code
3510 to set GCC=yes if CC=icc. This seems to end up at a point in the
3511 generated configure script that is early enough to affect the setting of
3512 compiler options, which is what is needed, but I have no means of testing
3513 whether it really works. (The user who reported this had patched the
3514 generated configure script, which of course I cannot do.)
3515
3516 LATER: After change 22 below (new libtool files), the configure script
3517 seems to know about icc (and also ecc). Therefore, I have commented out
3518 this hack in configure.in.
3519
3520 7. Added support for pkg-config (2 patches were sent in).
3521
3522 8. Negated POSIX character classes that used a combination of internal tables
3523 were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
3524 [[:^ascii]]. Typically, they would match almost any characters. The other
3525 POSIX classes were not broken in this way.
3526
3527 9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
3528 to find the match, as PCRE was deluded into thinking that the match had to
3529 start at the start point or following a newline. The same bug applied to
3530 patterns with negative forward assertions or any backward assertions
3531 preceding ".*" at the start, unless the pattern required a fixed first
3532 character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
3533
3534 10. In UTF-8 mode, when moving forwards in the subject after a failed match
3535 starting at the last subject character, bytes beyond the end of the subject
3536 string were read.
3537
3538 11. Renamed the variable "class" as "classbits" to make life easier for C++
3539 users. (Previously there was a macro definition, but it apparently wasn't
3540 enough.)
3541
3542 12. Added the new field "tables" to the extra data so that tables can be passed
3543 in at exec time, or the internal tables can be re-selected. This allows
3544 a compiled regex to be saved and re-used at a later time by a different
3545 program that might have everything at different addresses.
3546
3547 13. Modified the pcre-config script so that, when run on Solaris, it shows a
3548 -R library as well as a -L library.
3549
3550 14. The debugging options of pcretest (-d on the command line or D on a
3551 pattern) showed incorrect output for anything following an extended class
3552 that contained multibyte characters and which was followed by a quantifier.
3553
3554 15. Added optional support for general category Unicode character properties
3555 via the \p, \P, and \X escapes. Unicode property support implies UTF-8
3556 support. It adds about 90K to the size of the library. The meanings of the
3557 inbuilt class escapes such as \d and \s have NOT been changed.
3558
3559 16. Updated pcredemo.c to include calls to free() to release the memory for the
3560 compiled pattern.
3561
3562 17. The generated file chartables.c was being created in the source directory
3563 instead of in the building directory. This caused the build to fail if the
3564 source directory was different from the building directory, and was
3565 read-only.
3566
3567 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
3568 file. No doubt somebody will tell me if they don't make sense... Also added
3569 Dan Mooney's comments about building on OpenVMS.
3570
3571 19. Added support for partial matching via the PCRE_PARTIAL option for
3572 pcre_exec() and the \P data escape in pcretest.
3573
3574 20. Extended pcretest with 3 new pattern features:
3575
3576 (i) A pattern option of the form ">rest-of-line" causes pcretest to
3577 write the compiled pattern to the file whose name is "rest-of-line".
3578 This is a straight binary dump of the data, with the saved pointer to
3579 the character tables forced to be NULL. The study data, if any, is
3580 written too. After writing, pcretest reads a new pattern.
3581
3582 (ii) If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
3583 compiled pattern from the given file. There must not be any
3584 occurrences of "<" in the file name (pretty unlikely); if there are,
3585 pcretest will instead treat the initial "<" as a pattern delimiter.
3586 After reading in the pattern, pcretest goes on to read data lines as
3587 usual.
3588
3589 (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
3590 and 16-bit fields in a compiled pattern, to simulate a pattern that
3591 was compiled on a host of opposite endianness.
3592
3593 21. The pcre-exec() function can now cope with patterns that were compiled on
3594 hosts of opposite endianness, with this restriction:
3595
3596 As for any compiled expression that is saved and used later, the tables
3597 pointer field cannot be preserved; the extra_data field in the arguments
3598 to pcre_exec() should be used to pass in a tables address if a value
3599 other than the default internal tables were used at compile time.
3600
3601 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
3602 now diagnosed as an error. Previously, most of the time, a negative number
3603 would have been treated as zero, but if in addition "ovector" was passed as
3604 NULL, a crash could occur.
3605
3606 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
3607 new versions from the libtool 1.5 distribution (the last one is a copy of
3608 a file called libtool.m4). This seems to have fixed the need to patch
3609 "configure" to support Darwin 1.3 (which I used to do). However, I still
3610 had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
3611 workstation).
3612
3613 24. Changed the PCRE licence to be the more standard "BSD" licence.
3614
3615
3616 Version 4.5 01-Dec-03
3617 ---------------------
3618
3619 1. There has been some re-arrangement of the code for the match() function so
3620 that it can be compiled in a version that does not call itself recursively.
3621 Instead, it keeps those local variables that need separate instances for
3622 each "recursion" in a frame on the heap, and gets/frees frames whenever it
3623 needs to "recurse". Keeping track of where control must go is done by means
3624 of setjmp/longjmp. The whole thing is implemented by a set of macros that
3625 hide most of the details from the main code, and operates only if
3626 NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
3627 "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
3628 operating.
3629
3630 To make it easier for callers to provide specially tailored get/free
3631 functions for this usage, two new functions, pcre_stack_malloc, and
3632 pcre_stack_free, are used. They are always called in strict stacking order,
3633 and the size of block requested is always the same.
3634
3635 The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
3636 PCRE has been compiled to use the stack or the heap for recursion. The
3637 -C option of pcretest uses this to show which version is compiled.
3638
3639 A new data escape \S, is added to pcretest; it causes the amounts of store
3640 obtained and freed by both kinds of malloc/free at match time to be added
3641 to the output.
3642
3643 2. Changed the locale test to use "fr_FR" instead of "fr" because that's
3644 what's available on my current Linux desktop machine.
3645
3646 3. When matching a UTF-8 string, the test for a valid string at the start has
3647 been extended. If start_offset is not zero, PCRE now checks that it points
3648 to a byte that is the start of a UTF-8 character. If not, it returns
3649 PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
3650 this is necessary because there may be backward assertions in the pattern.
3651 When matching the same subject several times, it may save resources to use
3652 PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
3653
3654 4. The code for checking the validity of UTF-8 strings has been tightened so
3655 that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
3656 containing "overlong sequences".
3657
3658 5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
3659 I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
3660 should have been "&", but it just so happened that all the cases this let
3661 through by mistake were picked up later in the function.
3662
3663 6. I had used a variable called "isblank" - this is a C99 function, causing
3664 some compilers to warn. To avoid this, I renamed it (as "blankclass").
3665
3666 7. Cosmetic: (a) only output another newline at the end of pcretest if it is
3667 prompting; (b) run "./pcretest /dev/null" at the start of the test script
3668 so the version is shown; (c) stop "make test" echoing "./RunTest".
3669
3670 8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
3671
3672 9. The prototype for memmove() for systems that don't have it was using
3673 size_t, but the inclusion of the header that defines size_t was later. I've
3674 moved the #includes for the C headers earlier to avoid this.
3675
3676 10. Added some adjustments to the code to make it easier to compiler on certain
3677 special systems:
3678
3679 (a) Some "const" qualifiers were missing.
3680 (b) Added the macro EXPORT before all exported functions; by default this
3681 is defined to be empty.
3682 (c) Changed the dftables auxiliary program (that builds chartables.c) so
3683 that it reads its output file name as an argument instead of writing
3684 to the standard output and assuming this can be redirected.
3685
3686 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
3687 class containing characters with values greater than 255, PCRE compilation
3688 went into a loop.
3689
3690 12. A recursive reference to a subpattern that was within another subpattern
3691 that had a minimum quantifier of zero caused PCRE to crash. For example,
3692 (x(y(?2))z)? provoked this bug with a subject that got as far as the
3693 recursion. If the recursively-called subpattern itself had a zero repeat,
3694 that was OK.
3695
3696 13. In pcretest, the buffer for reading a data line was set at 30K, but the
3697 buffer into which it was copied (for escape processing) was still set at
3698 1024, so long lines caused crashes.
3699
3700 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
3701 "internal error: code overflow...". This applied to any character class
3702 that was followed by a possessive quantifier.
3703
3704 15. Modified the Makefile to add libpcre.la as a prerequisite for
3705 libpcreposix.la because I was told this is needed for a parallel build to
3706 work.
3707
3708 16. If a pattern that contained .* following optional items at the start was
3709 studied, the wrong optimizing data was generated, leading to matching
3710 errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
3711 matching string must start with a or b or c. The correct conclusion for
3712 this pattern is that a match can start with any character.
3713
3714
3715 Version 4.4 13-Aug-03
3716 ---------------------
3717
3718 1. In UTF-8 mode, a character class containing characters with values between
3719 127 and 255 was not handled correctly if the compiled pattern was studied.
3720 In fixing this, I have also improved the studying algorithm for such
3721 classes (slightly).
3722
3723 2. Three internal functions had redundant arguments passed to them. Removal
3724 might give a very teeny performance improvement.
3725
3726 3. Documentation bug: the value of the capture_top field in a callout is *one
3727 more than* the number of the hightest numbered captured substring.
3728
3729 4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
3730 in incorrectly linking with a previously installed version. They now link
3731 explicitly with libpcre.la.
3732
3733 5. configure.in no longer needs to recognize Cygwin specially.
3734
3735 6. A problem in pcre.in for Windows platforms is fixed.
3736
3737 7. If a pattern was successfully studied, and the -d (or /D) flag was given to
3738 pcretest, it used to include the size of the study block as part of its
3739 output. Unfortunately, the structure contains a field that has a different
3740 size on different hardware architectures. This meant that the tests that
3741 showed this size failed. As the block is currently always of a fixed size,
3742 this information isn't actually particularly useful in pcretest output, so
3743 I have just removed it.
3744
3745 8. Three pre-processor statements accidentally did not start in column 1.
3746 Sadly, there are *still* compilers around that complain, even though
3747 standard C has not required this for well over a decade. Sigh.
3748
3749 9. In pcretest, the code for checking callouts passed small integers in the
3750 callout_data field, which is a void * field. However, some picky compilers
3751 complained about the casts involved for this on 64-bit systems. Now
3752 pcretest passes the address of the small integer instead, which should get
3753 rid of the warnings.
3754
3755 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
3756 both compile and run time, and gives an error if an invalid UTF-8 sequence
3757 is found. There is a option for disabling this check in cases where the
3758 string is known to be correct and/or the maximum performance is wanted.
3759
3760 11. In response to a bug report, I changed one line in Makefile.in from
3761
3762 -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
3763 to
3764 -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
3765
3766 to look similar to other lines, but I have no way of telling whether this
3767 is the right thing to do, as I do not use Windows. No doubt I'll get told
3768 if it's wrong...
3769
3770
3771 Version 4.3 21-May-03
3772 ---------------------
3773
3774 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
3775 Makefile.
3776
3777 2. Some refactoring to improve the quality of the code:
3778
3779 (i) The utf8_table... variables are now declared "const".
3780
3781 (ii) The code for \cx, which used the "case flipping" table to upper case
3782 lower case letters, now just substracts 32. This is ASCII-specific,
3783 but the whole concept of \cx is ASCII-specific, so it seems
3784 reasonable.
3785
3786 (iii) PCRE was using its character types table to recognize decimal and
3787 hexadecimal digits in the pattern. This is silly, because it handles
3788 only 0-9, a-f, and A-F, but the character types table is locale-
3789 specific, which means strange things might happen. A private
3790 table is now used for this - though it costs 256 bytes, a table is
3791 much faster than multiple explicit tests. Of course, the standard
3792 character types table is still used for matching digits in subject
3793 strings against \d.
3794
3795 (iv) Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
3796 ending in _t are). So I've renamed it as ESC_tee.
3797
3798 3. The first argument for regexec() in the POSIX wrapper should have been
3799 defined as "const".
3800
3801 4. Changed pcretest to use malloc() for its buffers so that they can be
3802 Electric Fenced for debugging.
3803
3804 5. There were several places in the code where, in UTF-8 mode, PCRE would try
3805 to read one or more bytes before the start of the subject string. Often this
3806 had no effect on PCRE's behaviour, but in some circumstances it could
3807 provoke a segmentation fault.
3808
3809 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
3810 to try to read one or more bytes before the start of the subject string.
3811
3812 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
3813 UTF-8 support could misbehave in various ways if the subject string
3814 contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
3815 area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
3816 back over UTF-8 characters.)
3817
3818
3819 Version 4.2 14-Apr-03
3820 ---------------------
3821
3822 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
3823
3824 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
3825 [ON_WINDOWS]: new variable, "#" on non-Windows platforms
3826 [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
3827 [WIN_PREFIX]: new variable, "cyg" for Cygwin
3828 * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
3829 and BUILD_EXEEXT
3830 Note: automatic setting of the BUILD variables is not yet working
3831 set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
3832 compile-time but not at link-time
3833 [LINK]: use for linking executables only
3834 make different versions for Windows and non-Windows
3835 [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
3836 libraries
3837 [LINK_FOR_BUILD]: new variable
3838 [OBJEXT]: use throughout
3839 [EXEEXT]: use throughout
3840 <winshared>: new target
3841 <wininstall>: new target
3842 <dftables.o>: use native compiler
3843 <dftables>: use native linker
3844 <install>: handle Windows platform correctly
3845 <clean>: ditto
3846 <check>: ditto
3847 copy DLL to top builddir before testing
3848
3849 As part of these changes, -no-undefined was removed again. This was reported
3850 to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
3851 in any case.
3852
3853 3. Some tidies to get rid of compiler warnings:
3854
3855 . In the match_data structure, match_limit was an unsigned long int, whereas
3856 match_call_count was an int. I've made them both unsigned long ints.
3857
3858 . In pcretest the fact that a const uschar * doesn't automatically cast to
3859 a void * provoked a warning.
3860
3861 . Turning on some more compiler warnings threw up some "shadow" variables
3862 and a few more missing casts.
3863
3864 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
3865 option, a class that contained a single character with a value between 128
3866 and 255 (e.g. /[\xFF]/) caused PCRE to crash.
3867
3868 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
3869 option, a class that contained several characters, but with at least one
3870 whose value was between 128 and 255 caused PCRE to crash.
3871
3872
3873 Version 4.1 12-Mar-03
3874 ---------------------
3875
3876 1. Compiling with gcc -pedantic found a couple of places where casts were
3877 needed, and a string in dftables.c that was longer than standard compilers are
3878 required to support.
3879
3880 2. Compiling with Sun's compiler found a few more places where the code could
3881 be tidied up in order to avoid warnings.
3882
3883 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
3884 first of these names is deprecated in the latest Autoconf in favour of the name
3885 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
3886 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
3887 analogy I have changed it to CFLAGS_FOR_BUILD.
3888
3889 4. Added -no-undefined to the linking command in the Makefile, because this is
3890 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
3891 linking step for the pcreposix library.
3892
3893 5. PCRE was failing to diagnose the case of two named groups with the same
3894 name.
3895
3896 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
3897 literal character that is needed in the subject for a match, and scans along to
3898 ensure that it is present before embarking on the full matching process. This
3899 saves time in cases of nested unlimited repeats that are never going to match.
3900 Problem: the scan can take a lot of time if the subject is very long (e.g.
3901 megabytes), thus penalizing straightforward matches. It is now done only if the
3902 amount of subject to be scanned is less than 1000 bytes.
3903
3904 7. A lesser problem with the same optimization is that it was recording the
3905 first character of an anchored pattern as "needed", thus provoking a search
3906 right along the subject, even when the first match of the pattern was going to
3907 fail. The "needed" character is now not set for anchored patterns, unless it
3908 follows something in the pattern that is of non-fixed length. Thus, it still
3909 fulfils its original purpose of finding quick non-matches in cases of nested
3910 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
3911
3912
3913 Version 4.0 17-Feb-03
3914 ---------------------
3915
3916 1. If a comment in an extended regex that started immediately after a meta-item
3917 extended to the end of string, PCRE compiled incorrect data. This could lead to
3918 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
3919
3920 2. Moved to autoconf 2.53 and libtool 1.4.2.
3921
3922 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
3923 the special perltest8 script is no longer needed - all the tests can be run
3924 from a single perltest script.
3925
3926 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
3927 by \s. It has now been removed in PCRE. This means it isn't recognized as
3928 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
3929 class [:space:] *does* include VT, thereby creating a mess.
3930
3931 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
3932 space and tab.
3933
3934 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
3935 its new features into the main test script, reducing the number of scripts.
3936
3937 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
3938 were backward compatible, and made the (?i) apply to the whole pattern, as if
3939 /i were given. Now it behaves more logically, and applies the option setting
3940 only to what follows. PCRE has been changed to follow suit. However, if it
3941 finds options settings right at the start of the pattern, it extracts them into
3942 the global options, as before. Thus, they show up in the info data.
3943
3944 8. Added support for the \Q...\E escape sequence. Characters in between are
3945 treated as literals. This is slightly different from Perl in that $ and @ are
3946 also handled as literals inside the quotes. In Perl, they will cause variable
3947 interpolation. Note the following examples:
3948
3949 Pattern PCRE matches Perl matches
3950
3951 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
3952 \Qabc\$xyz\E abc\$xyz abc\$xyz
3953 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
3954
3955 For compatibility with Perl, \Q...\E sequences are recognized inside character
3956 classes as well as outside them.
3957
3958 9. Re-organized 3 code statements in pcretest to avoid "overflow in
3959 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
3960 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
3961 signed/unsigned warnings.
3962
3963 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
3964 option for pcretest, so I've replaced it by a simple function that does just
3965 that job.
3966
3967 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
3968 "pcregrep -".
3969
3970 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
3971 Java package. This provides some syntactic sugar for simple cases of what my
3972 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
3973 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
3974 item, you can use this simplified notation. Note that only makes sense with
3975 greedy quantifiers. Consequently, the use of the possessive quantifier forces
3976 greediness, whatever the setting of the PCRE_UNGREEDY option.
3977
3978 13. A change of greediness default within a pattern was not taking effect at
3979 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
3980 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
3981 was abstracted outside.
3982
3983 14. PCRE now supports the \G assertion. It is true when the current matching
3984 position is at the start point of the match. This differs from \A when the
3985 starting offset is non-zero. Used with the /g option of pcretest (or similar
3986 code), it works in the same way as it does for Perl's /g option. If all
3987 alternatives of a regex begin with \G, the expression is anchored to the start
3988 match position, and the "anchored" flag is set in the compiled expression.
3989
3990 15. Some bugs concerning the handling of certain option changes within patterns
3991 have been fixed. These applied to options other than (?ims). For example,
3992 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
3993 been the other way round. Some of this was related to change 7 above.
3994
3995 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
3996 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
3997 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
3998 POSIX classes only within a class (e.g. /[[:alpha:]]/).
3999
4000 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
4001 mode. Unlike ".", it always matches newline, whatever the setting of
4002 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
4003 assertions. Perl allows it, but it doesn't (in general) work because it can't
4004 calculate the length of the lookbehind. At least, that's the case for Perl
4005 5.8.0 - I've been told they are going to document that it doesn't work in
4006 future.
4007
4008 18. Added an error diagnosis for escapes that PCRE does not support: these are
4009 \L, \l, \N, \P, \p, \U, \u, and \X.
4010
4011 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
4012 reading past the end of the pattern in cases such as /[abcd/.
4013
4014 20. PCRE was getting more memory than necessary for patterns with classes that
4015 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
4016
4017 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
4018 compiling PCRE for use with Virtual Pascal.
4019
4020 22. Small fix to the Makefile to make it work properly if the build is done
4021 outside the source tree.
4022
4023 23. Added a new extension: a condition to go with recursion. If a conditional
4024 subpattern starts with (?(R) the "true" branch is used if recursion has
4025 happened, whereas the "false" branch is used only at the top level.
4026
4027 24. When there was a very long string of literal characters (over 255 bytes
4028 without UTF support, over 250 bytes with UTF support), the computation of how
4029 much memory was required could be incorrect, leading to segfaults or other
4030 strange effects.
4031
4032 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
4033 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
4034 there was a subsequent back reference to those brackets. This meant that, for
4035 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
4036 possible to check for precisely this case. All we can do is abandon the
4037 optimization if .* occurs inside capturing brackets when there are any back
4038 references whatsoever. (See below for a better fix that came later.)
4039
4040 26. The handling of the optimization for finding the first character of a
4041 non-anchored pattern, and for finding a character that is required later in the
4042 match were failing in some cases. This didn't break the matching; it just
4043 failed to optimize when it could. The way this is done has been re-implemented.
4044
4045 27. Fixed typo in error message for invalid (?R item (it said "(?p").
4046
4047 28. Added a new feature that provides some of the functionality that Perl
4048 provides with (?{...}). The facility is termed a "callout". The way it is done
4049 in PCRE is for the caller to provide an optional function, by setting
4050 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
4051 global variable. By default it is unset, which disables all calling out. To get
4052 the function called, the regex must include (?C) at appropriate points. This
4053 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
4054 This provides a means of identifying different callout points. When PCRE
4055 reaches such a point in the regex, if pcre_callout has been set, the external
4056 function is called. It is provided with data in a structure called
4057 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
4058 matching continues; if it returns a non-zero value, the match at the current
4059 point fails. However, backtracking will occur if possible. [This was changed
4060 later and other features added - see item 49 below.]
4061
4062 29. pcretest is upgraded to test the callout functionality. It provides a
4063 callout function that displays information. By default, it shows the start of
4064 the match and the current position in the text. There are some new data escapes
4065 to vary what happens:
4066
4067 \C+ in addition, show current contents of captured substrings
4068 \C- do not supply a callout function
4069 \C!n return 1 when callout number n is reached
4070 \C!n!m return 1 when callout number n is reached for the mth time
4071
4072 30. If pcregrep was called with the -l option and just a single file name, it
4073 output "<stdin>" if a match was found, instead of the file name.
4074
4075 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
4076 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
4077 pcre_exec(). This saves a malloc/free per call. The default value of
4078 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
4079 when configuring.
4080
4081 32. The default maximum size of a compiled pattern is 64K. There have been a
4082 few cases of people hitting this limit. The code now uses macros to handle the
4083 storing of links as offsets within the compiled pattern. It defaults to 2-byte
4084 links, but this can be changed to 3 or 4 bytes by --with-link-size when
4085 configuring. Tests 2 and 5 work only with 2-byte links because they output
4086 debugging information about compiled patterns.
4087
4088 33. Internal code re-arrangements:
4089
4090 (a) Moved the debugging function for printing out a compiled regex into
4091 its own source file (printint.c) and used #include to pull it into
4092 pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
4093 separate copies.
4094
4095 (b) Defined the list of op-code names for debugging as a macro in
4096 internal.h so that it is next to the definition of the opcodes.
4097
4098 (c) Defined a table of op-code lengths for simpler skipping along compiled
4099 code. This is again a macro in internal.h so that it is next to the
4100 definition of the opcodes.
4101
4102 34. Added support for recursive calls to individual subpatterns, along the
4103 lines of Robin Houston's patch (but implemented somewhat differently).
4104
4105 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
4106 allow it to read and process whole directories in Win32. This code was
4107 contributed by Lionel Fourquaux; it has not been tested by me.
4108
4109 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
4110 used to name a group. Names consist of alphanumerics and underscores, and must
4111 be unique. Back references use the syntax (?P=name) and recursive calls use
4112 (?P>name) which is a PCRE extension to the Python extension. Groups still have
4113 numbers. The function pcre_fullinfo() can be used after compilation to extract
4114 a name/number map. There are three relevant calls:
4115
4116 PCRE_INFO_NAMEENTRYSIZE yields the size of each entry in the map
4117 PCRE_INFO_NAMECOUNT yields the number of entries
4118 PCRE_INFO_NAMETABLE yields a pointer to the map.
4119
4120 The map is a vector of fixed-size entries. The size of each entry depends on
4121 the length of the longest name used. The first two bytes of each entry are the
4122 group number, most significant byte first. There follows the corresponding
4123 name, zero terminated. The names are in alphabetical order.
4124
4125 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
4126 case instead of 255. Making it the same both with and without UTF-8 support
4127 means that the same test output works with both.
4128
4129 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
4130 calling malloc() with a zero argument.
4131
4132 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
4133 optimization. I've improved things by keeping a bitmap of backreferences with
4134 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
4135 fact referenced, the optimization can be applied. It is unlikely that a
4136 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
4137 the match to follow \n) will appear inside brackets with a number greater than
4138 31, but if it does, any back reference > 31 suppresses the optimization.
4139
4140 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
4141 of disabling numbered capturing parentheses. Any opening parenthesis that is
4142 not followed by ? behaves as if it were followed by ?: but named parentheses
4143 can still be used for capturing (and they will acquire numbers in the usual
4144 way).
4145
4146 41. Redesigned the return codes from the match() function into yes/no/error so
4147 that errors can be passed back from deep inside the nested calls. A malloc
4148 failure while inside a recursive subpattern call now causes the
4149 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
4150
4151 42. It is now possible to set a limit on the number of times the match()
4152 function is called in a call to pcre_exec(). This facility makes it possible to
4153 limit the amount of recursion and backtracking, though not in a directly
4154 obvious way, because the match() function is used in a number of different
4155 circumstances. The count starts from zero for each position in the subject
4156 string (for non-anchored patterns). The default limit is, for compatibility, a
4157 large number, namely 10 000 000. You can change this in two ways:
4158
4159 (a) When configuring PCRE before making, you can use --with-match-limit=n
4160 to set a default value for the compiled library.
4161
4162 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
4163 a different value is set. See 45 below.
4164
4165 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
4166
4167 43. Added a new function pcre_config(int, void *) to enable run-time extraction
4168 of things that can be changed at compile time. The first argument specifies
4169 what is wanted and the second points to where the information is to be placed.
4170 The current list of available information is:
4171
4172 PCRE_CONFIG_UTF8
4173
4174 The output is an integer that is set to one if UTF-8 support is available;
4175 otherwise it is set to zero.
4176
4177 PCRE_CONFIG_NEWLINE
4178
4179 The output is an integer that it set to the value of the code that is used for
4180 newline. It is either LF (10) or CR (13).
4181
4182 PCRE_CONFIG_LINK_SIZE
4183
4184 The output is an integer that contains the number of bytes used for internal
4185 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
4186
4187 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
4188
4189 The output is an integer that contains the threshold above which the POSIX
4190 interface uses malloc() for output vectors. See item 31 above.
4191
4192 PCRE_CONFIG_MATCH_LIMIT
4193
4194 The output is an unsigned integer that contains the default limit of the number
4195 of match() calls in a pcre_exec() execution. See 42 above.
4196
4197 44. pcretest has been upgraded by the addition of the -C option. This causes it
4198 to extract all the available output from the new pcre_config() function, and to
4199 output it. The program then exits immediately.
4200
4201 45. A need has arisen to pass over additional data with calls to pcre_exec() in
4202 order to support additional features. One way would have been to define
4203 pcre_exec2() (for example) with extra arguments, but this would not have been
4204 extensible, and would also have required all calls to the original function to
4205 be mapped to the new one. Instead, I have chosen to extend the mechanism that
4206 is used for passing in "extra" data from pcre_study().
4207
4208 The pcre_extra structure is now exposed and defined in pcre.h. It currently
4209 contains the following fields:
4210
4211 flags a bitmap indicating which of the following fields are set
4212 study_data opaque data from pcre_study()
4213 match_limit a way of specifying a limit on match() calls for a specific
4214 call to pcre_exec()
4215 callout_data data for callouts (see 49 below)
4216
4217 The flag bits are also defined in pcre.h, and are
4218
4219 PCRE_EXTRA_STUDY_DATA
4220 PCRE_EXTRA_MATCH_LIMIT
4221 PCRE_EXTRA_CALLOUT_DATA
4222
4223 The pcre_study() function now returns one of these new pcre_extra blocks, with
4224 the actual study data pointed to by the study_data field, and the
4225 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
4226 before. That is, this change is entirely upwards-compatible and requires no
4227 change to existing code.
4228
4229 If you want to pass in additional data to pcre_exec(), you can either place it
4230 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
4231 block.
4232
4233 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
4234 data string contains the escape sequence \M, pcretest calls pcre_exec() several
4235 times with different match limits, until it finds the minimum value needed for
4236 pcre_exec() to complete. The value is then output. This can be instructive; for
4237 most simple matches the number is quite small, but for pathological cases it
4238 gets very large very quickly.
4239
4240 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
4241 returns the size of the data block pointed to by the study_data field in a
4242 pcre_extra block, that is, the value that was passed as the argument to
4243 pcre_malloc() when PCRE was getting memory in which to place the information
4244 created by pcre_study(). The fourth argument should point to a size_t variable.
4245 pcretest has been extended so that this information is shown after a successful
4246 pcre_study() call when information about the compiled regex is being displayed.
4247
4248 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
4249 because what follows is always an absolute path. (Later: it turns out that this
4250 is more than cosmetic for MinGW, because it doesn't like empty path
4251 components.)
4252
4253 49. Some changes have been made to the callout feature (see 28 above):
4254
4255 (i) A callout function now has three choices for what it returns:
4256
4257 0 => success, carry on matching
4258 > 0 => failure at this point, but backtrack if possible
4259 < 0 => serious error, return this value from pcre_exec()
4260
4261 Negative values should normally be chosen from the set of PCRE_ERROR_xxx
4262 values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
4263 "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
4264 use by callout functions. It will never be used by PCRE itself.
4265
4266 (ii) The pcre_extra structure (see 45 above) has a void * field called
4267 callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
4268 pcre_callout_block structure has a field of the same name. The contents of
4269 the field passed in the pcre_extra structure are passed to the callout
4270 function in the corresponding field in the callout block. This makes it
4271 easier to use the same callout-containing regex from multiple threads. For
4272 testing, the pcretest program has a new data escape
4273
4274 \C*n pass the number n (may be negative) as callout_data
4275
4276 If the callout function in pcretest receives a non-zero value as
4277 callout_data, it returns that value.
4278
4279 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
4280 there were some redundant $(CFLAGS) in commands that are now specified as
4281 $(LINK), which already includes $(CFLAGS).
4282
4283 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
4284 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
4285 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
4286 one-byte characters throughout. Note that case-insensitive matching applies
4287 only to characters whose values are less than 256. PCRE doesn't support the
4288 notion of cases for higher-valued characters.
4289
4290 (i) A character class whose characters are all within 0-255 is handled as
4291 a bit map, and the map is inverted for negative classes. Previously, a
4292 character > 255 always failed to match such a class; however it should
4293 match if the class was a negative one (e.g. [^ab]). This has been fixed.
4294
4295 (ii) A negated character class with a single character < 255 is coded as
4296 "not this character" (OP_NOT). This wasn't working properly when the test
4297 character was multibyte, either singly or repeated.
4298
4299 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
4300 mode, for example: \x{100}{2,3}.
4301
4302 (iv) The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
4303 singly or repeated) now correctly test multibyte characters. However,
4304 PCRE doesn't recognize any characters with values greater than 255 as
4305 digits, spaces, or word characters. Such characters always match \D, \S,
4306 and \W, and never match \d, \s, or \w.
4307
4308 (v) Classes may now contain characters and character ranges with values
4309 greater than 255. For example: [ab\x{100}-\x{400}].
4310
4311 (vi) pcregrep now has a --utf-8 option (synonym -u) which makes it call
4312 PCRE in UTF-8 mode.
4313
4314 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
4315 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
4316 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
4317 value.)
4318
4319 53. The single man page has become too large. I have therefore split it up into
4320 a number of separate man pages. These also give rise to individual HTML pages;
4321 these are now put in a separate directory, and there is an index.html page that
4322 lists them all. Some hyperlinking between the pages has been installed.
4323
4324 54. Added convenience functions for handling named capturing parentheses.
4325
4326 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
4327 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
4328 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
4329 are faulted.
4330
4331 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
4332 calling configure. These values are used when compiling the dftables.c program
4333 which is run to generate the source of the default character tables. They
4334 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
4335 you will need to set these values.
4336
4337 57. Updated the building process for Windows DLL, as provided by Fred Cox.
4338
4339
4340 Version 3.9 02-Jan-02
4341 ---------------------
4342
4343 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
4344
4345 2. If --disable-static was given, the building process failed when trying to
4346 build pcretest and pcregrep. (For some reason it was using libtool to compile
4347 them, which is not right, as they aren't part of the library.)
4348
4349
4350 Version 3.8 18-Dec-01
4351 ---------------------
4352
4353 1. The experimental UTF-8 code was completely screwed up. It was packing the
4354 bytes in the wrong order. How dumb can you get?
4355
4356
4357 Version 3.7 29-Oct-01
4358 ---------------------
4359
4360 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
4361 This caused pcretest, when used on the test data, to segfault. Unfortunately,
4362 this didn't happen under Solaris 8, where I normally test things.
4363
4364 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
4365 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
4366 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
4367 this fix an hour or so after the initial 3.7 release.)
4368
4369
4370 Version 3.6 23-Oct-01
4371 ---------------------
4372
4373 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
4374 offsets passed as NULL with zero offset count.
4375
4376 2. The config.guess and config.sub files had not been updated when I moved to
4377 the latest autoconf.
4378
4379
4380 Version 3.5 15-Aug-01
4381 ---------------------
4382
4383 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
4384 had been forgotten.
4385
4386 2. By using declared but undefined structures, we can avoid using "void"
4387 definitions in pcre.h while keeping the internal definitions of the structures
4388 private.
4389
4390 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
4391 user point of view, this means that both static and shared libraries are built
4392 by default, but this can be individually controlled. More of the work of
4393 handling this static/shared cases is now inside libtool instead of PCRE's make
4394 file.
4395
4396 4. The pcretest utility is now installed along with pcregrep because it is
4397 useful for users (to test regexs) and by doing this, it automatically gets
4398 relinked by libtool. The documentation has been turned into a man page, so
4399 there are now .1, .txt, and .html versions in /doc.
4400
4401 5. Upgrades to pcregrep:
4402 (i) Added long-form option names like gnu grep.
4403 (ii) Added --help to list all options with an explanatory phrase.
4404 (iii) Added -r, --recursive to recurse into sub-directories.
4405 (iv) Added -f, --file to read patterns from a file.
4406
4407 6. pcre_exec() was referring to its "code" argument before testing that
4408 argument for NULL (and giving an error if it was NULL).
4409
4410 7. Upgraded Makefile.in to allow for compiling in a different directory from
4411 the source directory.
4412
4413 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
4414 options bits, the pointer it was passed was to an int instead of to an unsigned
4415 long int. This mattered only on 64-bit systems.
4416
4417 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
4418 generated) instead of pcre.in, which it its source. Also made the same change
4419 in several of the .c files.
4420
4421 10. A new release of gcc defines printf() as a macro, which broke pcretest
4422 because it had an ifdef in the middle of a string argument for printf(). Fixed
4423 by using separate calls to printf().
4424
4425 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
4426 script, to force use of CR or LF instead of \n in the source. On non-Unix
4427 systems, the value can be set in config.h.
4428
4429 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
4430 absolute limit. Changed the text of the error message to make this clear, and
4431 likewise updated the man page.
4432
4433 13. The limit of 99 on the number of capturing subpatterns has been removed.
4434 The new limit is 65535, which I hope will not be a "real" limit.
4435
4436
4437 Version 3.4 22-Aug-00
4438 ---------------------
4439
4440 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
4441
4442 2. Diagnose condition (?(0) as an error instead of crashing on matching.
4443
4444
4445 Version 3.3 01-Aug-00
4446 ---------------------
4447
4448 1. If an octal character was given, but the value was greater than \377, it
4449 was not getting masked to the least significant bits, as documented. This could
4450 lead to crashes in some systems.
4451
4452 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
4453 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
4454
4455 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
4456 These just pass their arguments on to (pcre_free)(), but they are provided
4457 because some uses of PCRE bind it to non-C systems that can call its functions,
4458 but cannot call free() or pcre_free() directly.
4459
4460 4. Add "make test" as a synonym for "make check". Corrected some comments in
4461 the Makefile.
4462
4463 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
4464 Makefile.
4465
4466 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
4467 command called pgrep for grepping around the active processes.
4468
4469 7. Added the beginnings of support for UTF-8 character strings.
4470
4471 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
4472 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
4473 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
4474 out for the ar command.)
4475
4476
4477 Version 3.2 12-May-00
4478 ---------------------
4479
4480 This is purely a bug fixing release.
4481
4482 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
4483 of ZA. This was just one example of several cases that could provoke this bug,
4484 which was introduced by change 9 of version 2.00. The code for breaking
4485 infinite loops after an iteration that matches an empty string was't working
4486 correctly.
4487
4488 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
4489 when matched against abbab (for example). After matching an empty string, it
4490 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
4491 caused it to match further down the string than it should.
4492
4493 3. The code contained an inclusion of sys/types.h. It isn't clear why this
4494 was there because it doesn't seem to be needed, and it causes trouble on some
4495 systems, as it is not a Standard C header. It has been removed.
4496
4497 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
4498 were reported on the Macintosh. The changes were from
4499
4500 while ((c = *(++ptr)) != 0 && c != '\n');
4501 to
4502 while ((c = *(++ptr)) != 0 && c != '\n') ;
4503
4504 Totally extraordinary, but if that's what it takes...
4505
4506 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
4507 available. Added HAVE_BCOPY and an autoconf test for it; if neither
4508 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
4509 assumes the way PCRE uses memmove() (always moving upwards).
4510
4511 6. PCRE is being used in one environment where strchr() is not available. There
4512 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
4513 faster code anyway.
4514
4515
4516 Version 3.1 09-Feb-00
4517 ---------------------
4518
4519 The only change in this release is the fixing of some bugs in Makefile.in for
4520 the "install" target:
4521
4522 (1) It was failing to install pcreposix.h.
4523
4524 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
4525
4526
4527 Version 3.0 01-Feb-00
4528 ---------------------
4529
4530 1. Add support for the /+ modifier to perltest (to output $` like it does in
4531 pcretest).
4532
4533 2. Add support for the /g modifier to perltest.
4534
4535 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
4536 matches null strings.
4537
4538 4. Fix perltest so that it doesn't do unwanted things when fed an empty
4539 pattern. Perl treats empty patterns specially - it reuses the most recent
4540 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
4541 effect.
4542
4543 5. The POSIX interface was broken in that it was just handing over the POSIX
4544 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
4545 required a bigger vector, with some working space on the end. This means that
4546 the POSIX wrapper now has to get and free some memory, and copy the results.
4547
4548 6. Added some simple autoconf support, placing the test data and the
4549 documentation in separate directories, re-organizing some of the
4550 information files, and making it build pcre-config (a GNU standard). Also added
4551 libtool support for building PCRE as a shared library, which is now the
4552 default.
4553
4554 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
4555 09 are not valid octal constants. Single digits will be used for minor values
4556 less than 10.
4557
4558 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
4559 existing programs that set these in the POSIX interface can use PCRE without
4560 modification.
4561
4562 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
4563 return all that pcre_info() returns, plus additional data. The pcre_info()
4564 function is retained for compatibility, but is considered to be obsolete.
4565
4566 10. Added experimental recursion feature (?R) to handle one common case that
4567 Perl 5.6 will be able to do with (?p{...}).
4568
4569 11. Added support for POSIX character classes like [:alpha:], which Perl is
4570 adopting.
4571
4572
4573 Version 2.08 31-Aug-99
4574 ----------------------
4575
4576 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
4577 trying to match at the startoffset position, but instead was moving forward to
4578 the next newline as if a previous match had failed.
4579
4580 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
4581 and could get into a loop if a null string was matched other than at the start
4582 of the subject.
4583
4584 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
4585 be distinguished at compile time, and for completeness also added PCRE_DATE.
4586
4587 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
4588 in GnuWin32 environments.
4589
4590
4591 Version 2.07 29-Jul-99
4592 ----------------------
4593
4594 1. The documentation is now supplied in plain text form and HTML as well as in
4595 the form of man page sources.
4596
4597 2. C++ compilers don't like assigning (void *) values to other pointer types.
4598 In particular this affects malloc(). Although there is no problem in Standard
4599 C, I've put in casts to keep C++ compilers happy.
4600
4601 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
4602 should be (const char *).
4603
4604 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
4605 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
4606 However, I haven't made this a standard facility. The documentation doesn't
4607 mention it, and the Makefile doesn't support it.
4608
4609 5. The Makefile now contains an "install" target, with editable destinations at
4610 the top of the file. The pcretest program is not installed.
4611
4612 6. pgrep -V now gives the PCRE version number and date.
4613
4614 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
4615 causing the entire string to be ignored, instead of just the last character.
4616
4617 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
4618 non-matching string, it can take a very, very long time, even for strings of
4619 quite modest length, because of the nested recursion. PCRE now does better in
4620 some of these cases. It does this by remembering the last required literal
4621 character in the pattern, and pre-searching the subject to ensure it is present
4622 before running the real match. In other words, it applies a heuristic to detect
4623 some types of certain failure quickly, and in the above example, if presented
4624 with a string that has no trailing " it gives "no match" very quickly.
4625
4626 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
4627 other alternatives are tried instead.
4628
4629
4630 Version 2.06 09-Jun-99
4631 ----------------------
4632
4633 1. Change pcretest's output for amount of store used to show just the code
4634 space, because the remainder (the data block) varies in size between 32-bit and
4635 64-bit systems.
4636
4637 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
4638 start matching at. This allows lookbehinds to work when searching for multiple
4639 occurrences in a string.
4640
4641 3. Added additional options to pcretest for testing multiple occurrences:
4642
4643 /+ outputs the rest of the string that follows a match
4644 /g loops for multiple occurrences, using the new startoffset argument
4645 /G loops for multiple occurrences by passing an incremented pointer
4646
4647 4. PCRE wasn't doing the "first character" optimization for patterns starting
4648 with \b or \B, though it was doing it for other lookbehind assertions. That is,
4649 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
4650 the letter 'x'. On long subject strings, this gives a significant speed-up.
4651
4652
4653 Version 2.05 21-Apr-99
4654 ----------------------
4655
4656 1. Changed the type of magic_number from int to long int so that it works
4657 properly on 16-bit systems.
4658
4659 2. Fixed a bug which caused patterns starting with .* not to work correctly
4660 when the subject string contained newline characters. PCRE was assuming
4661 anchoring for such patterns in all cases, which is not correct because .* will
4662 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
4663 DOTALL is set at top level; otherwise it knows that patterns starting with .*
4664 must be retried after every newline in the subject.
4665
4666
4667 Version 2.04 18-Feb-99
4668 ----------------------
4669
4670 1. For parenthesized subpatterns with repeats whose minimum was zero, the
4671 computation of the store needed to hold the pattern was incorrect (too large).
4672 If such patterns were nested a few deep, this could multiply and become a real
4673 problem.
4674
4675 2. Added /M option to pcretest to show the memory requirement of a specific
4676 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
4677
4678 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
4679 compiled in such a way that the backtracking after subsequent failure was
4680 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
4681 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
4682
4683
4684 Version 2.03 02-Feb-99
4685 ----------------------
4686
4687 1. Fixed typo and small mistake in man page.
4688
4689 2. Added 4th condition (GPL supersedes if conflict) and created separate
4690 LICENCE file containing the conditions.
4691
4692 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
4693 Perl, that is the internal \ allows the delimiter to be included in the
4694 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
4695 the final delimiter, add \ to the end of the pattern (to test the error).
4696
4697 4. Added the convenience functions for extracting substrings after a successful
4698 match. Updated pcretest to make it able to test these functions.
4699
4700
4701 Version 2.02 14-Jan-99
4702 ----------------------
4703
4704 1. Initialized the working variables associated with each extraction so that
4705 their saving and restoring doesn't refer to uninitialized store.
4706
4707 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
4708 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
4709 fix the problem.
4710
4711 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
4712 calls, and wasn't printing the correct value for compiling calls. Increased the
4713 default value of LOOPREPEAT, and the number of significant figures in the
4714 times.
4715
4716 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
4717
4718 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
4719 a building problem on Windows NT with a FAT file system.
4720
4721
4722 Version 2.01 21-Oct-98
4723 ----------------------
4724
4725 1. Changed the API for pcre_compile() to allow for the provision of a pointer
4726 to character tables built by pcre_maketables() in the current locale. If NULL
4727 is passed, the default tables are used.
4728
4729
4730 Version 2.00 24-Sep-98
4731 ----------------------
4732
4733 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
4734 it any more.
4735
4736 2. Allow quantification of (?>) groups, and make it work correctly.
4737
4738 3. The first character computation wasn't working for (?>) groups.
4739
4740 4. Correct the implementation of \Z (it is permitted to match on the \n at the
4741 end of the subject) and add 5.005's \z, which really does match only at the
4742 very end of the subject.
4743
4744 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
4745
4746 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
4747 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
4748 localized options. All options to pcre_study() were also removed.
4749
4750 7. Add other new features from 5.005:
4751
4752 $(?<= positive lookbehind
4753 $(?<! negative lookbehind
4754 (?imsx-imsx) added the unsetting capability
4755 such a setting is global if at outer level; local otherwise
4756 (?imsx-imsx:) non-capturing groups with option setting
4757 (?(cond)re|re) conditional pattern matching
4758
4759 A backreference to itself in a repeated group matches the previous
4760 captured string.
4761
4762 8. General tidying up of studying (both automatic and via "study")
4763 consequential on the addition of new assertions.
4764
4765 9. As in 5.005, unlimited repeated groups that could match an empty substring
4766 are no longer faulted at compile time. Instead, the loop is forcibly broken at
4767 runtime if any iteration does actually match an empty substring.
4768
4769 10. Include the RunTest script in the distribution.
4770
4771 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
4772 discrepancies, some of which were old and were also with respect to 5.004. They
4773 have now been fixed.
4774
4775
4776 Version 1.09 28-Apr-98
4777 ----------------------
4778
4779 1. A negated single character class followed by a quantifier with a minimum
4780 value of one (e.g. [^x]{1,6} ) was not compiled correctly. This could lead to
4781 program crashes, or just wrong answers. This did not apply to negated classes
4782 containing more than one character, or to minima other than one.
4783
4784
4785 Version 1.08 27-Mar-98
4786 ----------------------
4787
4788 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
4789
4790 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
4791 latter must appear before anything that relies on it in the pattern.
4792
4793
4794 Version 1.07 16-Feb-98
4795 ----------------------
4796
4797 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
4798 repeat of a potentially empty string).
4799
4800
4801 Version 1.06 23-Jan-98
4802 ----------------------
4803
4804 1. Added Markus Oberhumer's little patches for C++.
4805
4806 2. Literal strings longer than 255 characters were broken.
4807
4808
4809 Version 1.05 23-Dec-97
4810 ----------------------
4811
4812 1. Negated character classes containing more than one character were failing if
4813 PCRE_CASELESS was set at run time.
4814
4815
4816 Version 1.04 19-Dec-97
4817 ----------------------
4818
4819 1. Corrected the man page, where some "const" qualifiers had been omitted.
4820
4821 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
4822 input syntax.
4823
4824 3. Fixed memory leak which occurred when a regex with back references was
4825 matched with an offsets vector that wasn't big enough. The temporary memory
4826 that is used in this case wasn't being freed if the match failed.
4827
4828 4. Tidied pcretest to ensure it frees memory that it gets.
4829
4830 5. Temporary memory was being obtained in the case where the passed offsets
4831 vector was exactly big enough.
4832
4833 6. Corrected definition of offsetof() from change 5 below.
4834
4835 7. I had screwed up change 6 below and broken the rules for the use of
4836 setjmp(). Now fixed.
4837
4838
4839 Version 1.03 18-Dec-97
4840 ----------------------
4841
4842 1. A erroneous regex with a missing opening parenthesis was correctly
4843 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
4844 on some systems.
4845
4846 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
4847 it was reported that one broken compiler failed on the former because "code" is
4848 also an independent variable.
4849
4850 3. The erroneous regex a[]b caused an array overrun reference.
4851
4852 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
4853 fail on data ending with that character. (It was going on too far, and checking
4854 the next character, typically a binary zero.) This was specific to the
4855 optimized code for single-character negative classes.
4856
4857 5. Added a contributed patch from the TIN world which does the following:
4858
4859 + Add an undef for memmove, in case the the system defines a macro for it.
4860
4861 + Add a definition of offsetof(), in case there isn't one. (I don't know
4862 the reason behind this - offsetof() is part of the ANSI standard - but
4863 it does no harm).
4864
4865 + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
4866 most of the places where whitespace preceded '#'. I have given up and
4867 allowed the remaining 2 cases to be at the margin.
4868
4869 + Rename some variables in pcre to eliminate shadowing. This seems very
4870 pedantic, but does no harm, of course.
4871
4872 6. Moved the call to setjmp() into its own function, to get rid of warnings
4873 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
4874
4875 7. Constructs such as \d{8,} were compiling into the equivalent of
4876 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
4877 outcome, but in this particular case used more store than had been allocated,
4878 which caused the bug to be discovered because it threw up an internal error.
4879
4880 8. The debugging code in both pcre and pcretest for outputting the compiled
4881 form of a regex was going wrong in the case of back references followed by
4882 curly-bracketed repeats.
4883
4884
4885 Version 1.02 12-Dec-97
4886 ----------------------
4887
4888 1. Typos in pcre.3 and comments in the source fixed.
4889
4890 2. Applied a contributed patch to get rid of places where it used to remove
4891 'const' from variables, and fixed some signed/unsigned and uninitialized
4892 variable warnings.
4893
4894 3. Added the "runtest" target to Makefile.
4895
4896 4. Set default compiler flag to -O2 rather than just -O.
4897
4898
4899 Version 1.01 19-Nov-97
4900 ----------------------
4901
4902 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
4903 like /([ab]*)*/, that is, for classes with more than one character in them.
4904
4905 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
4906 as /((?>a*))*/ (a PCRE_EXTRA facility).
4907
4908
4909 Version 1.00 18-Nov-97
4910 ----------------------
4911
4912 1. Added compile-time macros to support systems such as SunOS4 which don't have
4913 memmove() or strerror() but have other things that can be used instead.
4914
4915 2. Arranged that "make clean" removes the executables.
4916
4917
4918 Version 0.99 27-Oct-97
4919 ----------------------
4920
4921 1. Fixed bug in code for optimizing classes with only one character. It was
4922 initializing a 32-byte map regardless, which could cause it to run off the end
4923 of the memory it had got.
4924
4925 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
4926
4927
4928 Version 0.98 22-Oct-97
4929 ----------------------
4930
4931 1. Fixed bug in code for handling temporary memory usage when there are more
4932 back references than supplied space in the ovector. This could cause segfaults.
4933
4934
4935 Version 0.97 21-Oct-97
4936 ----------------------
4937
4938 1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
4939
4940 2. Optimized negated single characters not to use a bit map.
4941
4942 3. Brought error texts together as macro definitions; clarified some of them;
4943 fixed one that was wrong - it said "range out of order" when it meant "invalid
4944 escape sequence".
4945
4946 4. Changed some char * arguments to const char *.
4947
4948 5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
4949
4950 6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
4951 pcretest.
4952
4953
4954 Version 0.96 16-Oct-97
4955 ----------------------
4956
4957 1. Added a simple "pgrep" utility to the distribution.
4958
4959 2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
4960 unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
4961 where "ddd" means "one or more decimal digits".
4962
4963 3. Fixed serious bug. If a pattern had a back reference, but the call to
4964 pcre_exec() didn't supply a large enough ovector to record the related
4965 identifying subpattern, the match always failed. PCRE now remembers the number
4966 of the largest back reference, and gets some temporary memory in which to save
4967 the offsets during matching if necessary, in order to ensure that
4968 backreferences always work.
4969
4970 4. Increased the compatibility with Perl in a number of ways:
4971
4972 (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
4973 to request this handling. The option can be set at compile or exec time.
4974
4975 (b) $ matches before a terminating newline by default; an option
4976 PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
4977 mode). The option can be set at compile or exec time.
4978
4979 (c) The handling of \ followed by a digit other than 0 is now supposed to be
4980 the same as Perl's. If the decimal number it represents is less than 10
4981 or there aren't that many previous left capturing parentheses, an octal
4982 escape is read. Inside a character class, it's always an octal escape,
4983 even if it is a single digit.
4984
4985 (d) An escaped but undefined alphabetic character is taken as a literal,
4986 unless PCRE_EXTRA is set. Currently this just reserves the remaining
4987 escapes.
4988
4989 (e) {0} is now permitted. (The previous item is removed from the compiled
4990 pattern).
4991
4992 5. Changed all the names of code files so that the basic parts are no longer
4993 than 10 characters, and abolished the teeny "globals.c" file.
4994
4995 6. Changed the handling of character classes; they are now done with a 32-byte
4996 bit map always.
4997
4998 7. Added the -d and /D options to pcretest to make it possible to look at the
4999 internals of compilation without having to recompile pcre.
5000
5001
5002 Version 0.95 23-Sep-97
5003 ----------------------
5004
5005 1. Fixed bug in pre-pass concerning escaped "normal" characters such as \x5c or
5006 \x20 at the start of a run of normal characters. These were being treated as
5007 real characters, instead of the source characters being re-checked.
5008
5009
5010 Version 0.94 18-Sep-97
5011 ----------------------
5012
5013 1. The functions are now thread-safe, with the caveat that the global variables
5014 containing pointers to malloc() and free() or alternative functions are the
5015 same for all threads.
5016
5017 2. Get pcre_study() to generate a bitmap of initial characters for non-
5018 anchored patterns when this is possible, and use it if passed to pcre_exec().
5019
5020
5021 Version 0.93 15-Sep-97
5022 ----------------------
5023
5024 1. /(b)|(:+)/ was computing an incorrect first character.
5025
5026 2. Add pcre_study() to the API and the passing of pcre_extra to pcre_exec(),
5027 but not actually doing anything yet.
5028
5029 3. Treat "-" characters in classes that cannot be part of ranges as literals,
5030 as Perl does (e.g. [-az] or [az-]).
5031
5032 4. Set the anchored flag if a branch starts with .* or .*? because that tests
5033 all possible positions.
5034
5035 5. Split up into different modules to avoid including unneeded functions in a
5036 compiled binary. However, compile and exec are still in one module. The "study"
5037 function is split off.
5038
5039 6. The character tables are now in a separate module whose source is generated
5040 by an auxiliary program - but can then be edited by hand if required. There are
5041 now no calls to isalnum(), isspace(), isdigit(), isxdigit(), tolower() or
5042 toupper() in the code.
5043
5044 7. Turn the malloc/free funtions variables into pcre_malloc and pcre_free and
5045 make them global. Abolish the function for setting them, as the caller can now
5046 set them directly.
5047
5048
5049 Version 0.92 11-Sep-97
5050 ----------------------
5051
5052 1. A repeat with a fixed maximum and a minimum of 1 for an ordinary character
5053 (e.g. /a{1,3}/) was broken (I mis-optimized it).
5054
5055 2. Caseless matching was not working in character classes if the characters in
5056 the pattern were in upper case.
5057
5058 3. Make ranges like [W-c] work in the same way as Perl for caseless matching.
5059
5060 4. Make PCRE_ANCHORED public and accept as a compile option.
5061
5062 5. Add an options word to pcre_exec() and accept PCRE_ANCHORED and
5063 PCRE_CASELESS at run time. Add escapes \A and \I to pcretest to cause it to
5064 pass them.
5065
5066 6. Give an error if bad option bits passed at compile or run time.
5067
5068 7. Add PCRE_MULTILINE at compile and exec time, and (?m) as well. Add \M to
5069 pcretest to cause it to pass that flag.
5070
5071 8. Add pcre_info(), to get the number of identifying subpatterns, the stored
5072 options, and the first character, if set.
5073
5074 9. Recognize C+ or C{n,m} where n >= 1 as providing a fixed starting character.
5075
5076
5077 Version 0.91 10-Sep-97
5078 ----------------------
5079
5080 1. PCRE was failing to diagnose unlimited repeats of subpatterns that could
5081 match the empty string as in /(a*)*/. It was looping and ultimately crashing.
5082
5083 2. PCRE was looping on encountering an indefinitely repeated back reference to
5084 a subpattern that had matched an empty string, e.g. /(a|)\1*/. It now does what
5085 Perl does - treats the match as successful.
5086
5087 ****

Properties

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

  ViewVC Help
Powered by ViewVC 1.1.5