ViewVC logotype

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log

Revision 1040 - (show annotations)
Sun Sep 16 06:52:27 2012 UTC (7 years, 5 months ago) by zherczeg
File size: 236083 byte(s)
Improving the first n character searches.
1 ChangeLog for PCRE
2 ------------------
4 Version 8.32
5 ------------
7 1. Improved JIT compiler optimizations for first character search and single
8 character iterators.
10 2. Supporting IBM XL C compilers for PPC architectures in the JIT compiler.
11 Patch by Daniel Richard G.
13 3. Single character iterator optimizations in the JIT compiler.
15 4. Improved JIT compiler optimizations for character ranges.
17 5. Rename the "leave" variable names to "quit" to improve WinCE compatibility.
18 Reported by Giuseppe D'Angelo.
20 6. The PCRE_STARTLINE bit, indicating that a match can occur only at the start
21 of a line, was being set incorrectly in cases where .* appeared inside
22 atomic brackets at the start of a pattern, or where there was a subsequent
23 *PRUNE or *SKIP.
25 7. Improved instruction cache flush for POWER/PowerPC.
26 Patch by Daniel Richard G.
28 8. Fixed a number of issues in pcregrep, making it more compatible with GNU
29 grep:
31 (a) There is now no limit to the number of patterns to be matched.
33 (b) An error is given if a pattern is too long.
35 (c) Multiple uses of --exclude, --exclude-dir, --include, and --include-dir
36 are now supported.
38 (d) --exclude-from and --include-from (multiple use) have been added.
40 (e) Exclusions and inclusions now apply to all files and directories, not
41 just to those obtained from scanning a directory recursively.
43 (f) Multiple uses of -f and --file-list are now supported.
45 (g) In a Windows environment, the default for -d has been changed from
46 "read" (the GNU grep default) to "skip", because otherwise the presence
47 of a directory in the file list provokes an error.
49 (h) The documentation has been revised and clarified in places.
51 9. Improve the matching speed of capturing brackets.
53 10. Changed the meaning of \X so that it now matches a Unicode extended
54 grapheme cluster.
56 11. Patch by Daniel Richard G to the autoconf files to add a macro for sorting
57 out POSIX threads when JIT support is configured.
59 12. Added support for PCRE_STUDY_EXTRA_NEEDED.
61 13. In the POSIX wrapper regcomp() function, setting re_nsub field in the preg
62 structure could go wrong in environments where size_t is not the same size
63 as int.
65 14. Applied user-supplied patch to pcrecpp.cc to allow PCRE_NO_UTF8_CHECK to be
66 set.
68 15. The EBCDIC support had decayed; later updates to the code had included
69 explicit references to (e.g.) \x0a instead of CHAR_LF. There has been a
70 general tidy up of EBCDIC-related issues, and the documentation was also
71 not quite right. There is now a test that can be run on ASCII systems to
72 check some of the EBCDIC-related things (but is it not a full test).
74 16. The new PCRE_STUDY_EXTRA_NEEDED option is now used by pcregrep, resulting
75 in a small tidy to the code.
77 17. Fix JIT tests when UTF is disabled and both 8 and 16 bit mode are enabled.
79 18. If the --only-matching (-o) option in pcregrep is specified multiple
80 times, each one causes appropriate output. For example, -o1 -o2 outputs the
81 substrings matched by the 1st and 2nd capturing parentheses. A separating
82 string can be specified by --om-separator (default empty).
84 19. Improving the first n character searches.
87 Version 8.31 06-July-2012
88 -------------------------
90 1. Fixing a wrong JIT test case and some compiler warnings.
92 2. Removed a bashism from the RunTest script.
94 3. Add a cast to pcre_exec.c to fix the warning "unary minus operator applied
95 to unsigned type, result still unsigned" that was given by an MS compiler
96 on encountering the code "-sizeof(xxx)".
98 4. Partial matching support is added to the JIT compiler.
100 5. Fixed several bugs concerned with partial matching of items that consist
101 of more than one character:
103 (a) /^(..)\1/ did not partially match "aba" because checking references was
104 done on an "all or nothing" basis. This also applied to repeated
105 references.
107 (b) \R did not give a hard partial match if \r was found at the end of the
108 subject.
110 (c) \X did not give a hard partial match after matching one or more
111 characters at the end of the subject.
113 (d) When newline was set to CRLF, a pattern such as /a$/ did not recognize
114 a partial match for the string "\r".
116 (e) When newline was set to CRLF, the metacharacter "." did not recognize
117 a partial match for a CR character at the end of the subject string.
119 6. If JIT is requested using /S++ or -s++ (instead of just /S+ or -s+) when
120 running pcretest, the text "(JIT)" added to the output whenever JIT is
121 actually used to run the match.
123 7. Individual JIT compile options can be set in pcretest by following -s+[+]
124 or /S+[+] with a digit between 1 and 7.
126 8. OP_NOT now supports any UTF character not just single-byte ones.
128 9. (*MARK) control verb is now supported by the JIT compiler.
130 10. The command "./RunTest list" lists the available tests without actually
131 running any of them. (Because I keep forgetting what they all are.)
135 12. Applied a (slightly modified) user-supplied patch that improves performance
136 when the heap is used for recursion (compiled with --disable-stack-for-
137 recursion). Instead of malloc and free for each heap frame each time a
138 logical recursion happens, frames are retained on a chain and re-used where
139 possible. This sometimes gives as much as 30% improvement.
141 13. As documented, (*COMMIT) is now confined to within a recursive subpattern
142 call.
144 14. As documented, (*COMMIT) is now confined to within a positive assertion.
146 15. It is now possible to link pcretest with libedit as an alternative to
147 libreadline.
149 16. (*COMMIT) control verb is now supported by the JIT compiler.
151 17. The Unicode data tables have been updated to Unicode 6.1.0.
153 18. Added --file-list option to pcregrep.
155 19. Added binary file support to pcregrep, including the -a, --binary-files,
156 -I, and --text options.
158 20. The madvise function is renamed for posix_madvise for QNX compatibility
159 reasons. Fixed by Giuseppe D'Angelo.
161 21. Fixed a bug for backward assertions with REVERSE 0 in the JIT compiler.
163 22. Changed the option for creating symbolic links for 16-bit man pages from
164 -s to -sf so that re-installing does not cause issues.
166 23. Support PCRE_NO_START_OPTIMIZE in JIT as (*MARK) support requires it.
168 24. Fixed a very old bug in pcretest that caused errors with restarted DFA
169 matches in certain environments (the workspace was not being correctly
170 retained). Also added to pcre_dfa_exec() a simple plausibility check on
171 some of the workspace data at the beginning of a restart.
173 25. \s*\R was auto-possessifying the \s* when it should not, whereas \S*\R
174 was not doing so when it should - probably a typo introduced by SVN 528
175 (change 8.10/14).
177 26. When PCRE_UCP was not set, \w+\x{c4} was incorrectly auto-possessifying the
178 \w+ when the character tables indicated that \x{c4} was a word character.
179 There were several related cases, all because the tests for doing a table
180 lookup were testing for characters less than 127 instead of 255.
182 27. If a pattern contains capturing parentheses that are not used in a match,
183 their slots in the ovector are set to -1. For those that are higher than
184 any matched groups, this happens at the end of processing. In the case when
185 there were back references that the ovector was too small to contain
186 (causing temporary malloc'd memory to be used during matching), and the
187 highest capturing number was not used, memory off the end of the ovector
188 was incorrectly being set to -1. (It was using the size of the temporary
189 memory instead of the true size.)
191 28. To catch bugs like 27 using valgrind, when pcretest is asked to specify an
192 ovector size, it uses memory at the end of the block that it has got.
194 29. Check for an overlong MARK name and give an error at compile time. The
195 limit is 255 for the 8-bit library and 65535 for the 16-bit library.
197 30. JIT compiler update.
199 31. JIT is now supported on jailbroken iOS devices. Thanks for Ruiger
200 Rill for the patch.
202 32. Put spaces around SLJIT_PRINT_D in the JIT compiler. Required by CXX11.
204 33. Variable renamings in the PCRE-JIT compiler. No functionality change.
206 34. Fixed typos in pcregrep: in two places there was SUPPORT_LIBZ2 instead of
207 SUPPORT_LIBBZ2. This caused a build problem when bzip2 but not gzip (zlib)
208 was enabled.
210 35. Improve JIT code generation for greedy plus quantifier.
212 36. When /((?:a?)*)*c/ or /((?>a?)*)*c/ was matched against "aac", it set group
213 1 to "aa" instead of to an empty string. The bug affected repeated groups
214 that could potentially match an empty string.
216 37. Optimizing single character iterators in JIT.
218 38. Wide characters specified with \uxxxx in JavaScript mode are now subject to
219 the same checks as \x{...} characters in non-JavaScript mode. Specifically,
220 codepoints that are too big for the mode are faulted, and in a UTF mode,
221 disallowed codepoints are also faulted.
223 39. If PCRE was compiled with UTF support, in three places in the DFA
224 matcher there was code that should only have been obeyed in UTF mode, but
225 was being obeyed unconditionally. In 8-bit mode this could cause incorrect
226 processing when bytes with values greater than 127 were present. In 16-bit
227 mode the bug would be provoked by values in the range 0xfc00 to 0xdc00. In
228 both cases the values are those that cannot be the first data item in a UTF
229 character. The three items that might have provoked this were recursions,
230 possessively repeated groups, and atomic groups.
232 40. Ensure that libpcre is explicitly listed in the link commands for pcretest
233 and pcregrep, because some OS require shared objects to be explicitly
234 passed to ld, causing the link step to fail if they are not.
236 41. There were two incorrect #ifdefs in pcre_study.c, meaning that, in 16-bit
237 mode, patterns that started with \h* or \R* might be incorrectly matched.
240 Version 8.30 04-February-2012
241 -----------------------------
243 1. Renamed "isnumber" as "is_a_number" because in some Mac environments this
244 name is defined in ctype.h.
246 2. Fixed a bug in fixed-length calculation for lookbehinds that would show up
247 only in quite long subpatterns.
249 3. Removed the function pcre_info(), which has been obsolete and deprecated
250 since it was replaced by pcre_fullinfo() in February 2000.
252 4. For a non-anchored pattern, if (*SKIP) was given with a name that did not
253 match a (*MARK), and the match failed at the start of the subject, a
254 reference to memory before the start of the subject could occur. This bug
255 was introduced by fix 17 of release 8.21.
257 5. A reference to an unset group with zero minimum repetition was giving
258 totally wrong answers (in non-JavaScript-compatibility mode). For example,
259 /(another)?(\1?)test/ matched against "hello world test". This bug was
260 introduced in release 8.13.
262 6. Add support for 16-bit character strings (a large amount of work involving
263 many changes and refactorings).
265 7. RunGrepTest failed on msys because \r\n was replaced by whitespace when the
266 command "pattern=`printf 'xxx\r\njkl'`" was run. The pattern is now taken
267 from a file.
269 8. Ovector size of 2 is also supported by JIT based pcre_exec (the ovector size
270 rounding is not applied in this particular case).
272 9. The invalid Unicode surrogate codepoints U+D800 to U+DFFF are now rejected
273 if they appear, or are escaped, in patterns.
275 10. Get rid of a number of -Wunused-but-set-variable warnings.
277 11. The pattern /(?=(*:x))(q|)/ matches an empty string, and returns the mark
278 "x". The similar pattern /(?=(*:x))((*:y)q|)/ did not return a mark at all.
279 Oddly, Perl behaves the same way. PCRE has been fixed so that this pattern
280 also returns the mark "x". This bug applied to capturing parentheses,
281 non-capturing parentheses, and atomic parentheses. It also applied to some
282 assertions.
284 12. Stephen Kelly's patch to CMakeLists.txt allows it to parse the version
285 information out of configure.ac instead of relying on pcre.h.generic, which
286 is not stored in the repository.
288 13. Applied Dmitry V. Levin's patch for a more portable method for linking with
289 -lreadline.
291 14. ZH added PCRE_CONFIG_JITTARGET; added its output to pcretest -C.
293 15. Applied Graycode's patch to put the top-level frame on the stack rather
294 than the heap when not using the stack for recursion. This gives a
295 performance improvement in many cases when recursion is not deep.
297 16. Experimental code added to "pcretest -C" to output the stack frame size.
300 Version 8.21 12-Dec-2011
301 ------------------------
303 1. Updating the JIT compiler.
305 2. JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
306 are added as well.
308 3. Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
309 PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
310 calling _pcre_jit_exec. Some extra comments are added.
312 4. (*MARK) settings inside atomic groups that do not contain any capturing
313 parentheses, for example, (?>a(*:m)), were not being passed out. This bug
314 was introduced by change 18 for 8.20.
316 5. Supporting of \x, \U and \u in JavaScript compatibility mode based on the
317 ECMA-262 standard.
319 6. Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
320 erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
321 This bug was probably introduced by change 9 of 8.13.
323 7. While fixing 6 above, I noticed that a number of other items were being
324 incorrectly rejected as "not fixed length". This arose partly because newer
325 opcodes had not been added to the fixed-length checking code. I have (a)
326 corrected the bug and added tests for these items, and (b) arranged for an
327 error to occur if an unknown opcode is encountered while checking for fixed
328 length instead of just assuming "not fixed length". The items that were
329 rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
330 (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
331 repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
333 8. A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
334 being incorrectly compiled and would have given unpredicatble results.
336 9. A possessively repeated subpattern with minimum repeat count greater than
337 one behaved incorrectly. For example, (A){2,}+ behaved as if it was
338 (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
339 the first (A) could occur when it should not.
341 10. Add a cast and remove a redundant test from the code.
343 11. JIT should use pcre_malloc/pcre_free for allocation.
345 12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
346 best practice nowadays, and helps with cross-compiling. (If the exec_prefix
347 is anything other than /usr, -L is still shown).
349 13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
351 14. Perl does not support \N without a following name in a [] class; PCRE now
352 also gives an error.
354 15. If a forward reference was repeated with an upper limit of around 2000,
355 it caused the error "internal error: overran compiling workspace". The
356 maximum number of forward references (including repeats) was limited by the
357 internal workspace, and dependent on the LINK_SIZE. The code has been
358 rewritten so that the workspace expands (via pcre_malloc) if necessary, and
359 the default depends on LINK_SIZE. There is a new upper limit (for safety)
360 of around 200,000 forward references. While doing this, I also speeded up
361 the filling in of repeated forward references.
363 16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
364 incorrectly expecting the subject to contain another "a" after the start.
366 17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
367 in the match, the SKIP should be ignored. This was not happening; instead
368 the SKIP was being treated as NOMATCH. For patterns such as
369 /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
370 tested.
372 18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
373 now much more compatible with Perl, in particular in cases where the result
374 is a non-match for a non-anchored pattern. For example, if
375 /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
376 "m", where previously it did not return a name. A side effect of this
377 change is that for partial matches, the last encountered mark name is
378 returned, as for non matches. A number of tests that were previously not
379 Perl-compatible have been moved into the Perl-compatible test files. The
380 refactoring has had the pleasing side effect of removing one argument from
381 the match() function, thus reducing its stack requirements.
383 19. If the /S+ option was used in pcretest to study a pattern using JIT,
384 subsequent uses of /S (without +) incorrectly behaved like /S+.
386 21. Retrieve executable code size support for the JIT compiler and fixing
387 some warnings.
389 22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
390 not work when the shorter character appeared right at the end of the
391 subject string.
393 23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
394 systems.
396 24. Added PCRE_INFO_JITSIZE to pass on the value from (21) above, and also
397 output it when the /M option is used in pcretest.
399 25. The CheckMan script was not being included in the distribution. Also, added
400 an explicit "perl" to run Perl scripts from the PrepareRelease script
401 because this is reportedly needed in Windows.
403 26. If study data was being save in a file and studying had not found a set of
404 "starts with" bytes for the pattern, the data written to the file (though
405 never used) was taken from uninitialized memory and so caused valgrind to
406 complain.
408 27. Updated RunTest.bat as provided by Sheri Pierce.
410 28. Fixed a possible uninitialized memory bug in pcre_jit_compile.c.
412 29. Computation of memory usage for the table of capturing group names was
413 giving an unnecessarily large value.
416 Version 8.20 21-Oct-2011
417 ------------------------
419 1. Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
420 a POSIX class. After further experiments with Perl, which convinced me that
421 Perl has bugs and confusions, a closing square bracket is no longer allowed
422 in a POSIX name. This bug also affected patterns with classes that started
423 with full stops.
425 2. If a pattern such as /(a)b|ac/ is matched against "ac", there is no
426 captured substring, but while checking the failing first alternative,
427 substring 1 is temporarily captured. If the output vector supplied to
428 pcre_exec() was not big enough for this capture, the yield of the function
429 was still zero ("insufficient space for captured substrings"). This cannot
430 be totally fixed without adding another stack variable, which seems a lot
431 of expense for a edge case. However, I have improved the situation in cases
432 such as /(a)(b)x|abc/ matched against "abc", where the return code
433 indicates that fewer than the maximum number of slots in the ovector have
434 been set.
436 3. Related to (2) above: when there are more back references in a pattern than
437 slots in the output vector, pcre_exec() uses temporary memory during
438 matching, and copies in the captures as far as possible afterwards. It was
439 using the entire output vector, but this conflicts with the specification
440 that only 2/3 is used for passing back captured substrings. Now it uses
441 only the first 2/3, for compatibility. This is, of course, another edge
442 case.
444 4. Zoltan Herczeg's just-in-time compiler support has been integrated into the
445 main code base, and can be used by building with --enable-jit. When this is
446 done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
447 runtime --no-jit option is given.
449 5. When the number of matches in a pcre_dfa_exec() run exactly filled the
450 ovector, the return from the function was zero, implying that there were
451 other matches that did not fit. The correct "exactly full" value is now
452 returned.
454 6. If a subpattern that was called recursively or as a subroutine contained
455 (*PRUNE) or any other control that caused it to give a non-standard return,
456 invalid errors such as "Error -26 (nested recursion at the same subject
457 position)" or even infinite loops could occur.
459 7. If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
460 computing the minimum length on reaching *ACCEPT, and so ended up with the
461 wrong value of 1 rather than 0. Further investigation indicates that
462 computing a minimum subject length in the presence of *ACCEPT is difficult
463 (think back references, subroutine calls), and so I have changed the code
464 so that no minimum is registered for a pattern that contains *ACCEPT.
466 8. If (*THEN) was present in the first (true) branch of a conditional group,
467 it was not handled as intended. [But see 16 below.]
469 9. Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
470 Sheri Pierce.
472 10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
473 the first byte in a match must be "a".
475 11. Change 17 for 8.13 increased the recursion depth for patterns like
476 /a(?:.)*?a/ drastically. I've improved things by remembering whether a
477 pattern contains any instances of (*THEN). If it does not, the old
478 optimizations are restored. It would be nice to do this on a per-group
479 basis, but at the moment that is not feasible.
481 12. In some environments, the output of pcretest -C is CRLF terminated. This
482 broke RunTest's code that checks for the link size. A single white space
483 character after the value is now allowed for.
485 13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
486 For "fr", it uses the Windows-specific input and output files.
488 14. If (*THEN) appeared in a group that was called recursively or as a
489 subroutine, it did not work as intended. [But see next item.]
491 15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
492 pattern fragments (but not containing any | characters). If A and B are
493 matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
494 was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
495 D. In other words, Perl considers parentheses that do not contain any |
496 characters to be part of a surrounding alternative, whereas PCRE was
497 treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
498 differently. PCRE now behaves in the same way as Perl, except in the case
499 of subroutine/recursion calls such as (?1) which have in any case always
500 been different (but PCRE had them first :-).
502 16. Related to 15 above: Perl does not treat the | in a conditional group as
503 creating alternatives. Such a group is treated in the same way as an
504 ordinary group without any | characters when processing (*THEN). PCRE has
505 been changed to match Perl's behaviour.
507 17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
508 RunGrepTest script failed.
510 18. Change 22 for version 13 caused atomic groups to use more stack. This is
511 inevitable for groups that contain captures, but it can lead to a lot of
512 stack use in large patterns. The old behaviour has been restored for atomic
513 groups that do not contain any capturing parentheses.
515 19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
516 suppress the check for a minimum subject length at run time. (If it was
517 given to pcre_exec() or pcre_dfa_exec() it did work.)
519 20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
520 fail to work when decoding hex characters in data strings in EBCDIC
521 environments.
523 21. It appears that in at least one Mac OS environment, the isxdigit() function
524 is implemented as a macro that evaluates to its argument more than once,
525 contravening the C 90 Standard (I haven't checked a later standard). There
526 was an instance in pcretest which caused it to go wrong when processing
527 \x{...} escapes in subject strings. The has been rewritten to avoid using
528 things like p++ in the argument of isxdigit().
531 Version 8.13 16-Aug-2011
532 ------------------------
534 1. The Unicode data tables have been updated to Unicode 6.0.0.
536 2. Two minor typos in pcre_internal.h have been fixed.
538 3. Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
539 pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
540 in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
542 4. There were a number of related bugs in the code for matching backrefences
543 caselessly in UTF-8 mode when codes for the characters concerned were
544 different numbers of bytes. For example, U+023A and U+2C65 are an upper
545 and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
546 (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
547 code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
548 2-byte code at the end of the subject (it thought there wasn't enough data
549 left).
551 5. Comprehensive information about what went wrong is now returned by
552 pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
553 as the output vector has at least 2 elements. The offset of the start of
554 the failing character and a reason code are placed in the vector.
556 6. When the UTF-8 string check fails for pcre_compile(), the offset that is
557 now returned is for the first byte of the failing character, instead of the
558 last byte inspected. This is an incompatible change, but I hope it is small
559 enough not to be a problem. It makes the returned offset consistent with
560 pcre_exec() and pcre_dfa_exec().
562 7. pcretest now gives a text phrase as well as the error number when
563 pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
564 failure, the offset and reason code are output.
566 8. When \R was used with a maximizing quantifier it failed to skip backwards
567 over a \r\n pair if the subsequent match failed. Instead, it just skipped
568 back over a single character (\n). This seems wrong (because it treated the
569 two characters as a single entity when going forwards), conflicts with the
570 documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
571 behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
572 has been changed.
574 9. Some internal refactoring has changed the processing so that the handling
575 of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
576 time (the PCRE_DOTALL option was changed this way some time ago: version
577 7.7 change 16). This has made it possible to abolish the OP_OPT op code,
578 which was always a bit of a fudge. It also means that there is one less
579 argument for the match() function, which reduces its stack requirements
580 slightly. This change also fixes an incompatibility with Perl: the pattern
581 (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
583 10. More internal refactoring has drastically reduced the number of recursive
584 calls to match() for possessively repeated groups such as (abc)++ when
585 using pcre_exec().
587 11. While implementing 10, a number of bugs in the handling of groups were
588 discovered and fixed:
590 (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
591 (a|)*(?1) gave a compile-time internal error.
592 ((a|)+)+ did not notice that the outer group could match an empty string.
593 (^a|^)+ was not marked as anchored.
594 (.*a|.*)+ was not marked as matching at start or after a newline.
596 12. Yet more internal refactoring has removed another argument from the match()
597 function. Special calls to this function are now indicated by setting a
598 value in a variable in the "match data" data block.
600 13. Be more explicit in pcre_study() instead of relying on "default" for
601 opcodes that mean there is no starting character; this means that when new
602 ones are added and accidentally left out of pcre_study(), testing should
603 pick them up.
605 14. The -s option of pcretest has been documented for ages as being an old
606 synonym of -m (show memory usage). I have changed it to mean "force study
607 for every regex", that is, assume /S for every regex. This is similar to -i
608 and -d etc. It's slightly incompatible, but I'm hoping nobody is still
609 using it. It makes it easier to run collections of tests with and without
610 study enabled, and thereby test pcre_study() more easily. All the standard
611 tests are now run with and without -s (but some patterns can be marked as
612 "never study" - see 20 below).
614 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
615 restoration of the capturing data to the outer values was not happening
616 correctly.
618 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
619 empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
620 pattern had matched an empty string, and so incorrectly returned a no
621 match.
623 17. There was optimizing code for the last branch of non-capturing parentheses,
624 and also for the obeyed branch of a conditional subexpression, which used
625 tail recursion to cut down on stack usage. Unfortunately, now that there is
626 the possibility of (*THEN) occurring in these branches, tail recursion is
627 no longer possible because the return has to be checked for (*THEN). These
628 two optimizations have therefore been removed. [But see 8.20/11 above.]
630 18. If a pattern containing \R was studied, it was assumed that \R always
631 matched two bytes, thus causing the minimum subject length to be
632 incorrectly computed because \R can also match just one byte.
634 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
635 was incorrectly computed.
637 20. If /S is present twice on a test pattern in pcretest input, it now
638 *disables* studying, thereby overriding the use of -s on the command line
639 (see 14 above). This is necessary for one or two tests to keep the output
640 identical in both cases.
642 21. When (*ACCEPT) was used in an assertion that matched an empty string and
643 PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
645 22. When an atomic group that contained a capturing parenthesis was
646 successfully matched, but the branch in which it appeared failed, the
647 capturing was not being forgotten if a higher numbered group was later
648 captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
649 group 1 to "a", when in fact it should be unset. This applied to multi-
650 branched capturing and non-capturing groups, repeated or not, and also to
651 positive assertions (capturing in negative assertions does not happen
652 in PCRE) and also to nested atomic groups.
654 23. Add the ++ qualifier feature to pcretest, to show the remainder of the
655 subject after a captured substring, to make it easier to tell which of a
656 number of identical substrings has been captured.
658 24. The way atomic groups are processed by pcre_exec() has been changed so that
659 if they are repeated, backtracking one repetition now resets captured
660 values correctly. For example, if ((?>(a+)b)+aabab) is matched against
661 "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
662 "aaa". Previously, it would have been "a". As part of this code
663 refactoring, the way recursive calls are handled has also been changed.
665 25. If an assertion condition captured any substrings, they were not passed
666 back unless some other capturing happened later. For example, if
667 (?(?=(a))a) was matched against "a", no capturing was returned.
669 26. When studying a pattern that contained subroutine calls or assertions,
670 the code for finding the minimum length of a possible match was handling
671 direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
672 group 1 called group 2 while simultaneously a separate group 2 called group
673 1). A stack overflow occurred in this case. I have fixed this by limiting
674 the recursion depth to 10.
676 27. Updated RunTest.bat in the distribution to the version supplied by Tom
677 Fortmann. This supports explicit test numbers on the command line, and has
678 argument validation and error reporting.
680 28. An instance of \X with an unlimited repeat could fail if at any point the
681 first character it looked at was a mark character.
683 29. Some minor code refactoring concerning Unicode properties and scripts
684 should reduce the stack requirement of match() slightly.
686 30. Added the '=' option to pcretest to check the setting of unused capturing
687 slots at the end of the pattern, which are documented as being -1, but are
688 not included in the return count.
690 31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
691 compiled something random. Now it gives a compile-time error (as does
692 Perl).
694 32. A *MARK encountered during the processing of a positive assertion is now
695 recorded and passed back (compatible with Perl).
697 33. If --only-matching or --colour was set on a pcregrep call whose pattern
698 had alternative anchored branches, the search for a second match in a line
699 was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
700 matched the line "0102" twice. The same bug affected patterns that started
701 with a backwards assertion. For example /\b01|\b02/ also matched "0102"
702 twice.
704 34. Previously, PCRE did not allow quantification of assertions. However, Perl
705 does, and because of capturing effects, quantifying parenthesized
706 assertions may at times be useful. Quantifiers are now allowed for
707 parenthesized assertions.
709 35. A minor code tidy in pcre_compile() when checking options for \R usage.
711 36. \g was being checked for fancy things in a character class, when it should
712 just be a literal "g".
714 37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
715 appearance of a nested POSIX class supersedes an apparent external class.
716 For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
717 unescaped square brackets may also appear as part of class names. For
718 example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
719 more like Perl. (But see 8.20/1 above.)
721 38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
722 was because it thought it was \N{name}, which is not supported).
724 39. Add minix to OS list not supporting the -S option in pcretest.
726 40. PCRE tries to detect cases of infinite recursion at compile time, but it
727 cannot analyze patterns in sufficient detail to catch mutual recursions
728 such as ((?1))((?2)). There is now a runtime test that gives an error if a
729 subgroup is called recursively as a subpattern for a second time at the
730 same position in the subject string. In previous releases this might have
731 been caught by the recursion limit, or it might have run out of stack.
733 41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
734 happen only once. PCRE was, however incorrectly giving a compile time error
735 "recursive call could loop indefinitely" because it cannot analyze the
736 pattern in sufficient detail. The compile time test no longer happens when
737 PCRE is compiling a conditional subpattern, but actual runaway loops are
738 now caught at runtime (see 40 above).
740 42. It seems that Perl allows any characters other than a closing parenthesis
741 to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
742 has been changed to be the same.
744 43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
745 as not to get warnings when autogen.sh is called. Also changed
746 AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
748 44. To help people who use pcregrep to scan files containing exceedingly long
749 lines, the following changes have been made:
751 (a) The default value of the buffer size parameter has been increased from
752 8K to 20K. (The actual buffer used is three times this size.)
754 (b) The default can be changed by ./configure --with-pcregrep-bufsize when
755 PCRE is built.
757 (c) A --buffer-size=n option has been added to pcregrep, to allow the size
758 to be set at run time.
760 (d) Numerical values in pcregrep options can be followed by K or M, for
761 example --buffer-size=50K.
763 (e) If a line being scanned overflows pcregrep's buffer, an error is now
764 given and the return code is set to 2.
766 45. Add a pointer to the latest mark to the callout data block.
768 46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
769 partial match of an empty string instead of no match. This was specific to
770 the use of ".".
772 47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
773 complete match instead of a partial match. This bug was dependent on both
774 the PCRE_UTF8 and PCRE_DOTALL options being set.
776 48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
777 starting byte set, because \b was not being ignored.
780 Version 8.12 15-Jan-2011
781 ------------------------
783 1. Fixed some typos in the markup of the man pages, and wrote a script that
784 checks for such things as part of the documentation building process.
786 2. On a big-endian 64-bit system, pcregrep did not correctly process the
787 --match-limit and --recursion-limit options (added for 8.11). In
788 particular, this made one of the standard tests fail. (The integer value
789 went into the wrong half of a long int.)
791 3. If the --colour option was given to pcregrep with -v (invert match), it
792 did strange things, either producing crazy output, or crashing. It should,
793 of course, ignore a request for colour when reporting lines that do not
794 match.
796 4. Another pcregrep bug caused similar problems if --colour was specified with
797 -M (multiline) and the pattern match finished with a line ending.
799 5. In pcregrep, when a pattern that ended with a literal newline sequence was
800 matched in multiline mode, the following line was shown as part of the
801 match. This seems wrong, so I have changed it.
803 6. Another pcregrep bug in multiline mode, when --colour was specified, caused
804 the check for further matches in the same line (so they could be coloured)
805 to overrun the end of the current line. If another match was found, it was
806 incorrectly shown (and then shown again when found in the next line).
808 7. If pcregrep was compiled under Windows, there was a reference to the
809 function pcregrep_exit() before it was defined. I am assuming this was
810 the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
811 reported by a user. I've moved the definition above the reference.
814 Version 8.11 10-Dec-2010
815 ------------------------
817 1. (*THEN) was not working properly if there were untried alternatives prior
818 to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
819 backtracked to try for "b" instead of moving to the next alternative branch
820 at the same level (in this case, to look for "c"). The Perl documentation
821 is clear that when (*THEN) is backtracked onto, it goes to the "next
822 alternative in the innermost enclosing group".
824 2. (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
825 such as (A(*COMMIT)B(*THEN)C|D) any failure after matching A should
826 result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
827 (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
828 (*THEN).
830 3. If \s appeared in a character class, it removed the VT character from
831 the class, even if it had been included by some previous item, for example
832 in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
833 of \s, but is part of the POSIX "space" class.)
835 4. A partial match never returns an empty string (because you can always
836 match an empty string at the end of the subject); however the checking for
837 an empty string was starting at the "start of match" point. This has been
838 changed to the "earliest inspected character" point, because the returned
839 data for a partial match starts at this character. This means that, for
840 example, /(?<=abc)def/ gives a partial match for the subject "abc"
841 (previously it gave "no match").
843 5. Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
844 of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
845 previously a full match would be given. However, setting PCRE_PARTIAL_HARD
846 has an implication that the given string is incomplete (because a partial
847 match is preferred over a full match). For this reason, these items now
848 give a partial match in this situation. [Aside: previously, the one case
849 /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
850 match rather than a full match, which was wrong by the old rules, but is
851 now correct.]
853 6. There was a bug in the handling of #-introduced comments, recognized when
854 PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
855 If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
856 UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
857 scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
858 but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
859 places in pcre_compile().
861 7. Related to (6) above, when pcre_compile() was skipping #-introduced
862 comments when looking ahead for named forward references to subpatterns,
863 the only newline sequence it recognized was NL. It now handles newlines
864 according to the set newline convention.
866 8. SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
867 former, but used strtoul(), whereas pcretest avoided strtoul() but did not
868 cater for a lack of strerror(). These oversights have been fixed.
870 9. Added --match-limit and --recursion-limit to pcregrep.
872 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
874 11. When the -o option was used, pcregrep was setting a return code of 1, even
875 when matches were found, and --line-buffered was not being honoured.
877 12. Added an optional parentheses number to the -o and --only-matching options
878 of pcregrep.
880 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
881 can match an empty string. The code to do it in pcretest and pcredemo
882 needed fixing:
884 (a) When the newline convention was "crlf", pcretest got it wrong, skipping
885 only one byte after an empty string match just before CRLF (this case
886 just got forgotten; "any" and "anycrlf" were OK).
888 (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
889 mode when an empty string match preceded an ASCII character followed by
890 a non-ASCII character. (The code for advancing by one character rather
891 than one byte was nonsense.)
893 (c) The pcredemo.c sample program did not have any code at all to handle
894 the cases when CRLF is a valid newline sequence.
896 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
897 as a starting offset was within the subject string. There is now a new
898 error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
899 negative or greater than the length of the string. In order to test this,
900 pcretest is extended to allow the setting of negative starting offsets.
902 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
903 starting offset points to the beginning of a UTF-8 character was
904 unnecessarily clumsy. I tidied it up.
906 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
907 bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
909 17. Nobody had reported that the --include_dir option, which was added in
910 release 7.7 should have been called --include-dir (hyphen, not underscore)
911 for compatibility with GNU grep. I have changed it to --include-dir, but
912 left --include_dir as an undocumented synonym, and the same for
913 --exclude-dir, though that is not available in GNU grep, at least as of
914 release 2.5.4.
916 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
917 characters from a string of bytes) have been redefined so as not to use
918 loops, in order to improve performance in some environments. At the same
919 time, I abstracted some of the common code into auxiliary macros to save
920 repetition (this should not affect the compiled code).
922 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
923 compile-time error is now given if \c is not followed by an ASCII
924 character, that is, a byte less than 128. (In EBCDIC mode, the code is
925 different, and any byte value is allowed.)
927 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
928 START_OPTIMIZE option, which is now allowed at compile time - but just
929 passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
930 to pcregrep and other applications that have no direct access to PCRE
931 options. The new /Y option in pcretest sets this option when calling
932 pcre_compile().
934 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
935 back references. Groups containing recursive back references were forced to
936 be atomic by that change, but in the case of named groups, the amount of
937 memory required was incorrectly computed, leading to "Failed: internal
938 error: code overflow". This has been fixed.
940 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
941 pcretest.c, to avoid build problems in some Borland environments.
944 Version 8.10 25-Jun-2010
945 ------------------------
947 1. Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
948 THEN.
950 2. (*ACCEPT) was not working when inside an atomic group.
952 3. Inside a character class, \B is treated as a literal by default, but
953 faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
954 causes the error). The code is unchanged, but I tidied the documentation.
956 4. Inside a character class, PCRE always treated \R and \X as literals,
957 whereas Perl faults them if its -w option is set. I have changed PCRE so
958 that it faults them when PCRE_EXTRA is set.
960 5. Added support for \N, which always matches any character other than
961 newline. (It is the same as "." when PCRE_DOTALL is not set.)
963 6. When compiling pcregrep with newer versions of gcc which may have
964 FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
965 declared with attribute warn_unused_result" were given. Just casting the
966 result to (void) does not stop the warnings; a more elaborate fudge is
967 needed. I've used a macro to implement this.
969 7. Minor change to pcretest.c to avoid a compiler warning.
971 8. Added four artifical Unicode properties to help with an option to make
972 \s etc use properties (see next item). The new properties are: Xan
973 (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
975 9. Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
976 use Unicode properties. (*UCP) at the start of a pattern can be used to set
977 this option. Modified pcretest to add /W to test this facility. Added
978 REG_UCP to make it available via the POSIX interface.
980 10. Added --line-buffered to pcregrep.
982 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
983 studied, and the match started with a letter with a code point greater than
984 127 whose first byte was different to the first byte of the other case of
985 the letter, the other case of this starting letter was not recognized
986 (#976).
988 12. If a pattern that was studied started with a repeated Unicode property
989 test, for example, \p{Nd}+, there was the theoretical possibility of
990 setting up an incorrect bitmap of starting bytes, but fortunately it could
991 not have actually happened in practice until change 8 above was made (it
992 added property types that matched character-matching opcodes).
994 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
995 possible starting bytes for non-anchored patterns.
997 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
998 \R, and also a number of cases that involve Unicode properties, both
999 explicit and implicit when PCRE_UCP is set.
1001 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
1002 input, it could crash or give wrong results if characters with values
1003 greater than 0xc0 were present in the subject string. (Detail: it assumed
1004 UTF-8 input when processing these items.)
1006 16. Added a lot of (int) casts to avoid compiler warnings in systems where
1007 size_t is 64-bit (#991).
1009 17. Added a check for running out of memory when PCRE is compiled with
1010 --disable-stack-for-recursion (#990).
1012 18. If the last data line in a file for pcretest does not have a newline on
1013 the end, a newline was missing in the output.
1015 19. The default pcre_chartables.c file recognizes only ASCII characters (values
1016 less than 128) in its various bitmaps. However, there is a facility for
1017 generating tables according to the current locale when PCRE is compiled. It
1018 turns out that in some environments, 0x85 and 0xa0, which are Unicode space
1019 characters, are recognized by isspace() and therefore were getting set in
1020 these tables, and indeed these tables seem to approximate to ISO 8859. This
1021 caused a problem in UTF-8 mode when pcre_study() was used to create a list
1022 of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
1023 which of course cannot start UTF-8 characters. I have changed the code so
1024 that only real ASCII characters (less than 128) and the correct starting
1025 bytes for UTF-8 encodings are set for characters greater than 127 when in
1026 UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
1027 altogether.)
1029 20. Added the /T option to pcretest so as to be able to run tests with non-
1030 standard character tables, thus making it possible to include the tests
1031 used for 19 above in the standard set of tests.
1033 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
1034 reference to a subpattern the other side of a comment that contains an
1035 opening parenthesis caused either an internal compiling error, or a
1036 reference to the wrong subpattern.
1039 Version 8.02 19-Mar-2010
1040 ------------------------
1042 1. The Unicode data tables have been updated to Unicode 5.2.0.
1044 2. Added the option --libs-cpp to pcre-config, but only when C++ support is
1045 configured.
1047 3. Updated the licensing terms in the pcregexp.pas file, as agreed with the
1048 original author of that file, following a query about its status.
1050 4. On systems that do not have stdint.h (e.g. Solaris), check for and include
1051 inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
1053 5. A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
1054 quantifier applied to a forward-referencing subroutine call, could compile
1055 incorrect code or give the error "internal error: previously-checked
1056 referenced subpattern not found".
1058 6. Both MS Visual Studio and Symbian OS have problems with initializing
1059 variables to point to external functions. For these systems, therefore,
1060 pcre_malloc etc. are now initialized to local functions that call the
1061 relevant global functions.
1063 7. There were two entries missing in the vectors called coptable and poptable
1064 in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
1065 I've fixed the data, and added a kludgy way of testing at compile time that
1066 the lengths are correct (equal to the number of opcodes).
1068 8. Following on from 7, I added a similar kludge to check the length of the
1069 eint vector in pcreposix.c.
1071 9. Error texts for pcre_compile() are held as one long string to avoid too
1072 much relocation at load time. To find a text, the string is searched,
1073 counting zeros. There was no check for running off the end of the string,
1074 which could happen if a new error number was added without updating the
1075 string.
1077 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
1079 11. \K was not working if it appeared in an atomic group or in a group that
1080 was called as a "subroutine", or in an assertion. Perl 5.11 documents that
1081 \K is "not well defined" if used in an assertion. PCRE now accepts it if
1082 the assertion is positive, but not if it is negative.
1084 12. Change 11 fortuitously reduced the size of the stack frame used in the
1085 "match()" function of pcre_exec.c by one pointer. Forthcoming
1086 implementation of support for (*MARK) will need an extra pointer on the
1087 stack; I have reserved it now, so that the stack frame size does not
1088 decrease.
1090 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
1091 item in branch that calls a recursion is a subroutine call - as in the
1092 second branch in the above example - was incorrectly given the compile-
1093 time error "recursive call could loop indefinitely" because pcre_compile()
1094 was not correctly checking the subroutine for matching a non-empty string.
1096 14. The checks for overrunning compiling workspace could trigger after an
1097 overrun had occurred. This is a "should never occur" error, but it can be
1098 triggered by pathological patterns such as hundreds of nested parentheses.
1099 The checks now trigger 100 bytes before the end of the workspace.
1101 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
1104 Version 8.01 19-Jan-2010
1105 ------------------------
1107 1. If a pattern contained a conditional subpattern with only one branch (in
1108 particular, this includes all (*DEFINE) patterns), a call to pcre_study()
1109 computed the wrong minimum data length (which is of course zero for such
1110 subpatterns). This could cause incorrect "no match" results.
1112 2. For patterns such as (?i)a(?-i)b|c where an option setting at the start of
1113 the pattern is reset in the first branch, pcre_compile() failed with
1114 "internal error: code overflow at offset...". This happened only when
1115 the reset was to the original external option setting. (An optimization
1116 abstracts leading options settings into an external setting, which was the
1117 cause of this.)
1119 3. A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
1120 of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
1121 assertion pattern did not match (meaning that the assertion was true), it
1122 was incorrectly treated as false if the SKIP had been reached during the
1123 matching. This also applied to assertions used as conditions.
1125 4. If an item that is not supported by pcre_dfa_exec() was encountered in an
1126 assertion subpattern, including such a pattern used as a condition,
1127 unpredictable results occurred, instead of the error return
1130 5. The C++ GlobalReplace function was not working like Perl for the special
1131 situation when an empty string is matched. It now does the fancy magic
1132 stuff that is necessary.
1134 6. In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
1135 removed. (These were left over from very, very early versions of PCRE.)
1137 7. Some cosmetic changes to the code to make life easier when compiling it
1138 as part of something else:
1140 (a) Change DEBUG to PCRE_DEBUG.
1142 (b) In pcre_compile(), rename the member of the "branch_chain" structure
1143 called "current" as "current_branch", to prevent a collision with the
1144 Linux macro when compiled as a kernel module.
1146 (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
1147 prevent a collision with the Linux macro when compiled as a kernel
1148 module.
1150 8. In pcre_compile() there are some checks for integer overflows that used to
1151 cast potentially large values to (double). This has been changed to that
1152 when building, a check for int64_t is made, and if it is found, it is used
1153 instead, thus avoiding the use of floating point arithmetic. (There is no
1154 other use of FP in PCRE.) If int64_t is not found, the fallback is to
1155 double.
1157 9. Added two casts to avoid signed/unsigned warnings from VS Studio Express
1158 2005 (difference between two addresses compared to an unsigned value).
1160 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
1161 custom one, because of the following reported problem in Windows:
1163 - libbz2 uses the Pascal calling convention (WINAPI) for the functions
1164 under Win32.
1165 - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
1166 therefore missing the function definition.
1167 - The compiler thus generates a "C" signature for the test function.
1168 - The linker fails to find the "C" function.
1169 - PCRE fails to configure if asked to do so against libbz2.
1171 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
1172 messages were output:
1174 Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
1175 rerunning libtoolize, to keep the correct libtool macros in-tree.
1176 Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
1178 I have done both of these things.
1180 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
1181 most of the time, it *can* run out if it is given a pattern that contains a
1182 runaway infinite recursion. I updated the discussion in the pcrestack man
1183 page.
1185 13. Now that we have gone to the x.xx style of version numbers, the minor
1186 version may start with zero. Using 08 or 09 is a bad idea because users
1187 might check the value of PCRE_MINOR in their code, and 08 or 09 may be
1188 interpreted as invalid octal numbers. I've updated the previous comment in
1189 configure.ac, and also added a check that gives an error if 08 or 09 are
1190 used.
1192 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
1193 causing partial matching to fail when the end of the subject matched \W
1194 in a UTF-8 pattern where \W was quantified with a minimum of 3.
1196 15. There were some discrepancies between the declarations in pcre_internal.h
1197 of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
1198 their definitions. The declarations used "const uschar *" and the
1199 definitions used USPTR. Even though USPTR is normally defined as "const
1200 unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
1201 reported that: "This difference in casting confuses some C++ compilers, for
1202 example, SunCC recognizes above declarations as different functions and
1203 generates broken code for hbpcre." I have changed the declarations to use
1204 USPTR.
1206 16. GNU libtool is named differently on some systems. The autogen.sh script now
1207 tries several variants such as glibtoolize (MacOSX) and libtoolize1x
1208 (FreeBSD).
1210 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
1211 (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
1212 comment: "Figure out how to create a longlong from a string: strtoll and
1213 equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
1214 instance, but it only takes 2 args instead of 3!"
1216 18. A subtle bug concerned with back references has been fixed by a change of
1217 specification, with a corresponding code fix. A pattern such as
1218 ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
1219 refers, was giving matches when it shouldn't. For example, xa=xaaa would
1220 match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
1221 same bug. Such groups have to be quantified to be useful, or contained
1222 inside another quantified group. (If there's no repetition, the reference
1223 can never match.) The problem arises because, having left the group and
1224 moved on to the rest of the pattern, a later failure that backtracks into
1225 the group uses the captured value from the final iteration of the group
1226 rather than the correct earlier one. I have fixed this in PCRE by forcing
1227 any group that contains a reference to itself to be an atomic group; that
1228 is, there cannot be any backtracking into it once it has completed. This is
1229 similar to recursive and subroutine calls.
1232 Version 8.00 19-Oct-09
1233 ----------------------
1235 1. The table for translating pcre_compile() error codes into POSIX error codes
1236 was out-of-date, and there was no check on the pcre_compile() error code
1237 being within the table. This could lead to an OK return being given in
1238 error.
1240 2. Changed the call to open a subject file in pcregrep from fopen(pathname,
1241 "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
1242 in a Windows environment.
1244 3. The pcregrep --count option prints the count for each file even when it is
1245 zero, as does GNU grep. However, pcregrep was also printing all files when
1246 --files-with-matches was added. Now, when both options are given, it prints
1247 counts only for those files that have at least one match. (GNU grep just
1248 prints the file name in this circumstance, but including the count seems
1249 more useful - otherwise, why use --count?) Also ensured that the
1250 combination -clh just lists non-zero counts, with no names.
1252 4. The long form of the pcregrep -F option was incorrectly implemented as
1253 --fixed_strings instead of --fixed-strings. This is an incompatible change,
1254 but it seems right to fix it, and I didn't think it was worth preserving
1255 the old behaviour.
1257 5. The command line items --regex=pattern and --regexp=pattern were not
1258 recognized by pcregrep, which required --regex pattern or --regexp pattern
1259 (with a space rather than an '='). The man page documented the '=' forms,
1260 which are compatible with GNU grep; these now work.
1262 6. No libpcreposix.pc file was created for pkg-config; there was just
1263 libpcre.pc and libpcrecpp.pc. The omission has been rectified.
1265 7. Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
1266 when UCP support is not needed, by modifying the Python script that
1267 generates it from Unicode data files. This should not matter if the module
1268 is correctly used as a library, but I received one complaint about 50K of
1269 unwanted data. My guess is that the person linked everything into his
1270 program rather than using a library. Anyway, it does no harm.
1272 8. A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
1273 was a minimum greater than 1 for a wide character in a possessive
1274 repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
1275 which had an unlimited repeat of a nested, fixed maximum repeat of a wide
1276 character. Chaos in the form of incorrect output or a compiling loop could
1277 result.
1279 9. The restrictions on what a pattern can contain when partial matching is
1280 requested for pcre_exec() have been removed. All patterns can now be
1281 partially matched by this function. In addition, if there are at least two
1282 slots in the offset vector, the offset of the earliest inspected character
1283 for the match and the offset of the end of the subject are set in them when
1284 PCRE_ERROR_PARTIAL is returned.
1286 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
1287 synonymous with PCRE_PARTIAL, for backwards compatibility, and
1288 PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
1289 and may be more useful for multi-segment matching.
1291 11. Partial matching with pcre_exec() is now more intuitive. A partial match
1292 used to be given if ever the end of the subject was reached; now it is
1293 given only if matching could not proceed because another character was
1294 needed. This makes a difference in some odd cases such as Z(*FAIL) with the
1295 string "Z", which now yields "no match" instead of "partial match". In the
1296 case of pcre_dfa_exec(), "no match" is given if every matching path for the
1297 final character ended with (*FAIL).
1299 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
1300 if the pattern had a "must contain" character that was already found in the
1301 earlier partial match, unless partial matching was again requested. For
1302 example, with the pattern /dog.(body)?/, the "must contain" character is
1303 "g". If the first part-match was for the string "dog", restarting with
1304 "sbody" failed. This bug has been fixed.
1306 13. The string returned by pcre_dfa_exec() after a partial match has been
1307 changed so that it starts at the first inspected character rather than the
1308 first character of the match. This makes a difference only if the pattern
1309 starts with a lookbehind assertion or \b or \B (\K is not supported by
1310 pcre_dfa_exec()). It's an incompatible change, but it makes the two
1311 matching functions compatible, and I think it's the right thing to do.
1313 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
1314 so that the demonstration program is easily available in environments where
1315 PCRE has not been installed from source.
1317 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
1318 libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
1319 library.
1321 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
1322 It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
1323 is not the first non-POSIX option to be added. Clearly some people find
1324 these options useful.
1326 17. If a caller to the POSIX matching function regexec() passes a non-zero
1327 value for nmatch with a NULL value for pmatch, the value of
1328 nmatch is forced to zero.
1330 18. RunGrepTest did not have a test for the availability of the -u option of
1331 the diff command, as RunTest does. It now checks in the same way as
1332 RunTest, and also checks for the -b option.
1334 19. If an odd number of negated classes containing just a single character
1335 interposed, within parentheses, between a forward reference to a named
1336 subpattern and the definition of the subpattern, compilation crashed with
1337 an internal error, complaining that it could not find the referenced
1338 subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
1339 [The bug was that it was starting one character too far in when skipping
1340 over the character class, thus treating the ] as data rather than
1341 terminating the class. This meant it could skip too much.]
1343 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
1344 /g option in pcretest when the pattern contains \K, which makes it possible
1345 to have an empty string match not at the start, even when the pattern is
1346 anchored. Updated pcretest and pcredemo to use this option.
1348 21. If the maximum number of capturing subpatterns in a recursion was greater
1349 than the maximum at the outer level, the higher number was returned, but
1350 with unset values at the outer level. The correct (outer level) value is
1351 now given.
1353 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
1354 PCRE did not set those parentheses (unlike Perl). I have now found a way to
1355 make it do so. The string so far is captured, making this feature
1356 compatible with Perl.
1358 23. The tests have been re-organized, adding tests 11 and 12, to make it
1359 possible to check the Perl 5.10 features against Perl 5.10.
1361 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
1362 pattern matches a fixed length string. PCRE did not allow this; now it
1363 does. Neither allows recursion.
1365 25. I finally figured out how to implement a request to provide the minimum
1366 length of subject string that was needed in order to match a given pattern.
1367 (It was back references and recursion that I had previously got hung up
1368 on.) This code has now been added to pcre_study(); it finds a lower bound
1369 to the length of subject needed. It is not necessarily the greatest lower
1370 bound, but using it to avoid searching strings that are too short does give
1371 some useful speed-ups. The value is available to calling programs via
1372 pcre_fullinfo().
1374 26. While implementing 25, I discovered to my embarrassment that pcretest had
1375 not been passing the result of pcre_study() to pcre_dfa_exec(), so the
1376 study optimizations had never been tested with that matching function.
1377 Oops. What is worse, even when it was passed study data, there was a bug in
1378 pcre_dfa_exec() that meant it never actually used it. Double oops. There
1379 were also very few tests of studied patterns with pcre_dfa_exec().
1381 27. If (?| is used to create subpatterns with duplicate numbers, they are now
1382 allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
1383 on the other side of the coin, they are no longer allowed to have different
1384 names, because these cannot be distinguished in PCRE, and this has caused
1385 confusion. (This is a difference from Perl.)
1387 28. When duplicate subpattern names are present (necessarily with different
1388 numbers, as required by 27 above), and a test is made by name in a
1389 conditional pattern, either for a subpattern having been matched, or for
1390 recursion in such a pattern, all the associated numbered subpatterns are
1391 tested, and the overall condition is true if the condition is true for any
1392 one of them. This is the way Perl works, and is also more like the way
1393 testing by number works.
1396 Version 7.9 11-Apr-09
1397 ---------------------
1399 1. When building with support for bzlib/zlib (pcregrep) and/or readline
1400 (pcretest), all targets were linked against these libraries. This included
1401 libpcre, libpcreposix, and libpcrecpp, even though they do not use these
1402 libraries. This caused unwanted dependencies to be created. This problem
1403 has been fixed, and now only pcregrep is linked with bzlib/zlib and only
1404 pcretest is linked with readline.
1406 2. The "typedef int BOOL" in pcre_internal.h that was included inside the
1407 "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
1408 moved outside it again, because FALSE and TRUE are already defined in AIX,
1409 but BOOL is not.
1411 3. The pcre_config() function was treating the PCRE_MATCH_LIMIT and
1412 PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
1414 4. The pcregrep documentation said spaces were inserted as well as colons (or
1415 hyphens) following file names and line numbers when outputting matching
1416 lines. This is not true; no spaces are inserted. I have also clarified the
1417 wording for the --colour (or --color) option.
1419 5. In pcregrep, when --colour was used with -o, the list of matching strings
1420 was not coloured; this is different to GNU grep, so I have changed it to be
1421 the same.
1423 6. When --colo(u)r was used in pcregrep, only the first matching substring in
1424 each matching line was coloured. Now it goes on to look for further matches
1425 of any of the test patterns, which is the same behaviour as GNU grep.
1427 7. A pattern that could match an empty string could cause pcregrep to loop; it
1428 doesn't make sense to accept an empty string match in pcregrep, so I have
1429 locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
1430 seems to be how GNU grep behaves.
1432 8. The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
1433 start or after a newline", because the conditional assertion was not being
1434 correctly handled. The rule now is that both the assertion and what follows
1435 in the first alternative must satisfy the test.
1437 9. If auto-callout was enabled in a pattern with a conditional group whose
1438 condition was an assertion, PCRE could crash during matching, both with
1439 pcre_exec() and pcre_dfa_exec().
1441 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
1442 used for matching.
1444 11. Unicode property support in character classes was not working for
1445 characters (bytes) greater than 127 when not in UTF-8 mode.
1447 12. Added the -M command line option to pcretest.
1449 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
1451 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
1453 16. Added comments and documentation about mis-use of no_arg in the C++
1454 wrapper.
1456 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
1457 from Martin Jerabek that uses macro names for all relevant character and
1458 string constants.
1460 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
1461 SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
1462 SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
1463 these, but not everybody uses configure.
1465 19. A conditional group that had only one branch was not being correctly
1466 recognized as an item that could match an empty string. This meant that an
1467 enclosing group might also not be so recognized, causing infinite looping
1468 (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
1469 with the subject "ab", where knowledge that the repeated group can match
1470 nothing is needed in order to break the loop.
1472 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
1473 exec(), but without supplying a callout function, matching went wrong.
1475 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
1476 leak if the size of the offset vector was greater than 30. When the vector
1477 is smaller, the saved offsets during recursion go onto a local stack
1478 vector, but for larger vectors malloc() is used. It was failing to free
1479 when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
1480 error, in fact).
1482 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
1483 heapframe that is used only when UTF-8 support is enabled. This caused no
1484 problem, but was untidy.
1486 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
1487 CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
1488 included within another project.
1490 24. Steven Van Ingelgem's patches to add more options to the CMake support,
1491 slightly modified by me:
1493 (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
1494 not building pcregrep.
1496 (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
1497 if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
1499 25. Forward references, both numeric and by name, in patterns that made use of
1500 duplicate group numbers, could behave incorrectly or give incorrect errors,
1501 because when scanning forward to find the reference group, PCRE was not
1502 taking into account the duplicate group numbers. A pattern such as
1503 ^X(?3)(a)(?|(b)|(q))(Y) is an example.
1505 26. Changed a few more instances of "const unsigned char *" to USPTR, making
1506 the feature of a custom pointer more persuasive (as requested by a user).
1508 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
1509 pcretest.c, inside #ifndefs, because it seems they are sometimes already
1510 pre-defined.
1512 28. Added support for (*UTF8) at the start of a pattern.
1514 29. Arrange for flags added by the "release type" setting in CMake to be shown
1515 in the configuration summary.
1518 Version 7.8 05-Sep-08
1519 ---------------------
1521 1. Replaced UCP searching code with optimized version as implemented for Ad
1522 Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
1523 stage table and inline lookup instead of a function, giving speed ups of 2
1524 to 5 times on some simple patterns that I tested. Permission was given to
1525 distribute the MultiStage2.py script that generates the tables (it's not in
1526 the tarball, but is in the Subversion repository).
1528 2. Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
1529 scripts.
1531 3. Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
1532 a group with a zero qualifier. The result of the study could be incorrect,
1533 or the function might crash, depending on the pattern.
1535 4. Caseless matching was not working for non-ASCII characters in back
1536 references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
1537 It now works when Unicode Property Support is available.
1539 5. In pcretest, an escape such as \x{de} in the data was always generating
1540 a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
1541 non-UTF-8 mode. If the value is greater than 255, it gives a warning about
1542 truncation.
1544 6. Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
1546 7. Added two (int) casts to pcregrep when printing the difference of two
1547 pointers, in case they are 64-bit values.
1549 8. Added comments about Mac OS X stack usage to the pcrestack man page and to
1550 test 2 if it fails.
1552 9. Added PCRE_CALL_CONVENTION just before the names of all exported functions,
1553 and a #define of that name to empty if it is not externally set. This is to
1554 allow users of MSVC to set it if necessary.
1556 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
1557 the convenience functions in the pcre_get.c source file.
1559 11. An option change at the start of a pattern that had top-level alternatives
1560 could cause overwriting and/or a crash. This command provoked a crash in
1561 some environments:
1563 printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
1565 This potential security problem was recorded as CVE-2008-2371.
1567 12. For a pattern where the match had to start at the beginning or immediately
1568 after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
1569 pcre_dfa_exec() could read past the end of the passed subject if there was
1570 no match. To help with detecting such bugs (e.g. with valgrind), I modified
1571 pcretest so that it places the subject at the end of its malloc-ed buffer.
1573 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
1574 exec() might read past the end of the data buffer in UTF-8 mode.
1576 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
1577 the data contained the byte 0x85 as part of a UTF-8 character within its
1578 first line. This applied both to normal and DFA matching.
1580 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
1581 /^[^d]*?$/8 failed to match "abc".
1583 16. Added a missing copyright notice to pcrecpp_internal.h.
1585 17. Make it more clear in the documentation that values returned from
1586 pcre_exec() in ovector are byte offsets, not character counts.
1588 18. Tidied a few places to stop certain compilers from issuing warnings.
1590 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
1591 supplied by Stefan Weber. I made a further small update for 7.8 because
1592 there is a change of source arrangements: the pcre_searchfuncs.c module is
1593 replaced by pcre_ucd.c.
1596 Version 7.7 07-May-08
1597 ---------------------
1599 1. Applied Craig's patch to sort out a long long problem: "If we can't convert
1600 a string to a long long, pretend we don't even have a long long." This is
1601 done by checking for the strtoq, strtoll, and _strtoi64 functions.
1603 2. Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
1604 pre-7.6 versions, which defined a global no_arg variable instead of putting
1605 it in the RE class. (See also #8 below.)
1607 3. Remove a line of dead code, identified by coverity and reported by Nuno
1608 Lopes.
1610 4. Fixed two related pcregrep bugs involving -r with --include or --exclude:
1612 (1) The include/exclude patterns were being applied to the whole pathnames
1613 of files, instead of just to the final components.
1615 (2) If there was more than one level of directory, the subdirectories were
1616 skipped unless they satisfied the include/exclude conditions. This is
1617 inconsistent with GNU grep (and could even be seen as contrary to the
1618 pcregrep specification - which I improved to make it absolutely clear).
1619 The action now is always to scan all levels of directory, and just
1620 apply the include/exclude patterns to regular files.
1622 5. Added the --include_dir and --exclude_dir patterns to pcregrep, and used
1623 --exclude_dir in the tests to avoid scanning .svn directories.
1625 6. Applied Craig's patch to the QuoteMeta function so that it escapes the
1626 NUL character as backslash + 0 rather than backslash + NUL, because PCRE
1627 doesn't support NULs in patterns.
1629 7. Added some missing "const"s to declarations of static tables in
1630 pcre_compile.c and pcre_dfa_exec.c.
1632 8. Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
1633 caused by fix #2 above. (Subsequently also a second patch to fix the
1634 first patch. And a third patch - this was a messy problem.)
1636 9. Applied Craig's patch to remove the use of push_back().
1638 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
1639 matching function regexec().
1641 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
1642 which, however, unlike Perl's \g{...}, are subroutine calls, not back
1643 references. PCRE supports relative numbers with this syntax (I don't think
1644 Oniguruma does).
1646 12. Previously, a group with a zero repeat such as (...){0} was completely
1647 omitted from the compiled regex. However, this means that if the group
1648 was called as a subroutine from elsewhere in the pattern, things went wrong
1649 (an internal error was given). Such groups are now left in the compiled
1650 pattern, with a new opcode that causes them to be skipped at execution
1651 time.
1653 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
1654 to the way PCRE behaves:
1656 (a) A lone ] character is dis-allowed (Perl treats it as data).
1658 (b) A back reference to an unmatched subpattern matches an empty string
1659 (Perl fails the current match path).
1661 (c) A data ] in a character class must be notated as \] because if the
1662 first data character in a class is ], it defines an empty class. (In
1663 Perl it is not possible to have an empty class.) The empty class []
1664 never matches; it forces failure and is equivalent to (*FAIL) or (?!).
1665 The negative empty class [^] matches any one character, independently
1666 of the DOTALL setting.
1668 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
1669 non-existent subpattern following a character class starting with ']' and
1670 containing () gave an internal compiling error instead of "reference to
1671 non-existent subpattern". Fortunately, when the pattern did exist, the
1672 compiled code was correct. (When scanning forwards to check for the
1673 existencd of the subpattern, it was treating the data ']' as terminating
1674 the class, so got the count wrong. When actually compiling, the reference
1675 was subsequently set up correctly.)
1677 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
1678 it was being rejected as not supported by pcre_dfa_exec(), even though
1679 other assertions are supported. I have made pcre_dfa_exec() support
1680 (*FAIL).
1682 16. The implementation of 13c above involved the invention of a new opcode,
1683 OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
1684 cannot be changed at match time, I realized I could make a small
1685 improvement to matching performance by compiling OP_ALLANY instead of
1686 OP_ANY for "." when DOTALL was set, and then removing the runtime tests
1687 on the OP_ANY path.
1689 17. Compiling pcretest on Windows with readline support failed without the
1690 following two fixes: (1) Make the unistd.h include conditional on
1691 HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
1693 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
1694 ncurses library to be included for pcretest when ReadLine support is
1695 requested, but also to allow for it to be overridden. This patch came from
1696 Daniel Bergström.
1698 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
1699 as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
1700 any errors with the current Unicode tables. Thanks to Peter Kankowski for
1701 spotting this.
1704 Version 7.6 28-Jan-08
1705 ---------------------
1707 1. A character class containing a very large number of characters with
1708 codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
1709 overflow.
1711 2. Patch to cut out the "long long" test in pcrecpp_unittest when
1712 HAVE_LONG_LONG is not defined.
1714 3. Applied Christian Ehrlicher's patch to update the CMake build files to
1715 bring them up to date and include new features. This patch includes:
1717 - Fixed PH's badly added libz and libbz2 support.
1718 - Fixed a problem with static linking.
1719 - Added pcredemo. [But later removed - see 7 below.]
1720 - Fixed dftables problem and added an option.
1721 - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
1723 - Added readline support for pcretest.
1724 - Added an listing of the option settings after cmake has run.
1726 4. A user submitted a patch to Makefile that makes it easy to create
1727 "pcre.dll" under mingw when using Configure/Make. I added stuff to
1728 Makefile.am that cause it to include this special target, without
1729 affecting anything else. Note that the same mingw target plus all
1730 the other distribution libraries and programs are now supported
1731 when configuring with CMake (see 6 below) instead of with
1732 Configure/Make.
1734 5. Applied Craig's patch that moves no_arg into the RE class in the C++ code.
1735 This is an attempt to solve the reported problem "pcrecpp::no_arg is not
1736 exported in the Windows port". It has not yet been confirmed that the patch
1737 solves the problem, but it does no harm.
1739 6. Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
1740 NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
1741 with CMake, and also correct the comment about stack recursion.
1743 7. Remove the automatic building of pcredemo from the ./configure system and
1744 from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
1745 of a program that users should build themselves after PCRE is installed, so
1746 building it automatically is not really right. What is more, it gave
1747 trouble in some build environments.
1749 8. Further tidies to CMakeLists.txt from Sheri and Christian.
1752 Version 7.5 10-Jan-08
1753 ---------------------
1755 1. Applied a patch from Craig: "This patch makes it possible to 'ignore'
1756 values in parens when parsing an RE using the C++ wrapper."
1758 2. Negative specials like \S did not work in character classes in UTF-8 mode.
1759 Characters greater than 255 were excluded from the class instead of being
1760 included.
1762 3. The same bug as (2) above applied to negated POSIX classes such as
1763 [:^space:].
1765 4. PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
1766 defined or documented. It seems to have been a typo for PCRE_STATIC, so
1767 I have changed it.
1769 5. The construct (?&) was not diagnosed as a syntax error (it referenced the
1770 first named subpattern) and a construct such as (?&a) would reference the
1771 first named subpattern whose name started with "a" (in other words, the
1772 length check was missing). Both these problems are fixed. "Subpattern name
1773 expected" is now given for (?&) (a zero-length name), and this patch also
1774 makes it give the same error for \k'' (previously it complained that that
1775 was a reference to a non-existent subpattern).
1777 6. The erroneous patterns (?+-a) and (?-+a) give different error messages;
1778 this is right because (?- can be followed by option settings as well as by
1779 digits. I have, however, made the messages clearer.
1781 7. Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
1782 than the number used in the conditional) now cause a compile-time error.
1783 This is actually not compatible with Perl, which accepts such patterns, but
1784 treats the conditional as always being FALSE (as PCRE used to), but it
1785 seems to me that giving a diagnostic is better.
1787 8. Change "alphameric" to the more common word "alphanumeric" in comments
1788 and messages.
1790 9. Fix two occurrences of "backslash" in comments that should have been
1791 "backspace".
1793 10. Remove two redundant lines of code that can never be obeyed (their function
1794 was moved elsewhere).
1796 11. The program that makes PCRE's Unicode character property table had a bug
1797 which caused it to generate incorrect table entries for sequences of
1798 characters that have the same character type, but are in different scripts.
1799 It amalgamated them into a single range, with the script of the first of
1800 them. In other words, some characters were in the wrong script. There were
1801 thirteen such cases, affecting characters in the following ranges:
1803 U+002b0 - U+002c1
1804 U+0060c - U+0060d
1805 U+0061e - U+00612
1806 U+0064b - U+0065e
1807 U+0074d - U+0076d
1808 U+01800 - U+01805
1809 U+01d00 - U+01d77
1810 U+01d9b - U+01dbf
1811 U+0200b - U+0200f
1812 U+030fc - U+030fe
1813 U+03260 - U+0327f
1814 U+0fb46 - U+0fbb1
1815 U+10450 - U+1049d
1817 12. The -o option (show only the matching part of a line) for pcregrep was not
1818 compatible with GNU grep in that, if there was more than one match in a
1819 line, it showed only the first of them. It now behaves in the same way as
1820 GNU grep.
1822 13. If the -o and -v options were combined for pcregrep, it printed a blank
1823 line for every non-matching line. GNU grep prints nothing, and pcregrep now
1824 does the same. The return code can be used to tell if there were any
1825 non-matching lines.
1827 14. Added --file-offsets and --line-offsets to pcregrep.
1829 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
1830 infinitely looping recursion. The bug was that positive lookaheads were not
1831 being skipped when checking for a possible empty match (negative lookaheads
1832 and both kinds of lookbehind were skipped).
1834 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
1835 inclusion of <windows.h> to before rather than after the definition of
1836 INVALID_FILE_ATTRIBUTES (patch from David Byron).
1838 17. Specifying a possessive quantifier with a specific limit for a Unicode
1839 character property caused pcre_compile() to compile bad code, which led at
1840 runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
1841 are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
1842 caused the error; without that there was no problem.
1844 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
1846 19. Added --enable-pcretest-libreadline.
1848 20. In pcrecpp.cc, the variable 'count' was incremented twice in
1849 RE::GlobalReplace(). As a result, the number of replacements returned was
1850 double what it should be. I removed one of the increments, but Craig sent a
1851 later patch that removed the other one (the right fix) and added unit tests
1852 that check the return values (which was not done before).
1854 21. Several CMake things:
1856 (1) Arranged that, when cmake is used on Unix, the libraries end up with
1857 the names libpcre and libpcreposix, not just pcre and pcreposix.
1859 (2) The above change means that pcretest and pcregrep are now correctly
1860 linked with the newly-built libraries, not previously installed ones.
1864 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
1865 crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
1866 UTF-8 newline character). The key issue is that the pattern starts .*;
1867 this means that the match must be either at the beginning, or after a
1868 newline. The bug was in the code for advancing after a failed match and
1869 checking that the new position followed a newline. It was not taking
1870 account of UTF-8 characters correctly.
1872 23. PCRE was behaving differently from Perl in the way it recognized POSIX
1873 character classes. PCRE was not treating the sequence [:...:] as a
1874 character class unless the ... were all letters. Perl, however, seems to
1875 allow any characters between [: and :], though of course it rejects as
1876 unknown any "names" that contain non-letters, because all the known class
1877 names consist only of letters. Thus, Perl gives an error for [[:1234:]],
1878 for example, whereas PCRE did not - it did not recognize a POSIX character
1879 class. This seemed a bit dangerous, so the code has been changed to be
1880 closer to Perl. The behaviour is not identical to Perl, because PCRE will
1881 diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
1882 treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
1883 Perl does, and where it didn't before.
1885 24. Rewrite so as to remove the single use of %n from pcregrep because in some
1886 Windows environments %n is disabled by default.
1889 Version 7.4 21-Sep-07
1890 ---------------------
1892 1. Change 7.3/28 was implemented for classes by looking at the bitmap. This
1893 means that a class such as [\s] counted as "explicit reference to CR or
1894 LF". That isn't really right - the whole point of the change was to try to
1895 help when there was an actual mention of one of the two characters. So now
1896 the change happens only if \r or \n (or a literal CR or LF) character is
1897 encountered.
1899 2. The 32-bit options word was also used for 6 internal flags, but the numbers
1900 of both had grown to the point where there were only 3 bits left.
1901 Fortunately, there was spare space in the data structure, and so I have
1902 moved the internal flags into a new 16-bit field to free up more option
1903 bits.
1905 3. The appearance of (?J) at the start of a pattern set the DUPNAMES option,
1906 but did not set the internal JCHANGED flag - either of these is enough to
1907 control the way the "get" function works - but the PCRE_INFO_JCHANGED
1908 facility is supposed to tell if (?J) was ever used, so now (?J) at the
1909 start sets both bits.
1911 4. Added options (at build time, compile time, exec time) to change \R from
1912 matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
1914 5. doc/pcresyntax.html was missing from the distribution.
1916 6. Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
1917 compatibility, even though it is no longer used.
1919 7. Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
1920 strtoull to pcrecpp.cc to select the available functions in WIN32 when the
1921 windows.h file is present (where different names are used). [This was
1922 reversed later after testing - see 16 below.]
1924 8. Changed all #include <config.h> to #include "config.h". There were also
1925 some further <pcre.h> cases that I changed to "pcre.h".
1927 9. When pcregrep was used with the --colour option, it missed the line ending
1928 sequence off the lines that it output.
1930 10. It was pointed out to me that arrays of string pointers cause lots of
1931 relocations when a shared library is dynamically loaded. A technique of
1932 using a single long string with a table of offsets can drastically reduce
1933 these. I have refactored PCRE in four places to do this. The result is
1934 dramatic:
1936 Originally: 290
1937 After changing UCP table: 187
1938 After changing error message table: 43
1939 After changing table of "verbs" 36
1940 After changing table of Posix names 22
1942 Thanks to the folks working on Gregex for glib for this insight.
1944 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
1945 unicode-properties was also set.
1947 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
1949 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
1950 checked only for CRLF.
1952 14. Added casts to pcretest.c to avoid compiler warnings.
1954 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
1956 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
1957 and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
1958 entirely. This removes changes made in 7 above.
1960 17. The CMake files have been updated, and there is now more information about
1961 building with CMake in the NON-UNIX-USE document.
1964 Version 7.3 28-Aug-07
1965 ---------------------
1967 1. In the rejigging of the build system that eventually resulted in 7.1, the
1968 line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
1969 brackets there is not right, since it causes compilers to look for an
1970 installed pcre.h, not the version that is in the source that is being
1971 compiled (which of course may be different). I have changed it back to:
1973 #include "pcre.h"
1975 I have a vague recollection that the change was concerned with compiling in
1976 different directories, but in the new build system, that is taken care of
1977 by the VPATH setting the Makefile.
1979 2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
1980 when the subject happened to end in the byte 0x85 (e.g. if the last
1981 character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
1982 characters but of course it shouldn't be taken as a newline when it is part
1983 of another character. The bug was that, for an unlimited repeat of . in
1984 not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
1985 characters when looking for a newline.
1987 3. A small performance improvement in the DOTALL UTF-8 mode .* case.
1989 4. Debugging: adjusted the names of opcodes for different kinds of parentheses
1990 in debug output.
1992 5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
1993 long printing in the pcrecpp unittest when running under MinGW.
1995 6. ESC_K was left out of the EBCDIC table.
1997 7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
1998 parentheses; I made it 1000, which seemed large enough. Unfortunately, the
1999 limit also applies to "virtual nesting" when a pattern is recursive, and in
2000 this case 1000 isn't so big. I have been able to remove this limit at the
2001 expense of backing off one optimization in certain circumstances. Normally,
2002 when pcre_exec() would call its internal match() function recursively and
2003 immediately return the result unconditionally, it uses a "tail recursion"
2004 feature to save stack. However, when a subpattern that can match an empty
2005 string has an unlimited repetition quantifier, it no longer makes this
2006 optimization. That gives it a stack frame in which to save the data for
2007 checking that an empty string has been matched. Previously this was taken
2008 from the 1000-entry workspace that had been reserved. So now there is no
2009 explicit limit, but more stack is used.
2011 8. Applied Daniel's patches to solve problems with the import/export magic
2012 syntax that is required for Windows, and which was going wrong for the
2013 pcreposix and pcrecpp parts of the library. These were overlooked when this
2014 problem was solved for the main library.
2016 9. There were some crude static tests to avoid integer overflow when computing
2017 the size of patterns that contain repeated groups with explicit upper
2018 limits. As the maximum quantifier is 65535, the maximum group length was
2019 set at 30,000 so that the product of these two numbers did not overflow a
2020 32-bit integer. However, it turns out that people want to use groups that
2021 are longer than 30,000 bytes (though not repeat them that many times).
2022 Change 7.0/17 (the refactoring of the way the pattern size is computed) has
2023 made it possible to implement the integer overflow checks in a much more
2024 dynamic way, which I have now done. The artificial limitation on group
2025 length has been removed - we now have only the limit on the total length of
2026 the compiled pattern, which depends on the LINK_SIZE setting.
2028 10. Fixed a bug in the documentation for get/copy named substring when
2029 duplicate names are permitted. If none of the named substrings are set, the
2030 functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
2031 empty string.
2033 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
2034 instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
2035 because the ] is interpreted as the first data character and the
2036 terminating ] is not found. PCRE has been made compatible with Perl in this
2037 regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
2038 cause memory overwriting.
2040 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
2041 string has been matched (to stop an infinite loop). It was not recognizing
2042 a conditional subpattern that could match an empty string if that
2043 subpattern was within another subpattern. For example, it looped when
2044 trying to match (((?(1)X|))*) but it was OK with ((?(1)X|)*) where the
2045 condition was not nested. This bug has been fixed.
2047 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
2048 past the start of the subject in the presence of bytes with the top bit
2049 set, for example "\x8aBCD".
2051 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
2052 (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
2054 14. Optimized (?!) to (*FAIL).
2056 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
2057 This restricts code points to be within the range 0 to 0x10FFFF, excluding
2058 the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
2059 full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
2060 does: it's just the validity check that is more restrictive.
2062 16. Inserted checks for integer overflows during escape sequence (backslash)
2063 processing, and also fixed erroneous offset values for syntax errors during
2064 backslash processing.
2066 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
2067 for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
2069 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
2070 caused an overrun.
2072 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
2073 something other than just ASCII characters) inside a group that had an
2074 unlimited repeat caused a loop at compile time (while checking to see
2075 whether the group could match an empty string).
2077 20. Debugging a pattern containing \p or \P could cause a crash. For example,
2078 [\P{Any}] did so. (Error in the code for printing property names.)
2080 21. An orphan \E inside a character class could cause a crash.
2082 22. A repeated capturing bracket such as (A)? could cause a wild memory
2083 reference during compilation.
2085 23. There are several functions in pcre_compile() that scan along a compiled
2086 expression for various reasons (e.g. to see if it's fixed length for look
2087 behind). There were bugs in these functions when a repeated \p or \P was
2088 present in the pattern. These operators have additional parameters compared
2089 with \d, etc, and these were not being taken into account when moving along
2090 the compiled data. Specifically:
2092 (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
2093 length.
2095 (b) An item such as \pL+ within a repeated group could cause crashes or
2096 loops.
2098 (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
2099 "reference to non-existent subpattern" error.
2101 (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
2103 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
2104 characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
2106 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
2108 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
2109 character were causing crashes (broken optimization).
2111 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
2112 \p or \P) caused a compile-time loop.
2114 28. More problems have arisen in unanchored patterns when CRLF is a valid line
2115 break. For example, the unstudied pattern [\r\n]A does not match the string
2116 "\r\nA" because change 7.0/46 below moves the current point on by two
2117 characters after failing to match at the start. However, the pattern \nA
2118 *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
2119 the same is true. There doesn't seem any very clean way out of this, but
2120 what I have chosen to do makes the common cases work: PCRE now takes note
2121 of whether there can be an explicit match for \r or \n anywhere in the
2122 pattern, and if so, 7.0/46 no longer applies. As part of this change,
2123 there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
2124 pattern has explicit CR or LF references.
2126 29. Added (*CR) etc for changing newline setting at start of pattern.
2129 Version 7.2 19-Jun-07
2130 ---------------------
2132 1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
2133 which is apparently normally available under Windows.
2135 2. Re-jig the pcregrep tests with different newline settings in an attempt
2136 to make them independent of the local environment's newline setting.
2138 3. Add code to configure.ac to remove -g from the CFLAGS default settings.
2140 4. Some of the "internals" tests were previously cut out when the link size
2141 was not 2, because the output contained actual offsets. The recent new
2142 "Z" feature of pcretest means that these can be cut out, making the tests
2143 usable with all link sizes.
2145 5. Implemented Stan Switzer's goto replacement for longjmp() when not using
2146 stack recursion. This gives a massive performance boost under BSD, but just
2147 a small improvement under Linux. However, it saves one field in the frame
2148 in all cases.
2150 6. Added more features from the forthcoming Perl 5.10:
2152 (a) (?-n) (where n is a string of digits) is a relative subroutine or
2153 recursion call. It refers to the nth most recently opened parentheses.
2155 (b) (?+n) is also a relative subroutine call; it refers to the nth next
2156 to be opened parentheses.
2158 (c) Conditions that refer to capturing parentheses can be specified
2159 relatively, for example, (?(-2)... or (?(+3)...
2161 (d) \K resets the start of the current match so that everything before
2162 is not part of it.
2164 (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
2166 (f) \g{name} is another synonym - part of Perl 5.10's unification of
2167 reference syntax.
2169 (g) (?| introduces a group in which the numbering of parentheses in each
2170 alternative starts with the same number.
2172 (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
2174 7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
2177 8. A pattern such as (.*(.)?)* caused pcre_exec() to fail by either not
2178 terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
2179 for detecting groups that can match an empty string.
2181 9. A pattern with a very large number of alternatives (more than several
2182 hundred) was running out of internal workspace during the pre-compile
2183 phase, where pcre_compile() figures out how much memory will be needed. A
2184 bit of new cunning has reduced the workspace needed for groups with
2185 alternatives. The 1000-alternative test pattern now uses 12 bytes of
2186 workspace instead of running out of the 4096 that are available.
2188 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
2190 11. Applied patch from Google to remove an optimization that didn't quite work.
2191 The report of the bug said:
2193 pcrecpp::RE("a*").FullMatch("aaa") matches, while
2194 pcrecpp::RE("a*?").FullMatch("aaa") does not, and
2195 pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
2197 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
2198 it matched the wrong number of bytes.
2201 Version 7.1 24-Apr-07
2202 ---------------------
2204 1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
2205 that is more "standard", making use of automake and other Autotools. There
2206 is some re-arrangement of the files and adjustment of comments consequent
2207 on this.
2209 2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
2210 for recursive directory scanning broke on some systems because the files
2211 are not scanned in any specific order and on different systems the order
2212 was different. A call to "sort" has been inserted into RunGrepTest for the
2213 approprate test as a short-term fix. In the longer term there may be an
2214 alternative.
2216 3. I had an email from Eric Raymond about problems translating some of PCRE's
2217 man pages to HTML (despite the fact that I distribute HTML pages, some
2218 people do their own conversions for various reasons). The problems
2219 concerned the use of low-level troff macros .br and .in. I have therefore
2220 removed all such uses from the man pages (some were redundant, some could
2221 be replaced by .nf/.fi pairs). The 132html script that I use to generate
2222 HTML has been updated to handle .nf/.fi and to complain if it encounters
2223 .br or .in.
2225 4. Updated comments in configure.ac that get placed in config.h.in and also
2226 arranged for config.h to be included in the distribution, with the name
2227 config.h.generic, for the benefit of those who have to compile without
2228 Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
2230 5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
2231 Weber: (1) pcre_internal.h was missing some function renames; (2) updated
2232 makevp.bat for the current PCRE, using the additional files
2233 makevp_c.txt, makevp_l.txt, and pcregexp.pas.
2235 6. A Windows user reported a minor discrepancy with test 2, which turned out
2236 to be caused by a trailing space on an input line that had got lost in his
2237 copy. The trailing space was an accident, so I've just removed it.
2239 7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
2240 that is needed.
2242 8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
2243 as "const" (a) because they are and (b) because it helps the PHP
2244 maintainers who have recently made a script to detect big data structures
2245 in the php code that should be moved to the .rodata section. I remembered
2246 to update Builducptable as well, so it won't revert if ucptable.h is ever
2247 re-created.
2249 9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
2250 pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
2251 order to be able to cut out the UTF-8 tables in the latter when UTF-8
2252 support is not required. This saves 1.5-2K of code, which is important in
2253 some applications.
2255 Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
2256 so as not to refer to the tables, even though these functions will never be
2257 called when UTF-8 support is disabled. Otherwise there are problems with a
2258 shared library.
2260 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
2262 (a) It was defining its arguments as char * instead of void *.
2264 (b) It was assuming that all moves were upwards in memory; this was true
2265 a long time ago when I wrote it, but is no longer the case.
2267 The emulated memove() is provided for those environments that have neither
2268 memmove() nor bcopy(). I didn't think anyone used it these days, but that
2269 is clearly not the case, as these two bugs were recently reported.
2271 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
2272 and Detrail to create the HTML documentation, the .txt form of the man
2273 pages, and it removes trailing spaces from listed files. It also creates
2274 pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
2275 case, it wraps all the #defines with #ifndefs. This script should be run
2276 before "make dist".
2278 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
2279 with Unicode property support.
2281 (a) For a maximizing quantifier, if the two different cases of the
2282 character were of different lengths in their UTF-8 codings (there are
2283 some cases like this - I found 11), and the matching function had to
2284 back up over a mixture of the two cases, it incorrectly assumed they
2285 were both the same length.
2287 (b) When PCRE was configured to use the heap rather than the stack for
2288 recursion during matching, it was not correctly preserving the data for
2289 the other case of a UTF-8 character when checking ahead for a match
2290 while processing a minimizing repeat. If the check also involved
2291 matching a wide character, but failed, corruption could cause an
2292 erroneous result when trying to check for a repeat of the original
2293 character.
2295 13. Some tidying changes to the testing mechanism:
2297 (a) The RunTest script now detects the internal link size and whether there
2298 is UTF-8 and UCP support by running ./pcretest -C instead of relying on
2299 values substituted by "configure". (The RunGrepTest script already did
2300 this for UTF-8.) The configure.ac script no longer substitutes the
2301 relevant variables.
2303 (b) The debugging options /B and /D in pcretest show the compiled bytecode
2304 with length and offset values. This means that the output is different
2305 for different internal link sizes. Test 2 is skipped for link sizes
2306 other than 2 because of this, bypassing the problem. Unfortunately,
2307 there was also a test in test 3 (the locale tests) that used /B and
2308 failed for link sizes other than 2. Rather than cut the whole test out,
2309 I have added a new /Z option to pcretest that replaces the length and
2310 offset values with spaces. This is now used to make test 3 independent
2311 of link size. (Test 2 will be tidied up later.)
2313 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
2314 segmentation fault instead of returning the appropriate error message.
2316 15. In multiline mode when the newline sequence was set to "any", the pattern
2317 ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
2318 This doesn't seem right; it now treats the CRLF combination as the line
2319 ending, and so does not match in that case. It's only a pattern such as ^$
2320 that would hit this one: something like ^ABC$ would have failed after \r
2321 and then tried again after \r\n.
2323 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
2324 in an attempt to make files that differ only in their line terminators
2325 compare equal. This works on Linux.
2327 17. Under certain error circumstances pcregrep might try to free random memory
2328 as it exited. This is now fixed, thanks to valgrind.
2330 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
2331 "abc\r\n\r\n", it found an unwanted second match after the second \r. This
2332 was because its rules for how to advance for /g after matching an empty
2333 string at the end of a line did not allow for this case. They now check for
2334 it specially.
2336 20. pcretest is supposed to handle patterns and data of any length, by
2337 extending its buffers when necessary. It was getting this wrong when the
2338 buffer for a data line had to be extended.
2340 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
2341 CRLF as a newline sequence.
2343 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
2344 out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
2345 I have nevertheless tidied it up.
2347 23. Added some casts to kill warnings from HP-UX ia64 compiler.
2349 24. Added a man page for pcre-config.
2352 Version 7.0 19-Dec-06
2353 ---------------------
2355 1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
2356 moving to gcc 4.1.1.
2358 2. The -S option for pcretest uses setrlimit(); I had omitted to #include
2359 sys/time.h, which is documented as needed for this function. It doesn't
2360 seem to matter on Linux, but it showed up on some releases of OS X.
2362 3. It seems that there are systems where bytes whose values are greater than
2363 127 match isprint() in the "C" locale. The "C" locale should be the
2364 default when a C program starts up. In most systems, only ASCII printing
2365 characters match isprint(). This difference caused the output from pcretest
2366 to vary, making some of the tests fail. I have changed pcretest so that:
2368 (a) When it is outputting text in the compiled version of a pattern, bytes
2369 other than 32-126 are always shown as hex escapes.
2371 (b) When it is outputting text that is a matched part of a subject string,
2372 it does the same, unless a different locale has been set for the match
2373 (using the /L modifier). In this case, it uses isprint() to decide.
2375 4. Fixed a major bug that caused incorrect computation of the amount of memory
2376 required for a compiled pattern when options that changed within the
2377 pattern affected the logic of the preliminary scan that determines the
2378 length. The relevant options are -x, and -i in UTF-8 mode. The result was
2379 that the computed length was too small. The symptoms of this bug were
2380 either the PCRE error "internal error: code overflow" from pcre_compile(),
2381 or a glibc crash with a message such as "pcretest: free(): invalid next
2382 size (fast)". Examples of patterns that provoked this bug (shown in
2383 pcretest format) are:
2385 /(?-x: )/x
2386 /(?x)(?-x: \s*#\s*)/
2387 /((?i)[\x{c0}])/8
2388 /(?i:[\x{c0}])/8
2390 HOWEVER: Change 17 below makes this fix obsolete as the memory computation
2391 is now done differently.
2393 5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
2394 wrapper classes; (b) implement a new function in the C++ scanner that is
2395 more efficient than the old way of doing things because it avoids levels of
2396 recursion in the regex matching; (c) add a paragraph to the documentation
2397 for the FullMatch() function.
2399 6. The escape sequence \n was being treated as whatever was defined as
2400 "newline". Not only was this contrary to the documentation, which states
2401 that \n is character 10 (hex 0A), but it also went horribly wrong when
2402 "newline" was defined as CRLF. This has been fixed.
2404 7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
2405 was being set to -1 for the "end of line" case (supposedly a value that no
2406 character can have). Though this value is never used (the check for end of
2407 line is "zero bytes in current character"), it caused compiler complaints.
2408 I've changed it to 0xffffffff.
2410 8. In pcre_version.c, the version string was being built by a sequence of
2411 C macros that, in the event of PCRE_PRERELEASE being defined as an empty
2412 string (as it is for production releases) called a macro with an empty
2413 argument. The C standard says the result of this is undefined. The gcc
2414 compiler treats it as an empty string (which was what was wanted) but it is
2415 reported that Visual C gives an error. The source has been hacked around to
2416 avoid this problem.
2418 9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
2419 builds of pcretest, and changed the call to _setmode() to use _O_BINARY
2420 instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
2421 of them did).
2423 10. Originally, pcretest opened its input and output without "b"; then I was
2424 told that "b" was needed in some environments, so it was added for release
2425 5.0 to both the input and output. (It makes no difference on Unix-like
2426 systems.) Later I was told that it is wrong for the input on Windows. I've
2427 now abstracted the modes into two macros, to make it easier to fiddle with
2428 them, and removed "b" from the input mode under Windows.
2430 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
2432 12. Added -help and --help to pcretest as an official way of being reminded
2433 of the options.
2435 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
2436 and pcrecpp.cc because they annoy compilers at high warning levels.
2438 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
2440 15. Fixed an occurrence of == in configure.ac that should have been = (shell
2441 scripts are not C programs :-) and which was not noticed because it works
2442 on Linux.
2444 16. pcretest is supposed to handle any length of pattern and data line (as one
2445 line or as a continued sequence of lines) by extending its input buffer if
2446 necessary. This feature was broken for very long pattern lines, leading to
2447 a string of junk being passed to pcre_compile() if the pattern was longer
2448 than about 50K.
2450 17. I have done a major re-factoring of the way pcre_compile() computes the
2451 amount of memory needed for a compiled pattern. Previously, there was code
2452 that made a preliminary scan of the pattern in order to do this. That was
2453 OK when PCRE was new, but as the facilities have expanded, it has become
2454 harder and harder to keep it in step with the real compile phase, and there
2455 have been a number of bugs (see for example, 4 above). I have now found a
2456 cunning way of running the real compile function in a "fake" mode that
2457 enables it to compute how much memory it would need, while actually only
2458 ever using a few hundred bytes of working memory and without too many
2459 tests of the mode. This should make future maintenance and development
2460 easier. A side effect of this work is that the limit of 200 on the nesting
2461 depth of parentheses has been removed (though this was never a serious
2462 limitation, I suspect). However, there is a downside: pcre_compile() now
2463 runs more slowly than before (30% or more, depending on the pattern). I
2464 hope this isn't a big issue. There is no effect on runtime performance.
2466 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
2467 newline (only possible for the last line of a file) and it was a
2468 pattern that set a locale (followed by /Lsomething), pcretest crashed.
2470 19. Added additional timing features to pcretest. (1) The -tm option now times
2471 matching only, not compiling. (2) Both -t and -tm can be followed, as a
2472 separate command line item, by a number that specifies the number of
2473 repeats to use when timing. The default is 50000; this gives better
2474 precision, but takes uncomfortably long for very large patterns.
2476 20. Extended pcre_study() to be more clever in cases where a branch of a
2477 subpattern has no definite first character. For example, (a*|b*)[cd] would
2478 previously give no result from pcre_study(). Now it recognizes that the
2479 first character must be a, b, c, or d.
2481 21. There was an incorrect error "recursive call could loop indefinitely" if
2482 a subpattern (or the entire pattern) that was being tested for matching an
2483 empty string contained only one non-empty item after a nested subpattern.
2484 For example, the pattern (?>\x{100}*)\d(?R) provoked this error
2485 incorrectly, because the \d was being skipped in the check.
2487 22. The pcretest program now has a new pattern option /B and a command line
2488 option -b, which is equivalent to adding /B to every pattern. This causes
2489 it to show the compiled bytecode, without the additional information that
2490 -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
2491 is the same as /B/I).
2493 23. A new optimization is now able automatically to treat some sequences such
2494 as a*b as a*+b. More specifically, if something simple (such as a character
2495 or a simple class like \d) has an unlimited quantifier, and is followed by
2496 something that cannot possibly match the quantified thing, the quantifier
2497 is automatically "possessified".
2499 24. A recursive reference to a subpattern whose number was greater than 39
2500 went wrong under certain circumstances in UTF-8 mode. This bug could also
2501 have affected the operation of pcre_study().
2503 25. Realized that a little bit of performance could be had by replacing
2504 (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
2506 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
2508 27. Possessive quantifiers such as a++ were previously implemented by turning
2509 them into atomic groups such as ($>a+). Now they have their own opcodes,
2510 which improves performance. This includes the automatically created ones
2511 from 23 above.
2513 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
2514 lookahead was broken if it was not anchored. PCRE was mistakenly expecting
2515 the first matched character to be a colon. This applied both to named and
2516 numbered groups.
2518 29. The ucpinternal.h header file was missing its idempotency #ifdef.
2520 30. I was sent a "project" file called libpcre.a.dev which I understand makes
2521 building PCRE on Windows easier, so I have included it in the distribution.
2523 31. There is now a check in pcretest against a ridiculously large number being
2524 returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
2525 loop, the loop is abandoned.
2527 32. Forward references to subpatterns in conditions such as (?(2)...) where
2528 subpattern 2 is defined later cause pcre_compile() to search forwards in
2529 the pattern for the relevant set of parentheses. This search went wrong
2530 when there were unescaped parentheses in a character class, parentheses
2531 escaped with \Q...\E, or parentheses in a #-comment in /x mode.
2533 33. "Subroutine" calls and backreferences were previously restricted to
2534 referencing subpatterns earlier in the regex. This restriction has now
2535 been removed.
2537 34. Added a number of extra features that are going to be in Perl 5.10. On the
2538 whole, these are just syntactic alternatives for features that PCRE had
2539 previously implemented using the Python syntax or my own invention. The
2540 other formats are all retained for compatibility.
2542 (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
2543 as (?P<name>...). The new forms, as well as being in Perl 5.10, are
2544 also .NET compatible.
2546 (b) A recursion or subroutine call to a named group can now be defined as
2547 (?&name) as well as (?P>name).
2549 (c) A backreference to a named group can now be defined as \k<name> or
2550 \k'name' as well as (?P=name). The new forms, as well as being in Perl
2551 5.10, are also .NET compatible.
2553 (d) A conditional reference to a named group can now use the syntax
2554 (?(<name>) or (?('name') as well as (?(name).
2556 (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
2557 groups (named and numbered) that are never evaluated inline, but can be
2558 called as "subroutines" from elsewhere. In effect, the DEFINE condition
2559 is always false. There may be only one alternative in such a group.
2561 (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
2562 as the simple (?(R). The condition is true only if the most recent
2563 recursion is that of the given number or name. It does not search out
2564 through the entire recursion stack.
2566 (g) The escape \gN or \g{N} has been added, where N is a positive or
2567 negative number, specifying an absolute or relative reference.
2569 35. Tidied to get rid of some further signed/unsigned compiler warnings and
2570 some "unreachable code" warnings.
2572 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
2573 things, this adds five new scripts.
2575 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
2576 There were also incompatibilities regarding the handling of \Q..\E inside
2577 character classes, for example with patterns like [\Qa\E-\Qz\E] where the
2578 hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
2580 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
2581 matches an empty string, and forcibly breaks the loop. There were bugs in
2582 this code in non-simple cases. For a pattern such as ^(a()*)* matched
2583 against aaaa the result was just "a" rather than "aaaa", for example. Two
2584 separate and independent bugs (that affected different cases) have been
2585 fixed.
2587 39. Refactored the code to abolish the use of different opcodes for small
2588 capturing bracket numbers. This is a tidy that I avoided doing when I
2589 removed the limit on the number of capturing brackets for 3.5 back in 2001.
2590 The new approach is not only tidier, it makes it possible to reduce the
2591 memory needed to fix the previous bug (38).
2593 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
2594 sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
2595 processing dot, circumflex, or dollar metacharacters, or #-comments in /x
2596 mode.
2598 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
2599 report.
2601 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
2602 copy construction and assignment in the C++ wrapper.
2604 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
2605 couple of bugs that could have given wrong results in the "--newline=crlf"
2606 case.
2608 44. Added a number of casts and did some reorganization of signed/unsigned int
2609 variables following suggestions from Dair Grant. Also renamed the variable
2610 "this" as "item" because it is a C++ keyword.
2612 45. Arranged for dftables to add
2614 #include "pcre_internal.h"
2616 to pcre_chartables.c because without it, gcc 4.x may remove the array
2617 definition from the final binary if PCRE is built into a static library and
2618 dead code stripping is activated.
2620 46. For an unanchored pattern, if a match attempt fails at the start of a
2621 newline sequence, and the newline setting is CRLF or ANY, and the next two
2622 characters are CRLF, advance by two characters instead of one.
2625 Version 6.7 04-Jul-06
2626 ---------------------
2628 1. In order to handle tests when input lines are enormously long, pcretest has
2629 been re-factored so that it automatically extends its buffers when
2630 necessary. The code is crude, but this _is_ just a test program. The
2631 default size has been increased from 32K to 50K.
2633 2. The code in pcre_study() was using the value of the re argument before
2634 testing it for NULL. (Of course, in any sensible call of the function, it
2635 won't be NULL.)
2637 3. The memmove() emulation function in pcre_internal.h, which is used on
2638 systems that lack both memmove() and bcopy() - that is, hardly ever -
2639 was missing a "static" storage class specifier.
2641 4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
2642 containing an extended class (one that cannot be represented by a bitmap
2643 because it contains high-valued characters or Unicode property items, e.g.
2644 [\pZ]). Almost always one would set UTF-8 mode when processing such a
2645 pattern, but PCRE should not loop if you do not (it no longer does).
2646 [Detail: two cases were found: (a) a repeated subpattern containing an
2647 extended class; (b) a recursive reference to a subpattern that followed a
2648 previous extended class. It wasn't skipping over the extended class
2649 correctly when UTF-8 mode was not set.]
2651 5. A negated single-character class was not being recognized as fixed-length
2652 in lookbehind assertions such as (?<=[^f]), leading to an incorrect
2653 compile error "lookbehind assertion is not fixed length".
2655 6. The RunPerlTest auxiliary script was showing an unexpected difference
2656 between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
2657 write a Perl script that can interpret lines of an input file either as
2658 byte characters or as UTF-8, which is what "perltest" was being required to
2659 do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
2660 can't do is switch easily at run time between having the "use utf8;" pragma
2661 or not. In the end, I fudged it by using the RunPerlTest script to insert
2662 "use utf8;" explicitly for the UTF-8 tests.
2664 7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
2665 the end of the subject string, contrary to the documentation and to what
2666 Perl does. This was true of both matching functions. Now it matches only at
2667 the start of the subject and immediately after *internal* newlines.
2669 8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
2670 a pointer to an int instead of a pointer to an unsigned long int. This
2671 caused problems on 64-bit systems.
2673 9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
2674 instance of the 'standard' template library not being so standard".
2676 10. There was no check on the number of named subpatterns nor the maximum
2677 length of a subpattern name. The product of these values is used to compute
2678 the size of the memory block for a compiled pattern. By supplying a very
2679 long subpattern name and a large number of named subpatterns, the size
2680 computation could be caused to overflow. This is now prevented by limiting
2681 the length of names to 32 characters, and the number of named subpatterns
2682 to 10,000.
2684 11. Subpatterns that are repeated with specific counts have to be replicated in
2685 the compiled pattern. The size of memory for this was computed from the
2686 length of the subpattern and the repeat count. The latter is limited to
2687 65535, but there was no limit on the former, meaning that integer overflow
2688 could in principle occur. The compiled length of a repeated subpattern is
2689 now limited to 30,000 bytes in order to prevent this.
2691 12. Added the optional facility to have named substrings with the same name.
2693 13. Added the ability to use a named substring as a condition, using the
2694 Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
2695 are numbers (not recommended). Forward references are permitted.
2697 14. Added forward references in named backreferences (if you see what I mean).
2699 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
2700 pattern could run off the end of the subject. For example, the pattern
2701 "(?s)(.{1,5})"8 did this with the subject "ab".
2703 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
2704 PCRE_CASELESS was set when matching characters that were quantified with ?
2705 or *.
2707 17. A character class other than a single negated character that had a minimum
2708 but no maximum quantifier - for example [ab]{6,} - was not handled
2709 correctly by pce_dfa_exec(). It would match only one character.
2711 18. A valid (though odd) pattern that looked like a POSIX character
2712 class but used an invalid character after [ (for example [[,abc,]]) caused
2713 pcre_compile() to give the error "Failed: internal error: code overflow" or
2714 in some cases to crash with a glibc free() error. This could even happen if
2715 the pattern terminated after [[ but there just happened to be a sequence of
2716 letters, a binary zero, and a closing ] in the memory that followed.
2718 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
2719 over the years. Originally (before any Unicode support), just the bottom 8
2720 bits were taken. Thus, for example, \500 really meant \100. Nowadays the
2721 output from "man perlunicode" includes this:
2723 The regular expression compiler produces polymorphic opcodes. That
2724 is, the pattern adapts to the data and automatically switches to
2725 the Unicode character scheme when presented with Unicode data--or
2726 instead uses a traditional byte scheme when presented with byte
2727 data.
2729 Sadly, a wide octal escape does not cause a switch, and in a string with
2730 no other multibyte characters, these octal escapes are treated as before.
2731 Thus, in Perl, the pattern /\500/ actually matches \100 but the pattern
2732 /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
2733 Unicode string.
2735 I have not perpetrated such confusion in PCRE. Up till now, it took just
2736 the bottom 8 bits, as in old Perl. I have now made octal escapes with
2737 values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
2738 translate to the appropriate multibyte character.
2740 29. Applied some refactoring to reduce the number of warnings from Microsoft
2741 and Borland compilers. This has included removing the fudge introduced
2742 seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
2743 a warning about an unused variable.
2745 21. PCRE has not included VT (character 0x0b) in the set of whitespace
2746 characters since release 4.0, because Perl (from release 5.004) does not.
2747 [Or at least, is documented not to: some releases seem to be in conflict
2748 with the documentation.] However, when a pattern was studied with
2749 pcre_study() and all its branches started with \s, PCRE still included VT
2750 as a possible starting character. Of course, this did no harm; it just
2751 caused an unnecessary match attempt.
2753 22. Removed a now-redundant internal flag bit that recorded the fact that case
2754 dependency changed within the pattern. This was once needed for "required
2755 byte" processing, but is no longer used. This recovers a now-scarce options
2756 bit. Also moved the least significant internal flag bit to the most-
2757 significant bit of the word, which was not previously used (hangover from
2758 the days when it was an int rather than a uint) to free up another bit for
2759 the future.
2761 23. Added support for CRLF line endings as well as CR and LF. As well as the
2762 default being selectable at build time, it can now be changed at runtime
2763 via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
2764 specify that it is scanning data with non-default line endings.
2766 24. Changed the definition of CXXLINK to make it agree with the definition of
2767 LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
2769 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
2770 recursions. This makes a big different to stack usage for some patterns.
2772 26. If a subpattern containing a named recursion or subroutine reference such
2773 as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
2774 the space required for the compiled pattern went wrong and gave too small a
2775 value. Depending on the environment, this could lead to "Failed: internal
2776 error: code overflow at offset 49" or "glibc detected double free or
2777 corruption" errors.
2779 27. Applied patches from Google (a) to support the new newline modes and (b) to
2780 advance over multibyte UTF-8 characters in GlobalReplace.
2782 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
2783 difference for some implementation of PCRE in some Windows version.
2785 29. Added some extra testing facilities to pcretest:
2787 \q<number> in a data line sets the "match limit" value
2788 \Q<number> in a data line sets the "match recursion limt" value
2789 -S <number> sets the stack size, where <number> is in megabytes
2791 The -S option isn't available for Windows.
2794 Version 6.6 06-Feb-06
2795 ---------------------
2797 1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
2798 in pcreposix.h. I have copied the definition from pcre.h.
2800 2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
2801 because pcre.h is no longer a built file.
2803 3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
2804 not normally included in the compiled code.
2807 Version 6.5 01-Feb-06
2808 ---------------------
2810 1. When using the partial match feature with pcre_dfa_exec(), it was not
2811 anchoring the second and subsequent partial matches at the new starting
2812 point. This could lead to incorrect results. For example, with the pattern
2813 /1234/, partially matching against "123" and then "a4" gave a match.
2815 2. Changes to pcregrep:
2817 (a) All non-match returns from pcre_exec() were being treated as failures
2818 to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
2819 error message is output. Some extra information is given for the
2821 probably the only errors that are likely to be caused by users (by
2822 specifying a regex that has nested indefinite repeats, for instance).
2823 If there are more than 20 of these errors, pcregrep is abandoned.
2825 (b) A binary zero was treated as data while matching, but terminated the
2826 output line if it was written out. This has been fixed: binary zeroes
2827 are now no different to any other data bytes.
2829 (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
2830 used to set a locale for matching. The --locale=xxxx long option has
2831 been added (no short equivalent) to specify a locale explicitly on the
2832 pcregrep command, overriding the environment variables.
2834 (d) When -B was used with -n, some line numbers in the output were one less
2835 than they should have been.
2837 (e) Added the -o (--only-matching) option.
2839 (f) If -A or -C was used with -c (count only), some lines of context were
2840 accidentally printed for the final match.
2842 (g) Added the -H (--with-filename) option.
2844 (h) The combination of options -rh failed to suppress file names for files
2845 that were found from directory arguments.
2847 (i) Added the -D (--devices) and -d (--directories) options.
2849 (j) Added the -F (--fixed-strings) option.
2851 (k) Allow "-" to be used as a file name for -f as well as for a data file.
2853 (l) Added the --colo(u)r option.
2855 (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
2856 is not present by default.
2858 3. A nasty bug was discovered in the handling of recursive patterns, that is,
2859 items such as (?R) or (?1), when the recursion could match a number of
2860 alternatives. If it matched one of the alternatives, but subsequently,
2861 outside the recursion, there was a failure, the code tried to back up into
2862 the recursion. However, because of the way PCRE is implemented, this is not
2863 possible, and the result was an incorrect result from the match.
2865 In order to prevent this happening, the specification of recursion has
2866 been changed so that all such subpatterns are automatically treated as
2867 atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
2869 4. I had overlooked the fact that, in some locales, there are characters for
2870 which isalpha() is true but neither isupper() nor islower() are true. In
2871 the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
2872 and ordfeminine) are like this. This affected the treatment of \w and \W
2873 when they appeared in character classes, but not when they appeared outside
2874 a character class. The bit map for "word" characters is now created
2875 separately from the results of isalnum() instead of just taking it from the
2876 upper, lower, and digit maps. (Plus the underscore character, of course.)
2878 5. The above bug also affected the handling of POSIX character classes such as
2879 [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
2880 permanent tables. Instead, the bit maps for such a class were previously
2881 created as the appropriate unions of the upper, lower, and digit bitmaps.
2882 Now they are created by subtraction from the [[:word:]] class, which has
2883 its own bitmap.
2885 6. The [[:blank:]] character class matches horizontal, but not vertical space.
2886 It is created by subtracting the vertical space characters (\x09, \x0a,
2887 \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
2888 subtraction was done in the overall bitmap for a character class, meaning
2889 that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
2890 be recognized. This bug has been fixed.
2892 7. Patches from the folks at Google:
2894 (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
2895 real life, but is still worth protecting against".
2897 (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
2898 regular expressions".
2900 (c) pcre_scanner.cc: avoid use of std::count() because not all systems
2901 have it.
2903 (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
2904 "configure" and the latter not, in order to fix a problem somebody had
2905 with compiling the Arg class on HP-UX.
2907 (e) Improve the error-handling of the C++ wrapper a little bit.
2909 (f) New tests for checking recursion limiting.
2911 8. The pcre_memmove() function, which is used only if the environment does not
2912 have a standard memmove() function (and is therefore rarely compiled),
2913 contained two bugs: (a) use of int instead of size_t, and (b) it was not
2914 returning a result (though PCRE never actually uses the result).
2916 9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
2917 large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
2918 returned instead of calling malloc() with an overflowing number that would
2919 most likely cause subsequent chaos.
2921 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
2923 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
2924 with this option is matched, the nmatch and pmatch options of regexec() are
2925 ignored.
2927 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
2928 provided in case anyone wants to the the POSIX interface with UTF-8
2929 strings.
2931 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
2932 C++ linking (needed for some HP-UX environments).
2934 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
2935 (unused parameter) and in the pcre_printint() function (omitted "default"
2936 switch label when the default is to do nothing).
2938 15. Added some code to make it possible, when PCRE is compiled as a C++
2939 library, to replace subject pointers for pcre_exec() with a smart pointer
2940 class, thus making it possible to process discontinuous strings.
2942 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
2943 much the same function. They were added by different people who were trying
2944 to make PCRE easy to compile on non-Unix systems. It has been suggested
2945 that PCRE_EXPORT be abolished now that there is more automatic apparatus
2946 for compiling on Windows systems. I have therefore replaced it with
2947 PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
2948 defaults to "extern" for C or "extern C" for C++, which works fine on
2949 Unix-like systems. It is now possible to override the value of PCRE_DATA_
2950 SCOPE with something explicit in config.h. In addition:
2952 (a) pcreposix.h still had just "extern" instead of either of these macros;
2953 I have replaced it with PCRE_DATA_SCOPE.
2955 (b) Functions such as _pcre_xclass(), which are internal to the library,
2956 but external in the C sense, all had PCRE_EXPORT in their definitions.
2957 This is apparently wrong for the Windows case, so I have removed it.
2958 (It makes no difference on Unix-like systems.)
2960 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
2961 of recursive calls to match(). This is different to MATCH_LIMIT because
2962 that limits the total number of calls to match(), not all of which increase
2963 the depth of recursion. Limiting the recursion depth limits the amount of
2964 stack (or heap if NO_RECURSE is set) that is used. The default can be set
2965 when PCRE is compiled, and changed at run time. A patch from Google adds
2966 this functionality to the C++ interface.
2968 18. Changes to the handling of Unicode character properties:
2970 (a) Updated the table to Unicode 4.1.0.
2972 (b) Recognize characters that are not in the table as "Cn" (undefined).
2974 (c) I revised the way the table is implemented to a much improved format
2975 which includes recognition of ranges. It now supports the ranges that
2976 are defined in UnicodeData.txt, and it also amalgamates other
2977 characters into ranges. This has reduced the number of entries in the
2978 table from around 16,000 to around 3,000, thus reducing its size
2979 considerably. I realized I did not need to use a tree structure after
2980 all - a binary chop search is just as efficient. Having reduced the
2981 number of entries, I extended their size from 6 bytes to 8 bytes to
2982 allow for more data.
2984 (d) Added support for Unicode script names via properties such as \p{Han}.
2986 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
2987 matching that character.
2989 20. When matching a repeated Unicode property with a minimum greater than zero,
2990 (for example \pL{2,}), PCRE could look past the end of the subject if it
2991 reached it while seeking the minimum number of characters. This could
2992 happen only if some of the characters were more than one byte long, because
2993 there is a check for at least the minimum number of bytes.
2995 21. Refactored the implementation of \p and \P so as to be more general, to
2996 allow for more different types of property in future. This has changed the
2997 compiled form incompatibly. Anybody with saved compiled patterns that use
2998 \p or \P will have to recompile them.
3000 22. Added "Any" and "L&" to the supported property types.
3002 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
3003 but give a compile time error if the value is greater than 0xff.
3005 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
3006 accidentally not being installed or uninstalled.
3008 25. The pcre.h file was built from pcre.h.in, but the only changes that were
3009 made were to insert the current release number. This seemed silly, because
3010 it made things harder for people building PCRE on systems that don't run
3011 "configure". I have turned pcre.h into a distributed file, no longer built
3012 by "configure", with the version identification directly included. There is
3013 no longer a pcre.h.in file.
3015 However, this change necessitated a change to the pcre-config script as
3016 well. It is built from pcre-config.in, and one of the substitutions was the
3017 release number. I have updated configure.ac so that ./configure now finds
3018 the release number by grepping pcre.h.
3020 26. Added the ability to run the tests under valgrind.
3023 Version 6.4 05-Sep-05
3024 ---------------------
3026 1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
3027 "--" to be printed when multiple files were scanned, even when none of the
3028 -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
3029 consider it to be a bug, and have restored the previous behaviour.
3031 2. A couple of code tidies to get rid of compiler warnings.
3033 3. The pcretest program used to cheat by referring to symbols in the library
3034 whose names begin with _pcre_. These are internal symbols that are not
3035 really supposed to be visible externally, and in some environments it is
3036 possible to suppress them. The cheating is now confined to including
3037 certain files from the library's source, which is a bit cleaner.
3039 4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
3040 file's purpose clearer.
3042 5. Reorganized pcre_ucp_findchar().
3045 Version 6.3 15-Aug-05
3046 ---------------------
3048 1. The file libpcre.pc.in did not have general read permission in the tarball.
3050 2. There were some problems when building without C++ support:
3052 (a) If C++ support was not built, "make install" and "make test" still
3053 tried to test it.
3055 (b) There were problems when the value of CXX was explicitly set. Some
3056 changes have been made to try to fix these, and ...
3058 (c) --disable-cpp can now be used to explicitly disable C++ support.
3060 (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
3061 backslash in a target when C++ was disabled. This confuses some
3062 versions of "make", apparently. Using an intermediate variable solves
3063 this. (Same for CPP_LOBJ.)
3065 3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
3066 (non-Windows) now includes $(CFLAGS) because these flags are sometimes
3067 necessary on certain architectures.
3069 4. Added a setting of -export-symbols-regex to the link command to remove
3070 those symbols that are exported in the C sense, but actually are local
3071 within the library, and not documented. Their names all begin with
3072 "_pcre_". This is not a perfect job, because (a) we have to except some
3073 symbols that pcretest ("illegally") uses, and (b) the facility isn't always
3074 available (and never for static libraries). I have made a note to try to
3075 find a way round (a) in the future.
3078 Version 6.2 01-Aug-05
3079 ---------------------
3081 1. There was no test for integer overflow of quantifier values. A construction
3082 such as {1111111111111111} would give undefined results. What is worse, if
3083 a minimum quantifier for a parenthesized subpattern overflowed and became
3084 negative, the calculation of the memory size went wrong. This could have
3085 led to memory overwriting.
3087 2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
3089 3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
3090 operating environments where this matters.
3092 4. Applied Giuseppe Maxia's patch to add additional features for controlling
3093 PCRE options from within the C++ wrapper.
3095 5. Named capturing subpatterns were not being correctly counted when a pattern
3096 was compiled. This caused two problems: (a) If there were more than 100
3097 such subpatterns, the calculation of the memory needed for the whole
3098 compiled pattern went wrong, leading to an overflow error. (b) Numerical
3099 back references of the form \12, where the number was greater than 9, were
3100 not recognized as back references, even though there were sufficient
3101 previous subpatterns.
3103 6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
3104 versions of gcc, e.g. 2.95.4.
3107 Version 6.1 21-Jun-05
3108 ---------------------
3110 1. There was one reference to the variable "posix" in pcretest.c that was not
3111 surrounded by "#if !defined NOPOSIX".
3113 2. Make it possible to compile pcretest without DFA support, UTF8 support, or
3114 the cross-check on the old pcre_info() function, for the benefit of the
3115 cut-down version of PCRE that is currently imported into Exim.
3117 3. A (silly) pattern starting with (?i)(?-i) caused an internal space
3118 allocation error. I've done the easy fix, which wastes 2 bytes for sensible
3119 patterns that start (?i) but I don't think that matters. The use of (?i) is
3120 just an example; this all applies to the other options as well.
3122 4. Since libtool seems to echo the compile commands it is issuing, the output
3123 from "make" can be reduced a bit by putting "@" in front of each libtool
3124 compile command.
3126 5. Patch from the folks at Google for configure.in to be a bit more thorough
3127 in checking for a suitable C++ installation before trying to compile the
3128 C++ stuff. This should fix a reported problem when a compiler was present,
3129 but no suitable headers.
3131 6. The man pages all had just "PCRE" as their title. I have changed them to
3132 be the relevant file name. I have also arranged that these names are
3133 retained in the file doc/pcre.txt, which is a concatenation in text format
3134 of all the man pages except the little individual ones for each function.
3136 7. The NON-UNIX-USE file had not been updated for the different set of source
3137 files that come with release 6. I also added a few comments about the C++
3138 wrapper.
3141 Version 6.0 07-Jun-05
3142 ---------------------
3144 1. Some minor internal re-organization to help with my DFA experiments.
3146 2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
3147 didn't matter for the library itself when fully configured, but did matter
3148 when compiling without UCP support, or within Exim, where the ucp files are
3149 not imported.
3151 3. Refactoring of the library code to split up the various functions into
3152 different source modules. The addition of the new DFA matching code (see
3153 below) to a single monolithic source would have made it really too
3154 unwieldy, quite apart from causing all the code to be include in a
3155 statically linked application, when only some functions are used. This is
3156 relevant even without the DFA addition now that patterns can be compiled in
3157 one application and matched in another.
3159 The downside of splitting up is that there have to be some external
3160 functions and data tables that are used internally in different modules of
3161 the library but which are not part of the API. These have all had their
3162 names changed to start with "_pcre_" so that they are unlikely to clash
3163 with other external names.
3165 4. Added an alternate matching function, pcre_dfa_exec(), which matches using
3166 a different (DFA) algorithm. Although it is slower than the original
3167 function, it does have some advantages for certain types of matching
3168 problem.
3170 5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
3171 including restarting after a partial match.
3173 6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
3174 defined when compiling for Windows was sent to me. I have put it into the
3175 code, though I have no means of testing or verifying it.
3177 7. Added the pcre_refcount() auxiliary function.
3179 8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
3180 match before or at the first newline in the subject string. In pcretest,
3181 the /f option on a pattern can be used to set this.
3183 9. A repeated \w when used in UTF-8 mode with characters greater than 256
3184 would behave wrongly. This has been present in PCRE since release 4.0.
3186 10. A number of changes to the pcregrep command:
3188 (a) Refactored how -x works; insert ^(...)$ instead of setting
3189 PCRE_ANCHORED and checking the length, in preparation for adding
3190 something similar for -w.
3192 (b) Added the -w (match as a word) option.
3194 (c) Refactored the way lines are read and buffered so as to have more
3195 than one at a time available.
3197 (d) Implemented a pcregrep test script.
3199 (e) Added the -M (multiline match) option. This allows patterns to match
3200 over several lines of the subject. The buffering ensures that at least
3201 8K, or the rest of the document (whichever is the shorter) is available
3202 for matching (and similarly the previous 8K for lookbehind assertions).
3204 (f) Changed the --help output so that it now says
3206 -w, --word-regex(p)
3208 instead of two lines, one with "regex" and the other with "regexp"
3209 because that confused at least one person since the short forms are the
3210 same. (This required a bit of code, as the output is generated
3211 automatically from a table. It wasn't just a text change.)
3213 (g) -- can be used to terminate pcregrep options if the next thing isn't an
3214 option but starts with a hyphen. Could be a pattern or a path name
3215 starting with a hyphen, for instance.
3217 (h) "-" can be given as a file name to represent stdin.
3219 (i) When file names are being printed, "(standard input)" is used for
3220 the standard input, for compatibility with GNU grep. Previously
3221 "<stdin>" was used.
3223 (j) The option --label=xxx can be used to supply a name to be used for
3224 stdin when file names are being printed. There is no short form.
3226 (k) Re-factored the options decoding logic because we are going to add
3227 two more options that take data. Such options can now be given in four
3228 different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
3230 (l) Added the -A, -B, and -C options for requesting that lines of context
3231 around matches be printed.
3233 (m) Added the -L option to print the names of files that do not contain
3234 any matching lines, that is, the complement of -l.
3236 (n) The return code is 2 if any file cannot be opened, but pcregrep does
3237 continue to scan other files.
3239 (o) The -s option was incorrectly implemented. For compatibility with other
3240 greps, it now suppresses the error message for a non-existent or non-
3241 accessible file (but not the return code). There is a new option called
3242 -q that suppresses the output of matching lines, which was what -s was
3243 previously doing.
3245 (p) Added --include and --exclude options to specify files for inclusion
3246 and exclusion when recursing.
3248 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
3249 Hopefully, it now does.
3251 12. Missing cast in pcre_study().
3253 13. Added an "uninstall" target to the makefile.
3255 14. Replaced "extern" in the function prototypes in Makefile.in with
3256 "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
3257 world, but is set differently for Windows.
3259 15. Added a second compiling function called pcre_compile2(). The only
3260 difference is that it has an extra argument, which is a pointer to an
3261 integer error code. When there is a compile-time failure, this is set
3262 non-zero, in addition to the error test pointer being set to point to an
3263 error message. The new argument may be NULL if no error number is required
3264 (but then you may as well call pcre_compile(), which is now just a
3265 wrapper). This facility is provided because some applications need a
3266 numeric error indication, but it has also enabled me to tidy up the way
3267 compile-time errors are handled in the POSIX wrapper.
3269 16. Added VPATH=.libs to the makefile; this should help when building with one
3270 prefix path and installing with another. (Or so I'm told by someone who
3271 knows more about this stuff than I do.)
3273 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
3274 passes PCRE_DOTALL to the pcre_compile() function, making the "." character
3275 match everything, including newlines. This is not POSIX-compatible, but
3276 somebody wanted the feature. From pcretest it can be activated by using
3277 both the P and the s flags.
3279 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
3281 19. libpcre.pc was being incorrectly installed as executable.
3283 20. A couple of places in pcretest check for end-of-line by looking for '\n';
3284 it now also looks for '\r' so that it will work unmodified on Windows.
3286 21. Added Google's contributed C++ wrapper to the distribution.
3288 22. Added some untidy missing memory free() calls in pcretest, to keep
3289 Electric Fence happy when testing.
3293 Version 5.0 13-Sep-04
3294 ---------------------
3296 1. Internal change: literal characters are no longer packed up into items
3297 containing multiple characters in a single byte-string. Each character
3298 is now matched using a separate opcode. However, there may be more than one
3299 byte in the character in UTF-8 mode.
3301 2. The pcre_callout_block structure has two new fields: pattern_position and
3302 next_item_length. These contain the offset in the pattern to the next match
3303 item, and its length, respectively.
3305 3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
3306 insertion of callouts before each pattern item. Added the /C option to
3307 pcretest to make use of this.
3309 4. On the advice of a Windows user, the lines
3311 #if defined(_WIN32) || defined(WIN32)
3312 _setmode( _fileno( stdout ), 0x8000 );
3313 #endif /* defined(_WIN32) || defined(WIN32) */
3315 have been added to the source of pcretest. This apparently does useful
3316 magic in relation to line terminators.
3318 5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
3319 for the benefit of those environments where the "b" makes a difference.
3321 6. The icc compiler has the same options as gcc, but "configure" doesn't seem
3322 to know about it. I have put a hack into configure.in that adds in code
3323 to set GCC=yes if CC=icc. This seems to end up at a point in the
3324 generated configure script that is early enough to affect the setting of
3325 compiler options, which is what is needed, but I have no means of testing
3326 whether it really works. (The user who reported this had patched the
3327 generated configure script, which of course I cannot do.)
3329 LATER: After change 22 below (new libtool files), the configure script
3330 seems to know about icc (and also ecc). Therefore, I have commented out
3331 this hack in configure.in.
3333 7. Added support for pkg-config (2 patches were sent in).
3335 8. Negated POSIX character classes that used a combination of internal tables
3336 were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
3337 [[:^ascii]]. Typically, they would match almost any characters. The other
3338 POSIX classes were not broken in this way.
3340 9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
3341 to find the match, as PCRE was deluded into thinking that the match had to
3342 start at the start point or following a newline. The same bug applied to
3343 patterns with negative forward assertions or any backward assertions
3344 preceding ".*" at the start, unless the pattern required a fixed first
3345 character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
3347 10. In UTF-8 mode, when moving forwards in the subject after a failed match
3348 starting at the last subject character, bytes beyond the end of the subject
3349 string were read.
3351 11. Renamed the variable "class" as "classbits" to make life easier for C++
3352 users. (Previously there was a macro definition, but it apparently wasn't
3353 enough.)
3355 12. Added the new field "tables" to the extra data so that tables can be passed
3356 in at exec time, or the internal tables can be re-selected. This allows
3357 a compiled regex to be saved and re-used at a later time by a different
3358 program that might have everything at different addresses.
3360 13. Modified the pcre-config script so that, when run on Solaris, it shows a
3361 -R library as well as a -L library.
3363 14. The debugging options of pcretest (-d on the command line or D on a
3364 pattern) showed incorrect output for anything following an extended class
3365 that contained multibyte characters and which was followed by a quantifier.
3367 15. Added optional support for general category Unicode character properties
3368 via the \p, \P, and \X escapes. Unicode property support implies UTF-8
3369 support. It adds about 90K to the size of the library. The meanings of the
3370 inbuilt class escapes such as \d and \s have NOT been changed.
3372 16. Updated pcredemo.c to include calls to free() to release the memory for the
3373 compiled pattern.
3375 17. The generated file chartables.c was being created in the source directory
3376 instead of in the building directory. This caused the build to fail if the
3377 source directory was different from the building directory, and was
3378 read-only.
3380 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
3381 file. No doubt somebody will tell me if they don't make sense... Also added
3382 Dan Mooney's comments about building on OpenVMS.
3384 19. Added support for partial matching via the PCRE_PARTIAL option for
3385 pcre_exec() and the \P data escape in pcretest.
3387 20. Extended pcretest with 3 new pattern features:
3389 (i) A pattern option of the form ">rest-of-line" causes pcretest to
3390 write the compiled pattern to the file whose name is "rest-of-line".
3391 This is a straight binary dump of the data, with the saved pointer to
3392 the character tables forced to be NULL. The study data, if any, is
3393 written too. After writing, pcretest reads a new pattern.
3395 (ii) If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
3396 compiled pattern from the given file. There must not be any
3397 occurrences of "<" in the file name (pretty unlikely); if there are,
3398 pcretest will instead treat the initial "<" as a pattern delimiter.
3399 After reading in the pattern, pcretest goes on to read data lines as
3400 usual.
3402 (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
3403 and 16-bit fields in a compiled pattern, to simulate a pattern that
3404 was compiled on a host of opposite endianness.
3406 21. The pcre-exec() function can now cope with patterns that were compiled on
3407 hosts of opposite endianness, with this restriction:
3409 As for any compiled expression that is saved and used later, the tables
3410 pointer field cannot be preserved; the extra_data field in the arguments
3411 to pcre_exec() should be used to pass in a tables address if a value
3412 other than the default internal tables were used at compile time.
3414 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
3415 now diagnosed as an error. Previously, most of the time, a negative number
3416 would have been treated as zero, but if in addition "ovector" was passed as
3417 NULL, a crash could occur.
3419 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
3420 new versions from the libtool 1.5 distribution (the last one is a copy of
3421 a file called libtool.m4). This seems to have fixed the need to patch
3422 "configure" to support Darwin 1.3 (which I used to do). However, I still
3423 had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
3424 workstation).
3426 24. Changed the PCRE licence to be the more standard "BSD" licence.
3429 Version 4.5 01-Dec-03
3430 ---------------------
3432 1. There has been some re-arrangement of the code for the match() function so
3433 that it can be compiled in a version that does not call itself recursively.
3434 Instead, it keeps those local variables that need separate instances for
3435 each "recursion" in a frame on the heap, and gets/frees frames whenever it
3436 needs to "recurse". Keeping track of where control must go is done by means
3437 of setjmp/longjmp. The whole thing is implemented by a set of macros that
3438 hide most of the details from the main code, and operates only if
3439 NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
3440 "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
3441 operating.
3443 To make it easier for callers to provide specially tailored get/free
3444 functions for this usage, two new functions, pcre_stack_malloc, and
3445 pcre_stack_free, are used. They are always called in strict stacking order,
3446 and the size of block requested is always the same.
3448 The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
3449 PCRE has been compiled to use the stack or the heap for recursion. The
3450 -C option of pcretest uses this to show which version is compiled.
3452 A new data escape \S, is added to pcretest; it causes the amounts of store
3453 obtained and freed by both kinds of malloc/free at match time to be added
3454 to the output.
3456 2. Changed the locale test to use "fr_FR" instead of "fr" because that's
3457 what's available on my current Linux desktop machine.
3459 3. When matching a UTF-8 string, the test for a valid string at the start has
3460 been extended. If start_offset is not zero, PCRE now checks that it points
3461 to a byte that is the start of a UTF-8 character. If not, it returns
3462 PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
3463 this is necessary because there may be backward assertions in the pattern.
3464 When matching the same subject several times, it may save resources to use
3465 PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
3467 4. The code for checking the validity of UTF-8 strings has been tightened so
3468 that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
3469 containing "overlong sequences".
3471 5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
3472 I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
3473 should have been "&", but it just so happened that all the cases this let
3474 through by mistake were picked up later in the function.
3476 6. I had used a variable called "isblank" - this is a C99 function, causing
3477 some compilers to warn. To avoid this, I renamed it (as "blankclass").
3479 7. Cosmetic: (a) only output another newline at the end of pcretest if it is
3480 prompting; (b) run "./pcretest /dev/null" at the start of the test script
3481 so the version is shown; (c) stop "make test" echoing "./RunTest".
3483 8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
3485 9. The prototype for memmove() for systems that don't have it was using
3486 size_t, but the inclusion of the header that defines size_t was later. I've
3487 moved the #includes for the C headers earlier to avoid this.
3489 10. Added some adjustments to the code to make it easier to compiler on certain
3490 special systems:
3492 (a) Some "const" qualifiers were missing.
3493 (b) Added the macro EXPORT before all exported functions; by default this
3494 is defined to be empty.
3495 (c) Changed the dftables auxiliary program (that builds chartables.c) so
3496 that it reads its output file name as an argument instead of writing
3497 to the standard output and assuming this can be redirected.
3499 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
3500 class containing characters with values greater than 255, PCRE compilation
3501 went into a loop.
3503 12. A recursive reference to a subpattern that was within another subpattern
3504 that had a minimum quantifier of zero caused PCRE to crash. For example,
3505 (x(y(?2))z)? provoked this bug with a subject that got as far as the
3506 recursion. If the recursively-called subpattern itself had a zero repeat,
3507 that was OK.
3509 13. In pcretest, the buffer for reading a data line was set at 30K, but the
3510 buffer into which it was copied (for escape processing) was still set at
3511 1024, so long lines caused crashes.
3513 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
3514 "internal error: code overflow...". This applied to any character class
3515 that was followed by a possessive quantifier.
3517 15. Modified the Makefile to add libpcre.la as a prerequisite for
3518 libpcreposix.la because I was told this is needed for a parallel build to
3519 work.
3521 16. If a pattern that contained .* following optional items at the start was
3522 studied, the wrong optimizing data was generated, leading to matching
3523 errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
3524 matching string must start with a or b or c. The correct conclusion for
3525 this pattern is that a match can start with any character.
3528 Version 4.4 13-Aug-03
3529 ---------------------
3531 1. In UTF-8 mode, a character class containing characters with values between
3532 127 and 255 was not handled correctly if the compiled pattern was studied.
3533 In fixing this, I have also improved the studying algorithm for such
3534 classes (slightly).
3536 2. Three internal functions had redundant arguments passed to them. Removal
3537 might give a very teeny performance improvement.
3539 3. Documentation bug: the value of the capture_top field in a callout is *one
3540 more than* the number of the hightest numbered captured substring.
3542 4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
3543 in incorrectly linking with a previously installed version. They now link
3544 explicitly with libpcre.la.
3546 5. configure.in no longer needs to recognize Cygwin specially.
3548 6. A problem in pcre.in for Windows platforms is fixed.
3550 7. If a pattern was successfully studied, and the -d (or /D) flag was given to
3551 pcretest, it used to include the size of the study block as part of its
3552 output. Unfortunately, the structure contains a field that has a different
3553 size on different hardware architectures. This meant that the tests that
3554 showed this size failed. As the block is currently always of a fixed size,
3555 this information isn't actually particularly useful in pcretest output, so
3556 I have just removed it.
3558 8. Three pre-processor statements accidentally did not start in column 1.
3559 Sadly, there are *still* compilers around that complain, even though
3560 standard C has not required this for well over a decade. Sigh.
3562 9. In pcretest, the code for checking callouts passed small integers in the
3563 callout_data field, which is a void * field. However, some picky compilers
3564 complained about the casts involved for this on 64-bit systems. Now
3565 pcretest passes the address of the small integer instead, which should get
3566 rid of the warnings.
3568 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
3569 both compile and run time, and gives an error if an invalid UTF-8 sequence
3570 is found. There is a option for disabling this check in cases where the
3571 string is known to be correct and/or the maximum performance is wanted.
3573 11. In response to a bug report, I changed one line in Makefile.in from
3575 -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
3576 to
3577 -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
3579 to look similar to other lines, but I have no way of telling whether this
3580 is the right thing to do, as I do not use Windows. No doubt I'll get told
3581 if it's wrong...
3584 Version 4.3 21-May-03
3585 ---------------------
3587 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
3588 Makefile.
3590 2. Some refactoring to improve the quality of the code:
3592 (i) The utf8_table... variables are now declared "const".
3594 (ii) The code for \cx, which used the "case flipping" table to upper case
3595 lower case letters, now just substracts 32. This is ASCII-specific,
3596 but the whole concept of \cx is ASCII-specific, so it seems
3597 reasonable.
3599 (iii) PCRE was using its character types table to recognize decimal and
3600 hexadecimal digits in the pattern. This is silly, because it handles
3601 only 0-9, a-f, and A-F, but the character types table is locale-
3602 specific, which means strange things might happen. A private
3603 table is now used for this - though it costs 256 bytes, a table is
3604 much faster than multiple explicit tests. Of course, the standard
3605 character types table is still used for matching digits in subject
3606 strings against \d.
3608 (iv) Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
3609 ending in _t are). So I've renamed it as ESC_tee.
3611 3. The first argument for regexec() in the POSIX wrapper should have been
3612 defined as "const".
3614 4. Changed pcretest to use malloc() for its buffers so that they can be
3615 Electric Fenced for debugging.
3617 5. There were several places in the code where, in UTF-8 mode, PCRE would try
3618 to read one or more bytes before the start of the subject string. Often this
3619 had no effect on PCRE's behaviour, but in some circumstances it could
3620 provoke a segmentation fault.
3622 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
3623 to try to read one or more bytes before the start of the subject string.
3625 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
3626 UTF-8 support could misbehave in various ways if the subject string
3627 contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
3628 area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
3629 back over UTF-8 characters.)
3632 Version 4.2 14-Apr-03
3633 ---------------------
3635 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
3637 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
3638 [ON_WINDOWS]: new variable, "#" on non-Windows platforms
3639 [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
3640 [WIN_PREFIX]: new variable, "cyg" for Cygwin
3641 * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
3643 Note: automatic setting of the BUILD variables is not yet working
3644 set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
3645 compile-time but not at link-time
3646 [LINK]: use for linking executables only
3647 make different versions for Windows and non-Windows
3648 [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
3649 libraries
3650 [LINK_FOR_BUILD]: new variable
3651 [OBJEXT]: use throughout
3652 [EXEEXT]: use throughout
3653 <winshared>: new target
3654 <wininstall>: new target
3655 <dftables.o>: use native compiler
3656 <dftables>: use native linker
3657 <install>: handle Windows platform correctly
3658 <clean>: ditto
3659 <check>: ditto
3660 copy DLL to top builddir before testing
3662 As part of these changes, -no-undefined was removed again. This was reported
3663 to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
3664 in any case.
3666 3. Some tidies to get rid of compiler warnings:
3668 . In the match_data structure, match_limit was an unsigned long int, whereas
3669 match_call_count was an int. I've made them both unsigned long ints.
3671 . In pcretest the fact that a const uschar * doesn't automatically cast to
3672 a void * provoked a warning.
3674 . Turning on some more compiler warnings threw up some "shadow" variables
3675 and a few more missing casts.
3677 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
3678 option, a class that contained a single character with a value between 128
3679 and 255 (e.g. /[\xFF]/) caused PCRE to crash.
3681 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
3682 option, a class that contained several characters, but with at least one
3683 whose value was between 128 and 255 caused PCRE to crash.
3686 Version 4.1 12-Mar-03
3687 ---------------------
3689 1. Compiling with gcc -pedantic found a couple of places where casts were
3690 needed, and a string in dftables.c that was longer than standard compilers are
3691 required to support.
3693 2. Compiling with Sun's compiler found a few more places where the code could
3694 be tidied up in order to avoid warnings.
3696 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
3697 first of these names is deprecated in the latest Autoconf in favour of the name
3698 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
3699 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
3700 analogy I have changed it to CFLAGS_FOR_BUILD.
3702 4. Added -no-undefined to the linking command in the Makefile, because this is
3703 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
3704 linking step for the pcreposix library.
3706 5. PCRE was failing to diagnose the case of two named groups with the same
3707 name.
3709 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
3710 literal character that is needed in the subject for a match, and scans along to
3711 ensure that it is present before embarking on the full matching process. This
3712 saves time in cases of nested unlimited repeats that are never going to match.
3713 Problem: the scan can take a lot of time if the subject is very long (e.g.
3714 megabytes), thus penalizing straightforward matches. It is now done only if the
3715 amount of subject to be scanned is less than 1000 bytes.
3717 7. A lesser problem with the same optimization is that it was recording the
3718 first character of an anchored pattern as "needed", thus provoking a search
3719 right along the subject, even when the first match of the pattern was going to
3720 fail. The "needed" character is now not set for anchored patterns, unless it
3721 follows something in the pattern that is of non-fixed length. Thus, it still
3722 fulfils its original purpose of finding quick non-matches in cases of nested
3723 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
3726 Version 4.0 17-Feb-03
3727 ---------------------
3729 1. If a comment in an extended regex that started immediately after a meta-item
3730 extended to the end of string, PCRE compiled incorrect data. This could lead to
3731 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
3733 2. Moved to autoconf 2.53 and libtool 1.4.2.
3735 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
3736 the special perltest8 script is no longer needed - all the tests can be run
3737 from a single perltest script.
3739 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
3740 by \s. It has now been removed in PCRE. This means it isn't recognized as
3741 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
3742 class [:space:] *does* include VT, thereby creating a mess.
3744 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
3745 space and tab.
3747 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
3748 its new features into the main test script, reducing the number of scripts.
3750 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
3751 were backward compatible, and made the (?i) apply to the whole pattern, as if
3752 /i were given. Now it behaves more logically, and applies the option setting
3753 only to what follows. PCRE has been changed to follow suit. However, if it
3754 finds options settings right at the start of the pattern, it extracts them into
3755 the global options, as before. Thus, they show up in the info data.
3757 8. Added support for the \Q...\E escape sequence. Characters in between are
3758 treated as literals. This is slightly different from Perl in that $ and @ are
3759 also handled as literals inside the quotes. In Perl, they will cause variable
3760 interpolation. Note the following examples:
3762 Pattern PCRE matches Perl matches
3764 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
3765 \Qabc\$xyz\E abc\$xyz abc\$xyz
3766 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
3768 For compatibility with Perl, \Q...\E sequences are recognized inside character
3769 classes as well as outside them.
3771 9. Re-organized 3 code statements in pcretest to avoid "overflow in
3772 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
3773 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
3774 signed/unsigned warnings.
3776 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
3777 option for pcretest, so I've replaced it by a simple function that does just
3778 that job.
3780 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
3781 "pcregrep -".
3783 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
3784 Java package. This provides some syntactic sugar for simple cases of what my
3785 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
3786 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
3787 item, you can use this simplified notation. Note that only makes sense with
3788 greedy quantifiers. Consequently, the use of the possessive quantifier forces
3789 greediness, whatever the setting of the PCRE_UNGREEDY option.
3791 13. A change of greediness default within a pattern was not taking effect at
3792 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
3793 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
3794 was abstracted outside.
3796 14. PCRE now supports the \G assertion. It is true when the current matching
3797 position is at the start point of the match. This differs from \A when the
3798 starting offset is non-zero. Used with the /g option of pcretest (or similar
3799 code), it works in the same way as it does for Perl's /g option. If all
3800 alternatives of a regex begin with \G, the expression is anchored to the start
3801 match position, and the "anchored" flag is set in the compiled expression.
3803 15. Some bugs concerning the handling of certain option changes within patterns
3804 have been fixed. These applied to options other than (?ims). For example,
3805 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
3806 been the other way round. Some of this was related to change 7 above.
3808 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
3809 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
3810 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
3811 POSIX classes only within a class (e.g. /[[:alpha:]]/).
3813 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
3814 mode. Unlike ".", it always matches newline, whatever the setting of
3815 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
3816 assertions. Perl allows it, but it doesn't (in general) work because it can't
3817 calculate the length of the lookbehind. At least, that's the case for Perl
3818 5.8.0 - I've been told they are going to document that it doesn't work in
3819 future.
3821 18. Added an error diagnosis for escapes that PCRE does not support: these are
3822 \L, \l, \N, \P, \p, \U, \u, and \X.
3824 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
3825 reading past the end of the pattern in cases such as /[abcd/.
3827 20. PCRE was getting more memory than necessary for patterns with classes that
3828 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
3830 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
3831 compiling PCRE for use with Virtual Pascal.
3833 22. Small fix to the Makefile to make it work properly if the build is done
3834 outside the source tree.
3836 23. Added a new extension: a condition to go with recursion. If a conditional
3837 subpattern starts with (?(R) the "true" branch is used if recursion has
3838 happened, whereas the "false" branch is used only at the top level.
3840 24. When there was a very long string of literal characters (over 255 bytes
3841 without UTF support, over 250 bytes with UTF support), the computation of how
3842 much memory was required could be incorrect, leading to segfaults or other
3843 strange effects.
3845 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
3846 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
3847 there was a subsequent back reference to those brackets. This meant that, for
3848 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
3849 possible to check for precisely this case. All we can do is abandon the
3850 optimization if .* occurs inside capturing brackets when there are any back
3851 references whatsoever. (See below for a better fix that came later.)
3853 26. The handling of the optimization for finding the first character of a
3854 non-anchored pattern, and for finding a character that is required later in the
3855 match were failing in some cases. This didn't break the matching; it just
3856 failed to optimize when it could. The way this is done has been re-implemented.
3858 27. Fixed typo in error message for invalid (?R item (it said "(?p").
3860 28. Added a new feature that provides some of the functionality that Perl
3861 provides with (?{...}). The facility is termed a "callout". The way it is done
3862 in PCRE is for the caller to provide an optional function, by setting
3863 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
3864 global variable. By default it is unset, which disables all calling out. To get
3865 the function called, the regex must include (?C) at appropriate points. This
3866 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
3867 This provides a means of identifying different callout points. When PCRE
3868 reaches such a point in the regex, if pcre_callout has been set, the external
3869 function is called. It is provided with data in a structure called
3870 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
3871 matching continues; if it returns a non-zero value, the match at the current
3872 point fails. However, backtracking will occur if possible. [This was changed
3873 later and other features added - see item 49 below.]
3875 29. pcretest is upgraded to test the callout functionality. It provides a
3876 callout function that displays information. By default, it shows the start of
3877 the match and the current position in the text. There are some new data escapes
3878 to vary what happens:
3880 \C+ in addition, show current contents of captured substrings
3881 \C- do not supply a callout function
3882 \C!n return 1 when callout number n is reached
3883 \C!n!m return 1 when callout number n is reached for the mth time
3885 30. If pcregrep was called with the -l option and just a single file name, it
3886 output "<stdin>" if a match was found, instead of the file name.
3888 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
3889 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
3890 pcre_exec(). This saves a malloc/free per call. The default value of
3891 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
3892 when configuring.
3894 32. The default maximum size of a compiled pattern is 64K. There have been a
3895 few cases of people hitting this limit. The code now uses macros to handle the
3896 storing of links as offsets within the compiled pattern. It defaults to 2-byte
3897 links, but this can be changed to 3 or 4 bytes by --with-link-size when
3898 configuring. Tests 2 and 5 work only with 2-byte links because they output
3899 debugging information about compiled patterns.
3901 33. Internal code re-arrangements:
3903 (a) Moved the debugging function for printing out a compiled regex into
3904 its own source file (printint.c) and used #include to pull it into
3905 pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
3906 separate copies.
3908 (b) Defined the list of op-code names for debugging as a macro in
3909 internal.h so that it is next to the definition of the opcodes.
3911 (c) Defined a table of op-code lengths for simpler skipping along compiled
3912 code. This is again a macro in internal.h so that it is next to the
3913 definition of the opcodes.
3915 34. Added support for recursive calls to individual subpatterns, along the
3916 lines of Robin Houston's patch (but implemented somewhat differently).
3918 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
3919 allow it to read and process whole directories in Win32. This code was
3920 contributed by Lionel Fourquaux; it has not been tested by me.
3922 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
3923 used to name a group. Names consist of alphanumerics and underscores, and must
3924 be unique. Back references use the syntax (?P=name) and recursive calls use
3925 (?P>name) which is a PCRE extension to the Python extension. Groups still have
3926 numbers. The function pcre_fullinfo() can be used after compilation to extract
3927 a name/number map. There are three relevant calls:
3929 PCRE_INFO_NAMEENTRYSIZE yields the size of each entry in the map
3930 PCRE_INFO_NAMECOUNT yields the number of entries
3931 PCRE_INFO_NAMETABLE yields a pointer to the map.
3933 The map is a vector of fixed-size entries. The size of each entry depends on
3934 the length of the longest name used. The first two bytes of each entry are the
3935 group number, most significant byte first. There follows the corresponding
3936 name, zero terminated. The names are in alphabetical order.
3938 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
3939 case instead of 255. Making it the same both with and without UTF-8 support
3940 means that the same test output works with both.
3942 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
3943 calling malloc() with a zero argument.
3945 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
3946 optimization. I've improved things by keeping a bitmap of backreferences with
3947 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
3948 fact referenced, the optimization can be applied. It is unlikely that a
3949 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
3950 the match to follow \n) will appear inside brackets with a number greater than
3951 31, but if it does, any back reference > 31 suppresses the optimization.
3953 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
3954 of disabling numbered capturing parentheses. Any opening parenthesis that is
3955 not followed by ? behaves as if it were followed by ?: but named parentheses
3956 can still be used for capturing (and they will acquire numbers in the usual
3957 way).
3959 41. Redesigned the return codes from the match() function into yes/no/error so
3960 that errors can be passed back from deep inside the nested calls. A malloc
3961 failure while inside a recursive subpattern call now causes the
3962 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
3964 42. It is now possible to set a limit on the number of times the match()
3965 function is called in a call to pcre_exec(). This facility makes it possible to
3966 limit the amount of recursion and backtracking, though not in a directly
3967 obvious way, because the match() function is used in a number of different
3968 circumstances. The count starts from zero for each position in the subject
3969 string (for non-anchored patterns). The default limit is, for compatibility, a
3970 large number, namely 10 000 000. You can change this in two ways:
3972 (a) When configuring PCRE before making, you can use --with-match-limit=n
3973 to set a default value for the compiled library.
3975 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
3976 a different value is set. See 45 below.
3978 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
3980 43. Added a new function pcre_config(int, void *) to enable run-time extraction
3981 of things that can be changed at compile time. The first argument specifies
3982 what is wanted and the second points to where the information is to be placed.
3983 The current list of available information is:
3987 The output is an integer that is set to one if UTF-8 support is available;
3988 otherwise it is set to zero.
3992 The output is an integer that it set to the value of the code that is used for
3993 newline. It is either LF (10) or CR (13).
3997 The output is an integer that contains the number of bytes used for internal
3998 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
4002 The output is an integer that contains the threshold above which the POSIX
4003 interface uses malloc() for output vectors. See item 31 above.
4007 The output is an unsigned integer that contains the default limit of the number
4008 of match() calls in a pcre_exec() execution. See 42 above.
4010 44. pcretest has been upgraded by the addition of the -C option. This causes it
4011 to extract all the available output from the new pcre_config() function, and to
4012 output it. The program then exits immediately.
4014 45. A need has arisen to pass over additional data with calls to pcre_exec() in
4015 order to support additional features. One way would have been to define
4016 pcre_exec2() (for example) with extra arguments, but this would not have been
4017 extensible, and would also have required all calls to the original function to
4018 be mapped to the new one. Instead, I have chosen to extend the mechanism that
4019 is used for passing in "extra" data from pcre_study().
4021 The pcre_extra structure is now exposed and defined in pcre.h. It currently
4022 contains the following fields:
4024 flags a bitmap indicating which of the following fields are set
4025 study_data opaque data from pcre_study()
4026 match_limit a way of specifying a limit on match() calls for a specific
4027 call to pcre_exec()
4028 callout_data data for callouts (see 49 below)
4030 The flag bits are also defined in pcre.h, and are
4036 The pcre_study() function now returns one of these new pcre_extra blocks, with
4037 the actual study data pointed to by the study_data field, and the
4038 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
4039 before. That is, this change is entirely upwards-compatible and requires no
4040 change to existing code.
4042 If you want to pass in additional data to pcre_exec(), you can either place it
4043 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
4044 block.
4046 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
4047 data string contains the escape sequence \M, pcretest calls pcre_exec() several
4048 times with different match limits, until it finds the minimum value needed for
4049 pcre_exec() to complete. The value is then output. This can be instructive; for
4050 most simple matches the number is quite small, but for pathological cases it
4051 gets very large very quickly.
4053 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
4054 returns the size of the data block pointed to by the study_data field in a
4055 pcre_extra block, that is, the value that was passed as the argument to
4056 pcre_malloc() when PCRE was getting memory in which to place the information
4057 created by pcre_study(). The fourth argument should point to a size_t variable.
4058 pcretest has been extended so that this information is shown after a successful
4059 pcre_study() call when information about the compiled regex is being displayed.
4061 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
4062 because what follows is always an absolute path. (Later: it turns out that this
4063 is more than cosmetic for MinGW, because it doesn't like empty path
4064 components.)
4066 49. Some changes have been made to the callout feature (see 28 above):
4068 (i) A callout function now has three choices for what it returns:
4070 0 => success, carry on matching
4071 > 0 => failure at this point, but backtrack if possible
4072 < 0 => serious error, return this value from pcre_exec()
4074 Negative values should normally be chosen from the set of PCRE_ERROR_xxx
4075 values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
4076 "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
4077 use by callout functions. It will never be used by PCRE itself.
4079 (ii) The pcre_extra structure (see 45 above) has a void * field called
4080 callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
4081 pcre_callout_block structure has a field of the same name. The contents of
4082 the field passed in the pcre_extra structure are passed to the callout
4083 function in the corresponding field in the callout block. This makes it
4084 easier to use the same callout-containing regex from multiple threads. For
4085 testing, the pcretest program has a new data escape
4087 \C*n pass the number n (may be negative) as callout_data
4089 If the callout function in pcretest receives a non-zero value as
4090 callout_data, it returns that value.
4092 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
4093 there were some redundant $(CFLAGS) in commands that are now specified as
4094 $(LINK), which already includes $(CFLAGS).
4096 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
4097 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
4098 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
4099 one-byte characters throughout. Note that case-insensitive matching applies
4100 only to characters whose values are less than 256. PCRE doesn't support the
4101 notion of cases for higher-valued characters.
4103 (i) A character class whose characters are all within 0-255 is handled as
4104 a bit map, and the map is inverted for negative classes. Previously, a
4105 character > 255 always failed to match such a class; however it should
4106 match if the class was a negative one (e.g. [^ab]). This has been fixed.
4108 (ii) A negated character class with a single character < 255 is coded as
4109 "not this character" (OP_NOT). This wasn't working properly when the test
4110 character was multibyte, either singly or repeated.
4112 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
4113 mode, for example: \x{100}{2,3}.
4115 (iv) The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
4116 singly or repeated) now correctly test multibyte characters. However,
4117 PCRE doesn't recognize any characters with values greater than 255 as
4118 digits, spaces, or word characters. Such characters always match \D, \S,
4119 and \W, and never match \d, \s, or \w.
4121 (v) Classes may now contain characters and character ranges with values
4122 greater than 255. For example: [ab\x{100}-\x{400}].
4124 (vi) pcregrep now has a --utf-8 option (synonym -u) which makes it call
4125 PCRE in UTF-8 mode.
4127 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
4128 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
4129 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
4130 value.)
4132 53. The single man page has become too large. I have therefore split it up into
4133 a number of separate man pages. These also give rise to individual HTML pages;
4134 these are now put in a separate directory, and there is an index.html page that
4135 lists them all. Some hyperlinking between the pages has been installed.
4137 54. Added convenience functions for handling named capturing parentheses.
4139 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
4140 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
4141 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
4142 are faulted.
4144 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
4145 calling configure. These values are used when compiling the dftables.c program
4146 which is run to generate the source of the default character tables. They
4147 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
4148 you will need to set these values.
4150 57. Updated the building process for Windows DLL, as provided by Fred Cox.
4153 Version 3.9 02-Jan-02
4154 ---------------------
4156 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
4158 2. If --disable-static was given, the building process failed when trying to
4159 build pcretest and pcregrep. (For some reason it was using libtool to compile
4160 them, which is not right, as they aren't part of the library.)
4163 Version 3.8 18-Dec-01
4164 ---------------------
4166 1. The experimental UTF-8 code was completely screwed up. It was packing the
4167 bytes in the wrong order. How dumb can you get?
4170 Version 3.7 29-Oct-01
4171 ---------------------
4173 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
4174 This caused pcretest, when used on the test data, to segfault. Unfortunately,
4175 this didn't happen under Solaris 8, where I normally test things.
4177 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
4178 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
4179 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
4180 this fix an hour or so after the initial 3.7 release.)
4183 Version 3.6 23-Oct-01
4184 ---------------------
4186 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
4187 offsets passed as NULL with zero offset count.
4189 2. The config.guess and config.sub files had not been updated when I moved to
4190 the latest autoconf.
4193 Version 3.5 15-Aug-01
4194 ---------------------
4196 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
4197 had been forgotten.
4199 2. By using declared but undefined structures, we can avoid using "void"
4200 definitions in pcre.h while keeping the internal definitions of the structures
4201 private.
4203 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
4204 user point of view, this means that both static and shared libraries are built
4205 by default, but this can be individually controlled. More of the work of
4206 handling this static/shared cases is now inside libtool instead of PCRE's make
4207 file.
4209 4. The pcretest utility is now installed along with pcregrep because it is
4210 useful for users (to test regexs) and by doing this, it automatically gets
4211 relinked by libtool. The documentation has been turned into a man page, so
4212 there are now .1, .txt, and .html versions in /doc.
4214 5. Upgrades to pcregrep:
4215 (i) Added long-form option names like gnu grep.
4216 (ii) Added --help to list all options with an explanatory phrase.
4217 (iii) Added -r, --recursive to recurse into sub-directories.
4218 (iv) Added -f, --file to read patterns from a file.
4220 6. pcre_exec() was referring to its "code" argument before testing that
4221 argument for NULL (and giving an error if it was NULL).
4223 7. Upgraded Makefile.in to allow for compiling in a different directory from
4224 the source directory.
4226 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
4227 options bits, the pointer it was passed was to an int instead of to an unsigned
4228 long int. This mattered only on 64-bit systems.
4230 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
4231 generated) instead of pcre.in, which it its source. Also made the same change
4232 in several of the .c files.
4234 10. A new release of gcc defines printf() as a macro, which broke pcretest
4235 because it had an ifdef in the middle of a string argument for printf(). Fixed
4236 by using separate calls to printf().
4238 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
4239 script, to force use of CR or LF instead of \n in the source. On non-Unix
4240 systems, the value can be set in config.h.
4242 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
4243 absolute limit. Changed the text of the error message to make this clear, and
4244 likewise updated the man page.
4246 13. The limit of 99 on the number of capturing subpatterns has been removed.
4247 The new limit is 65535, which I hope will not be a "real" limit.
4250 Version 3.4 22-Aug-00
4251 ---------------------
4253 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
4255 2. Diagnose condition (?(0) as an error instead of crashing on matching.
4258 Version 3.3 01-Aug-00
4259 ---------------------
4261 1. If an octal character was given, but the value was greater than \377, it
4262 was not getting masked to the least significant bits, as documented. This could
4263 lead to crashes in some systems.
4265 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
4266 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
4268 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
4269 These just pass their arguments on to (pcre_free)(), but they are provided
4270 because some uses of PCRE bind it to non-C systems that can call its functions,
4271 but cannot call free() or pcre_free() directly.
4273 4. Add "make test" as a synonym for "make check". Corrected some comments in
4274 the Makefile.
4276 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
4277 Makefile.
4279 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
4280 command called pgrep for grepping around the active processes.
4282 7. Added the beginnings of support for UTF-8 character strings.
4284 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
4285 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
4286 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
4287 out for the ar command.)
4290 Version 3.2 12-May-00
4291 ---------------------
4293 This is purely a bug fixing release.
4295 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
4296 of ZA. This was just one example of several cases that could provoke this bug,
4297 which was introduced by change 9 of version 2.00. The code for breaking
4298 infinite loops after an iteration that matches an empty string was't working
4299 correctly.
4301 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
4302 when matched against abbab (for example). After matching an empty string, it
4303 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
4304 caused it to match further down the string than it should.
4306 3. The code contained an inclusion of sys/types.h. It isn't clear why this
4307 was there because it doesn't seem to be needed, and it causes trouble on some
4308 systems, as it is not a Standard C header. It has been removed.
4310 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
4311 were reported on the Macintosh. The changes were from
4313 while ((c = *(++ptr)) != 0 && c != '\n');
4314 to
4315 while ((c = *(++ptr)) != 0 && c != '\n') ;
4317 Totally extraordinary, but if that's what it takes...
4319 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
4320 available. Added HAVE_BCOPY and an autoconf test for it; if neither
4321 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
4322 assumes the way PCRE uses memmove() (always moving upwards).
4324 6. PCRE is being used in one environment where strchr() is not available. There
4325 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
4326 faster code anyway.
4329 Version 3.1 09-Feb-00
4330 ---------------------
4332 The only change in this release is the fixing of some bugs in Makefile.in for
4333 the "install" target:
4335 (1) It was failing to install pcreposix.h.
4337 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
4340 Version 3.0 01-Feb-00
4341 ---------------------
4343 1. Add support for the /+ modifier to perltest (to output $` like it does in
4344 pcretest).
4346 2. Add support for the /g modifier to perltest.
4348 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
4349 matches null strings.
4351 4. Fix perltest so that it doesn't do unwanted things when fed an empty
4352 pattern. Perl treats empty patterns specially - it reuses the most recent
4353 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
4354 effect.
4356 5. The POSIX interface was broken in that it was just handing over the POSIX
4357 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
4358 required a bigger vector, with some working space on the end. This means that
4359 the POSIX wrapper now has to get and free some memory, and copy the results.
4361 6. Added some simple autoconf support, placing the test data and the
4362 documentation in separate directories, re-organizing some of the
4363 information files, and making it build pcre-config (a GNU standard). Also added
4364 libtool support for building PCRE as a shared library, which is now the
4365 default.
4367 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
4368 09 are not valid octal constants. Single digits will be used for minor values
4369 less than 10.
4371 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
4372 existing programs that set these in the POSIX interface can use PCRE without
4373 modification.
4375 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
4376 return all that pcre_info() returns, plus additional data. The pcre_info()
4377 function is retained for compatibility, but is considered to be obsolete.
4379 10. Added experimental recursion feature (?R) to handle one common case that
4380 Perl 5.6 will be able to do with (?p{...}).
4382 11. Added support for POSIX character classes like [:alpha:], which Perl is
4383 adopting.
4386 Version 2.08 31-Aug-99
4387 ----------------------
4389 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
4390 trying to match at the startoffset position, but instead was moving forward to
4391 the next newline as if a previous match had failed.
4393 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
4394 and could get into a loop if a null string was matched other than at the start
4395 of the subject.
4397 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
4398 be distinguished at compile time, and for completeness also added PCRE_DATE.
4400 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
4401 in GnuWin32 environments.
4404 Version 2.07 29-Jul-99
4405 ----------------------
4407 1. The documentation is now supplied in plain text form and HTML as well as in
4408 the form of man page sources.
4410 2. C++ compilers don't like assigning (void *) values to other pointer types.
4411 In particular this affects malloc(). Although there is no problem in Standard
4412 C, I've put in casts to keep C++ compilers happy.
4414 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
4415 should be (const char *).
4417 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
4418 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
4419 However, I haven't made this a standard facility. The documentation doesn't
4420 mention it, and the Makefile doesn't support it.
4422 5. The Makefile now contains an "install" target, with editable destinations at
4423 the top of the file. The pcretest program is not installed.
4425 6. pgrep -V now gives the PCRE version number and date.
4427 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
4428 causing the entire string to be ignored, instead of just the last character.
4430 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
4431 non-matching string, it can take a very, very long time, even for strings of
4432 quite modest length, because of the nested recursion. PCRE now does better in
4433 some of these cases. It does this by remembering the last required literal
4434 character in the pattern, and pre-searching the subject to ensure it is present
4435 before running the real match. In other words, it applies a heuristic to detect
4436 some types of certain failure quickly, and in the above example, if presented
4437 with a string that has no trailing " it gives "no match" very quickly.
4439 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
4440 other alternatives are tried instead.
4443 Version 2.06 09-Jun-99
4444 ----------------------
4446 1. Change pcretest's output for amount of store used to show just the code
4447 space, because the remainder (the data block) varies in size between 32-bit and
4448 64-bit systems.
4450 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
4451 start matching at. This allows lookbehinds to work when searching for multiple
4452 occurrences in a string.
4454 3. Added additional options to pcretest for testing multiple occurrences:
4456 /+ outputs the rest of the string that follows a match
4457 /g loops for multiple occurrences, using the new startoffset argument
4458 /G loops for multiple occurrences by passing an incremented pointer
4460 4. PCRE wasn't doing the "first character" optimization for patterns starting
4461 with \b or \B, though it was doing it for other lookbehind assertions. That is,
4462 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
4463 the letter 'x'. On long subject strings, this gives a significant speed-up.
4466 Version 2.05 21-Apr-99
4467 ----------------------
4469 1. Changed the type of magic_number from int to long int so that it works
4470 properly on 16-bit systems.
4472 2. Fixed a bug which caused patterns starting with .* not to work correctly
4473 when the subject string contained newline characters. PCRE was assuming
4474 anchoring for such patterns in all cases, which is not correct because .* will
4475 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
4476 DOTALL is set at top level; otherwise it knows that patterns starting with .*
4477 must be retried after every newline in the subject.
4480 Version 2.04 18-Feb-99
4481 ----------------------
4483 1. For parenthesized subpatterns with repeats whose minimum was zero, the
4484 computation of the store needed to hold the pattern was incorrect (too large).
4485 If such patterns were nested a few deep, this could multiply and become a real
4486 problem.
4488 2. Added /M option to pcretest to show the memory requirement of a specific
4489 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
4491 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
4492 compiled in such a way that the backtracking after subsequent failure was
4493 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
4494 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
4497 Version 2.03 02-Feb-99
4498 ----------------------
4500 1. Fixed typo and small mistake in man page.
4502 2. Added 4th condition (GPL supersedes if conflict) and created separate
4503 LICENCE file containing the conditions.
4505 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
4506 Perl, that is the internal \ allows the delimiter to be included in the
4507 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
4508 the final delimiter, add \ to the end of the pattern (to test the error).
4510 4. Added the convenience functions for extracting substrings after a successful
4511 match. Updated pcretest to make it able to test these functions.
4514 Version 2.02 14-Jan-99
4515 ----------------------
4517 1. Initialized the working variables associated with each extraction so that
4518 their saving and restoring doesn't refer to uninitialized store.
4520 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
4521 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
4522 fix the problem.
4524 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
4525 calls, and wasn't printing the correct value for compiling calls. Increased the
4526 default value of LOOPREPEAT, and the number of significant figures in the
4527 times.
4529 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
4531 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
4532 a building problem on Windows NT with a FAT file system.
4535 Version 2.01 21-Oct-98
4536 ----------------------
4538 1. Changed the API for pcre_compile() to allow for the provision of a pointer
4539 to character tables built by pcre_maketables() in the current locale. If NULL
4540 is passed, the default tables are used.
4543 Version 2.00 24-Sep-98
4544 ----------------------
4546 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
4547 it any more.
4549 2. Allow quantification of (?>) groups, and make it work correctly.
4551 3. The first character computation wasn't working for (?>) groups.
4553 4. Correct the implementation of \Z (it is permitted to match on the \n at the
4554 end of the subject) and add 5.005's \z, which really does match only at the
4555 very end of the subject.
4557 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
4559 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
4560 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
4561 localized options. All options to pcre_study() were also removed.
4563 7. Add other new features from 5.005:
4565 $(?<= positive lookbehind
4566 $(?<! negative lookbehind
4567 (?imsx-imsx) added the unsetting capability
4568 such a setting is global if at outer level; local otherwise
4569 (?imsx-imsx:) non-capturing groups with option setting
4570 (?(cond)re|re) conditional pattern matching
4572 A backreference to itself in a repeated group matches the previous
4573 captured string.
4575 8. General tidying up of studying (both automatic and via "study")
4576 consequential on the addition of new assertions.
4578 9. As in 5.005, unlimited repeated groups that could match an empty substring
4579 are no longer faulted at compile time. Instead, the loop is forcibly broken at
4580 runtime if any iteration does actually match an empty substring.
4582 10. Include the RunTest script in the distribution.
4584 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
4585 discrepancies, some of which were old and were also with respect to 5.004. They
4586 have now been fixed.
4589 Version 1.09 28-Apr-98
4590 ----------------------
4592 1. A negated single character class followed by a quantifier with a minimum
4593 value of one (e.g. [^x]{1,6} ) was not compiled correctly. This could lead to
4594 program crashes, or just wrong answers. This did not apply to negated classes
4595 containing more than one character, or to minima other than one.
4598 Version 1.08 27-Mar-98
4599 ----------------------
4601 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
4603 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
4604 latter must appear before anything that relies on it in the pattern.
4607 Version 1.07 16-Feb-98
4608 ----------------------
4610 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
4611 repeat of a potentially empty string).
4614 Version 1.06 23-Jan-98
4615 ----------------------
4617 1. Added Markus Oberhumer's little patches for C++.
4619 2. Literal strings longer than 255 characters were broken.
4622 Version 1.05 23-Dec-97
4623 ----------------------
4625 1. Negated character classes containing more than one character were failing if
4626 PCRE_CASELESS was set at run time.
4629 Version 1.04 19-Dec-97
4630 ----------------------
4632 1. Corrected the man page, where some "const" qualifiers had been omitted.
4634 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
4635 input syntax.
4637 3. Fixed memory leak which occurred when a regex with back references was
4638 matched with an offsets vector that wasn't big enough. The temporary memory
4639 that is used in this case wasn't being freed if the match failed.
4641 4. Tidied pcretest to ensure it frees memory that it gets.
4643 5. Temporary memory was being obtained in the case where the passed offsets
4644 vector was exactly big enough.
4646 6. Corrected definition of offsetof() from change 5 below.
4648 7. I had screwed up change 6 below and broken the rules for the use of
4649 setjmp(). Now fixed.
4652 Version 1.03 18-Dec-97
4653 ----------------------
4655 1. A erroneous regex with a missing opening parenthesis was correctly
4656 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
4657 on some systems.
4659 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
4660 it was reported that one broken compiler failed on the former because "code" is
4661 also an independent variable.
4663 3. The erroneous regex a[]b caused an array overrun reference.
4665 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
4666 fail on data ending with that character. (It was going on too far, and checking
4667 the next character, typically a binary zero.) This was specific to the
4668 optimized code for single-character negative classes.
4670 5. Added a contributed patch from the TIN world which does the following:
4672 + Add an undef for memmove, in case the the system defines a macro for it.
4674 + Add a definition of offsetof(), in case there isn't one. (I don't know
4675 the reason behind this - offsetof() is part of the ANSI standard - but
4676 it does no harm).
4678 + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
4679 most of the places where whitespace preceded '#'. I have given up and
4680 allowed the remaining 2 cases to be at the margin.
4682 + Rename some variables in pcre to eliminate shadowing. This seems very
4683 pedantic, but does no harm, of course.
4685 6. Moved the call to setjmp() into its own function, to get rid of warnings
4686 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
4688 7. Constructs such as \d{8,} were compiling into the equivalent of
4689 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
4690 outcome, but in this particular case used more store than had been allocated,
4691 which caused the bug to be discovered because it threw up an internal error.
4693 8. The debugging code in both pcre and pcretest for outputting the compiled
4694 form of a regex was going wrong in the case of back references followed by
4695 curly-bracketed repeats.
4698 Version 1.02 12-Dec-97
4699 ----------------------
4701 1. Typos in pcre.3 and comments in the source fixed.
4703 2. Applied a contributed patch to get rid of places where it used to remove
4704 'const' from variables, and fixed some signed/unsigned and uninitialized
4705 variable warnings.
4707 3. Added the "runtest" target to Makefile.
4709 4. Set default compiler flag to -O2 rather than just -O.
4712 Version 1.01 19-Nov-97
4713 ----------------------
4715 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
4716 like /([ab]*)*/, that is, for classes with more than one character in them.
4718 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
4719 as /((?>a*))*/ (a PCRE_EXTRA facility).
4722 Version 1.00 18-Nov-97
4723 ----------------------
4725 1. Added compile-time macros to support systems such as SunOS4 which don't have
4726 memmove() or strerror() but have other things that can be used instead.
4728 2. Arranged that "make clean" removes the executables.
4731 Version 0.99 27-Oct-97
4732 ----------------------
4734 1. Fixed bug in code for optimizing classes with only one character. It was
4735 initializing a 32-byte map regardless, which could cause it to run off the end
4736 of the memory it had got.
4738 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
4741 Version 0.98 22-Oct-97
4742 ----------------------
4744 1. Fixed bug in code for handling temporary memory usage when there are more
4745 back references than supplied space in the ovector. This could cause segfaults.
4748 Version 0.97 21-Oct-97
4749 ----------------------
4751 1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
4753 2. Optimized negated single characters not to use a bit map.
4755 3. Brought error texts together as macro definitions; clarified some of them;
4756 fixed one that was wrong - it said "range out of order" when it meant "invalid
4757 escape sequence".
4759 4. Changed some char * arguments to const char *.
4761 5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
4763 6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
4764 pcretest.
4767 Version 0.96 16-Oct-97
4768 ----------------------
4770 1. Added a simple "pgrep" utility to the distribution.
4772 2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
4773 unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
4774 where "ddd" means "one or more decimal digits".
4776 3. Fixed serious bug. If a pattern had a back reference, but the call to
4777 pcre_exec() didn't supply a large enough ovector to record the related
4778 identifying subpattern, the match always failed. PCRE now remembers the number
4779 of the largest back reference, and gets some temporary memory in which to save
4780 the offsets during matching if necessary, in order to ensure that
4781 backreferences always work.
4783 4. Increased the compatibility with Perl in a number of ways:
4785 (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
4786 to request this handling. The option can be set at compile or exec time.
4788 (b) $ matches before a terminating newline by default; an option
4789 PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
4790 mode). The option can be set at compile or exec time.
4792 (c) The handling of \ followed by a digit other than 0 is now supposed to be
4793 the same as Perl's. If the decimal number it represents is less than 10
4794 or there aren't that many previous left capturing parentheses, an octal
4795 escape is read. Inside a character class, it's always an octal escape,
4796 even if it is a single digit.
4798 (d) An escaped but undefined alphabetic character is taken as a literal,
4799 unless PCRE_EXTRA is set. Currently this just reserves the remaining
4800 escapes.
4802 (e) {0} is now permitted. (The previous item is removed from the compiled
4803 pattern).
4805 5. Changed all the names of code files so that the basic parts are no longer
4806 than 10 characters, and abolished the teeny "globals.c" file.
4808 6. Changed the handling of character classes; they are now done with a 32-byte
4809 bit map always.
4811 7. Added the -d and /D options to pcretest to make it possible to look at the
4812 internals of compilation without having to recompile pcre.
4815 Version 0.95 23-Sep-97
4816 ----------------------
4818 1. Fixed bug in pre-pass concerning escaped "normal" characters such as \x5c or
4819 \x20 at the start of a run of normal characters. These were being treated as
4820 real characters, instead of the source characters being re-checked.
4823 Version 0.94 18-Sep-97
4824 ----------------------
4826 1. The functions are now thread-safe, with the caveat that the global variables
4827 containing pointers to malloc() and free() or alternative functions are the
4828 same for all threads.
4830 2. Get pcre_study() to generate a bitmap of initial characters for non-
4831 anchored patterns when this is possible, and use it if passed to pcre_exec().
4834 Version 0.93 15-Sep-97
4835 ----------------------
4837 1. /(b)|(:+)/ was computing an incorrect first character.
4839 2. Add pcre_study() to the API and the passing of pcre_extra to pcre_exec(),
4840 but not actually doing anything yet.
4842 3. Treat "-" characters in classes that cannot be part of ranges as literals,
4843 as Perl does (e.g. [-az] or [az-]).
4845 4. Set the anchored flag if a branch starts with .* or .*? because that tests
4846 all possible positions.
4848 5. Split up into different modules to avoid including unneeded functions in a
4849 compiled binary. However, compile and exec are still in one module. The "study"
4850 function is split off.
4852 6. The character tables are now in a separate module whose source is generated
4853 by an auxiliary program - but can then be edited by hand if required. There are
4854 now no calls to isalnum(), isspace(), isdigit(), isxdigit(), tolower() or
4855 toupper() in the code.
4857 7. Turn the malloc/free funtions variables into pcre_malloc and pcre_free and
4858 make them global. Abolish the function for setting them, as the caller can now
4859 set them directly.
4862 Version 0.92 11-Sep-97
4863 ----------------------
4865 1. A repeat with a fixed maximum and a minimum of 1 for an ordinary character
4866 (e.g. /a{1,3}/) was broken (I mis-optimized it).
4868 2. Caseless matching was not working in character classes if the characters in
4869 the pattern were in upper case.
4871 3. Make ranges like [W-c] work in the same way as Perl for caseless matching.
4873 4. Make PCRE_ANCHORED public and accept as a compile option.
4875 5. Add an options word to pcre_exec() and accept PCRE_ANCHORED and
4876 PCRE_CASELESS at run time. Add escapes \A and \I to pcretest to cause it to
4877 pass them.
4879 6. Give an error if bad option bits passed at compile or run time.
4881 7. Add PCRE_MULTILINE at compile and exec time, and (?m) as well. Add \M to
4882 pcretest to cause it to pass that flag.
4884 8. Add pcre_info(), to get the number of identifying subpatterns, the stored
4885 options, and the first character, if set.
4887 9. Recognize C+ or C{n,m} where n >= 1 as providing a fixed starting character.
4890 Version 0.91 10-Sep-97
4891 ----------------------
4893 1. PCRE was failing to diagnose unlimited repeats of subpatterns that could
4894 match the empty string as in /(a*)*/. It was looping and ultimately crashing.
4896 2. PCRE was looping on encountering an indefinitely repeated back reference to
4897 a subpattern that had matched an empty string, e.g. /(a|)\1*/. It now does what
4898 Perl does - treats the match as successful.
4900 ****


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

  ViewVC Help
Powered by ViewVC 1.1.5