ViewVC logotype

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log

Revision 990 - (show annotations)
Sun Jul 8 16:32:22 2012 UTC (7 years, 10 months ago) by zherczeg
File size: 233124 byte(s)
Improved JIT compiler optimizations for character ranges.
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.
18 Version 8.31 06-July-2012
19 -------------------------
21 1. Fixing a wrong JIT test case and some compiler warnings.
23 2. Removed a bashism from the RunTest script.
25 3. Add a cast to pcre_exec.c to fix the warning "unary minus operator applied
26 to unsigned type, result still unsigned" that was given by an MS compiler
27 on encountering the code "-sizeof(xxx)".
29 4. Partial matching support is added to the JIT compiler.
31 5. Fixed several bugs concerned with partial matching of items that consist
32 of more than one character:
34 (a) /^(..)\1/ did not partially match "aba" because checking references was
35 done on an "all or nothing" basis. This also applied to repeated
36 references.
38 (b) \R did not give a hard partial match if \r was found at the end of the
39 subject.
41 (c) \X did not give a hard partial match after matching one or more
42 characters at the end of the subject.
44 (d) When newline was set to CRLF, a pattern such as /a$/ did not recognize
45 a partial match for the string "\r".
47 (e) When newline was set to CRLF, the metacharacter "." did not recognize
48 a partial match for a CR character at the end of the subject string.
50 6. If JIT is requested using /S++ or -s++ (instead of just /S+ or -s+) when
51 running pcretest, the text "(JIT)" added to the output whenever JIT is
52 actually used to run the match.
54 7. Individual JIT compile options can be set in pcretest by following -s+[+]
55 or /S+[+] with a digit between 1 and 7.
57 8. OP_NOT now supports any UTF character not just single-byte ones.
59 9. (*MARK) control verb is now supported by the JIT compiler.
61 10. The command "./RunTest list" lists the available tests without actually
62 running any of them. (Because I keep forgetting what they all are.)
66 12. Applied a (slightly modified) user-supplied patch that improves performance
67 when the heap is used for recursion (compiled with --disable-stack-for-
68 recursion). Instead of malloc and free for each heap frame each time a
69 logical recursion happens, frames are retained on a chain and re-used where
70 possible. This sometimes gives as much as 30% improvement.
72 13. As documented, (*COMMIT) is now confined to within a recursive subpattern
73 call.
75 14. As documented, (*COMMIT) is now confined to within a positive assertion.
77 15. It is now possible to link pcretest with libedit as an alternative to
78 libreadline.
80 16. (*COMMIT) control verb is now supported by the JIT compiler.
82 17. The Unicode data tables have been updated to Unicode 6.1.0.
84 18. Added --file-list option to pcregrep.
86 19. Added binary file support to pcregrep, including the -a, --binary-files,
87 -I, and --text options.
89 20. The madvise function is renamed for posix_madvise for QNX compatibility
90 reasons. Fixed by Giuseppe D'Angelo.
92 21. Fixed a bug for backward assertions with REVERSE 0 in the JIT compiler.
94 22. Changed the option for creating symbolic links for 16-bit man pages from
95 -s to -sf so that re-installing does not cause issues.
97 23. Support PCRE_NO_START_OPTIMIZE in JIT as (*MARK) support requires it.
99 24. Fixed a very old bug in pcretest that caused errors with restarted DFA
100 matches in certain environments (the workspace was not being correctly
101 retained). Also added to pcre_dfa_exec() a simple plausibility check on
102 some of the workspace data at the beginning of a restart.
104 25. \s*\R was auto-possessifying the \s* when it should not, whereas \S*\R
105 was not doing so when it should - probably a typo introduced by SVN 528
106 (change 8.10/14).
108 26. When PCRE_UCP was not set, \w+\x{c4} was incorrectly auto-possessifying the
109 \w+ when the character tables indicated that \x{c4} was a word character.
110 There were several related cases, all because the tests for doing a table
111 lookup were testing for characters less than 127 instead of 255.
113 27. If a pattern contains capturing parentheses that are not used in a match,
114 their slots in the ovector are set to -1. For those that are higher than
115 any matched groups, this happens at the end of processing. In the case when
116 there were back references that the ovector was too small to contain
117 (causing temporary malloc'd memory to be used during matching), and the
118 highest capturing number was not used, memory off the end of the ovector
119 was incorrectly being set to -1. (It was using the size of the temporary
120 memory instead of the true size.)
122 28. To catch bugs like 27 using valgrind, when pcretest is asked to specify an
123 ovector size, it uses memory at the end of the block that it has got.
125 29. Check for an overlong MARK name and give an error at compile time. The
126 limit is 255 for the 8-bit library and 65535 for the 16-bit library.
128 30. JIT compiler update.
130 31. JIT is now supported on jailbroken iOS devices. Thanks for Ruiger
131 Rill for the patch.
133 32. Put spaces around SLJIT_PRINT_D in the JIT compiler. Required by CXX11.
135 33. Variable renamings in the PCRE-JIT compiler. No functionality change.
137 34. Fixed typos in pcregrep: in two places there was SUPPORT_LIBZ2 instead of
138 SUPPORT_LIBBZ2. This caused a build problem when bzip2 but not gzip (zlib)
139 was enabled.
141 35. Improve JIT code generation for greedy plus quantifier.
143 36. When /((?:a?)*)*c/ or /((?>a?)*)*c/ was matched against "aac", it set group
144 1 to "aa" instead of to an empty string. The bug affected repeated groups
145 that could potentially match an empty string.
147 37. Optimizing single character iterators in JIT.
149 38. Wide characters specified with \uxxxx in JavaScript mode are now subject to
150 the same checks as \x{...} characters in non-JavaScript mode. Specifically,
151 codepoints that are too big for the mode are faulted, and in a UTF mode,
152 disallowed codepoints are also faulted.
154 39. If PCRE was compiled with UTF support, in three places in the DFA
155 matcher there was code that should only have been obeyed in UTF mode, but
156 was being obeyed unconditionally. In 8-bit mode this could cause incorrect
157 processing when bytes with values greater than 127 were present. In 16-bit
158 mode the bug would be provoked by values in the range 0xfc00 to 0xdc00. In
159 both cases the values are those that cannot be the first data item in a UTF
160 character. The three items that might have provoked this were recursions,
161 possessively repeated groups, and atomic groups.
163 40. Ensure that libpcre is explicitly listed in the link commands for pcretest
164 and pcregrep, because some OS require shared objects to be explicitly
165 passed to ld, causing the link step to fail if they are not.
167 41. There were two incorrect #ifdefs in pcre_study.c, meaning that, in 16-bit
168 mode, patterns that started with \h* or \R* might be incorrectly matched.
171 Version 8.30 04-February-2012
172 -----------------------------
174 1. Renamed "isnumber" as "is_a_number" because in some Mac environments this
175 name is defined in ctype.h.
177 2. Fixed a bug in fixed-length calculation for lookbehinds that would show up
178 only in quite long subpatterns.
180 3. Removed the function pcre_info(), which has been obsolete and deprecated
181 since it was replaced by pcre_fullinfo() in February 2000.
183 4. For a non-anchored pattern, if (*SKIP) was given with a name that did not
184 match a (*MARK), and the match failed at the start of the subject, a
185 reference to memory before the start of the subject could occur. This bug
186 was introduced by fix 17 of release 8.21.
188 5. A reference to an unset group with zero minimum repetition was giving
189 totally wrong answers (in non-JavaScript-compatibility mode). For example,
190 /(another)?(\1?)test/ matched against "hello world test". This bug was
191 introduced in release 8.13.
193 6. Add support for 16-bit character strings (a large amount of work involving
194 many changes and refactorings).
196 7. RunGrepTest failed on msys because \r\n was replaced by whitespace when the
197 command "pattern=`printf 'xxx\r\njkl'`" was run. The pattern is now taken
198 from a file.
200 8. Ovector size of 2 is also supported by JIT based pcre_exec (the ovector size
201 rounding is not applied in this particular case).
203 9. The invalid Unicode surrogate codepoints U+D800 to U+DFFF are now rejected
204 if they appear, or are escaped, in patterns.
206 10. Get rid of a number of -Wunused-but-set-variable warnings.
208 11. The pattern /(?=(*:x))(q|)/ matches an empty string, and returns the mark
209 "x". The similar pattern /(?=(*:x))((*:y)q|)/ did not return a mark at all.
210 Oddly, Perl behaves the same way. PCRE has been fixed so that this pattern
211 also returns the mark "x". This bug applied to capturing parentheses,
212 non-capturing parentheses, and atomic parentheses. It also applied to some
213 assertions.
215 12. Stephen Kelly's patch to CMakeLists.txt allows it to parse the version
216 information out of configure.ac instead of relying on pcre.h.generic, which
217 is not stored in the repository.
219 13. Applied Dmitry V. Levin's patch for a more portable method for linking with
220 -lreadline.
222 14. ZH added PCRE_CONFIG_JITTARGET; added its output to pcretest -C.
224 15. Applied Graycode's patch to put the top-level frame on the stack rather
225 than the heap when not using the stack for recursion. This gives a
226 performance improvement in many cases when recursion is not deep.
228 16. Experimental code added to "pcretest -C" to output the stack frame size.
231 Version 8.21 12-Dec-2011
232 ------------------------
234 1. Updating the JIT compiler.
236 2. JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
237 are added as well.
239 3. Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
240 PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
241 calling _pcre_jit_exec. Some extra comments are added.
243 4. (*MARK) settings inside atomic groups that do not contain any capturing
244 parentheses, for example, (?>a(*:m)), were not being passed out. This bug
245 was introduced by change 18 for 8.20.
247 5. Supporting of \x, \U and \u in JavaScript compatibility mode based on the
248 ECMA-262 standard.
250 6. Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
251 erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
252 This bug was probably introduced by change 9 of 8.13.
254 7. While fixing 6 above, I noticed that a number of other items were being
255 incorrectly rejected as "not fixed length". This arose partly because newer
256 opcodes had not been added to the fixed-length checking code. I have (a)
257 corrected the bug and added tests for these items, and (b) arranged for an
258 error to occur if an unknown opcode is encountered while checking for fixed
259 length instead of just assuming "not fixed length". The items that were
260 rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
261 (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
262 repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
264 8. A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
265 being incorrectly compiled and would have given unpredicatble results.
267 9. A possessively repeated subpattern with minimum repeat count greater than
268 one behaved incorrectly. For example, (A){2,}+ behaved as if it was
269 (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
270 the first (A) could occur when it should not.
272 10. Add a cast and remove a redundant test from the code.
274 11. JIT should use pcre_malloc/pcre_free for allocation.
276 12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
277 best practice nowadays, and helps with cross-compiling. (If the exec_prefix
278 is anything other than /usr, -L is still shown).
280 13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
282 14. Perl does not support \N without a following name in a [] class; PCRE now
283 also gives an error.
285 15. If a forward reference was repeated with an upper limit of around 2000,
286 it caused the error "internal error: overran compiling workspace". The
287 maximum number of forward references (including repeats) was limited by the
288 internal workspace, and dependent on the LINK_SIZE. The code has been
289 rewritten so that the workspace expands (via pcre_malloc) if necessary, and
290 the default depends on LINK_SIZE. There is a new upper limit (for safety)
291 of around 200,000 forward references. While doing this, I also speeded up
292 the filling in of repeated forward references.
294 16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
295 incorrectly expecting the subject to contain another "a" after the start.
297 17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
298 in the match, the SKIP should be ignored. This was not happening; instead
299 the SKIP was being treated as NOMATCH. For patterns such as
300 /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
301 tested.
303 18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
304 now much more compatible with Perl, in particular in cases where the result
305 is a non-match for a non-anchored pattern. For example, if
306 /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
307 "m", where previously it did not return a name. A side effect of this
308 change is that for partial matches, the last encountered mark name is
309 returned, as for non matches. A number of tests that were previously not
310 Perl-compatible have been moved into the Perl-compatible test files. The
311 refactoring has had the pleasing side effect of removing one argument from
312 the match() function, thus reducing its stack requirements.
314 19. If the /S+ option was used in pcretest to study a pattern using JIT,
315 subsequent uses of /S (without +) incorrectly behaved like /S+.
317 21. Retrieve executable code size support for the JIT compiler and fixing
318 some warnings.
320 22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
321 not work when the shorter character appeared right at the end of the
322 subject string.
324 23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
325 systems.
327 24. Added PCRE_INFO_JITSIZE to pass on the value from (21) above, and also
328 output it when the /M option is used in pcretest.
330 25. The CheckMan script was not being included in the distribution. Also, added
331 an explicit "perl" to run Perl scripts from the PrepareRelease script
332 because this is reportedly needed in Windows.
334 26. If study data was being save in a file and studying had not found a set of
335 "starts with" bytes for the pattern, the data written to the file (though
336 never used) was taken from uninitialized memory and so caused valgrind to
337 complain.
339 27. Updated RunTest.bat as provided by Sheri Pierce.
341 28. Fixed a possible uninitialized memory bug in pcre_jit_compile.c.
343 29. Computation of memory usage for the table of capturing group names was
344 giving an unnecessarily large value.
347 Version 8.20 21-Oct-2011
348 ------------------------
350 1. Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
351 a POSIX class. After further experiments with Perl, which convinced me that
352 Perl has bugs and confusions, a closing square bracket is no longer allowed
353 in a POSIX name. This bug also affected patterns with classes that started
354 with full stops.
356 2. If a pattern such as /(a)b|ac/ is matched against "ac", there is no
357 captured substring, but while checking the failing first alternative,
358 substring 1 is temporarily captured. If the output vector supplied to
359 pcre_exec() was not big enough for this capture, the yield of the function
360 was still zero ("insufficient space for captured substrings"). This cannot
361 be totally fixed without adding another stack variable, which seems a lot
362 of expense for a edge case. However, I have improved the situation in cases
363 such as /(a)(b)x|abc/ matched against "abc", where the return code
364 indicates that fewer than the maximum number of slots in the ovector have
365 been set.
367 3. Related to (2) above: when there are more back references in a pattern than
368 slots in the output vector, pcre_exec() uses temporary memory during
369 matching, and copies in the captures as far as possible afterwards. It was
370 using the entire output vector, but this conflicts with the specification
371 that only 2/3 is used for passing back captured substrings. Now it uses
372 only the first 2/3, for compatibility. This is, of course, another edge
373 case.
375 4. Zoltan Herczeg's just-in-time compiler support has been integrated into the
376 main code base, and can be used by building with --enable-jit. When this is
377 done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
378 runtime --no-jit option is given.
380 5. When the number of matches in a pcre_dfa_exec() run exactly filled the
381 ovector, the return from the function was zero, implying that there were
382 other matches that did not fit. The correct "exactly full" value is now
383 returned.
385 6. If a subpattern that was called recursively or as a subroutine contained
386 (*PRUNE) or any other control that caused it to give a non-standard return,
387 invalid errors such as "Error -26 (nested recursion at the same subject
388 position)" or even infinite loops could occur.
390 7. If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
391 computing the minimum length on reaching *ACCEPT, and so ended up with the
392 wrong value of 1 rather than 0. Further investigation indicates that
393 computing a minimum subject length in the presence of *ACCEPT is difficult
394 (think back references, subroutine calls), and so I have changed the code
395 so that no minimum is registered for a pattern that contains *ACCEPT.
397 8. If (*THEN) was present in the first (true) branch of a conditional group,
398 it was not handled as intended. [But see 16 below.]
400 9. Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
401 Sheri Pierce.
403 10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
404 the first byte in a match must be "a".
406 11. Change 17 for 8.13 increased the recursion depth for patterns like
407 /a(?:.)*?a/ drastically. I've improved things by remembering whether a
408 pattern contains any instances of (*THEN). If it does not, the old
409 optimizations are restored. It would be nice to do this on a per-group
410 basis, but at the moment that is not feasible.
412 12. In some environments, the output of pcretest -C is CRLF terminated. This
413 broke RunTest's code that checks for the link size. A single white space
414 character after the value is now allowed for.
416 13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
417 For "fr", it uses the Windows-specific input and output files.
419 14. If (*THEN) appeared in a group that was called recursively or as a
420 subroutine, it did not work as intended. [But see next item.]
422 15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
423 pattern fragments (but not containing any | characters). If A and B are
424 matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
425 was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
426 D. In other words, Perl considers parentheses that do not contain any |
427 characters to be part of a surrounding alternative, whereas PCRE was
428 treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
429 differently. PCRE now behaves in the same way as Perl, except in the case
430 of subroutine/recursion calls such as (?1) which have in any case always
431 been different (but PCRE had them first :-).
433 16. Related to 15 above: Perl does not treat the | in a conditional group as
434 creating alternatives. Such a group is treated in the same way as an
435 ordinary group without any | characters when processing (*THEN). PCRE has
436 been changed to match Perl's behaviour.
438 17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
439 RunGrepTest script failed.
441 18. Change 22 for version 13 caused atomic groups to use more stack. This is
442 inevitable for groups that contain captures, but it can lead to a lot of
443 stack use in large patterns. The old behaviour has been restored for atomic
444 groups that do not contain any capturing parentheses.
446 19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
447 suppress the check for a minimum subject length at run time. (If it was
448 given to pcre_exec() or pcre_dfa_exec() it did work.)
450 20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
451 fail to work when decoding hex characters in data strings in EBCDIC
452 environments.
454 21. It appears that in at least one Mac OS environment, the isxdigit() function
455 is implemented as a macro that evaluates to its argument more than once,
456 contravening the C 90 Standard (I haven't checked a later standard). There
457 was an instance in pcretest which caused it to go wrong when processing
458 \x{...} escapes in subject strings. The has been rewritten to avoid using
459 things like p++ in the argument of isxdigit().
462 Version 8.13 16-Aug-2011
463 ------------------------
465 1. The Unicode data tables have been updated to Unicode 6.0.0.
467 2. Two minor typos in pcre_internal.h have been fixed.
469 3. Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
470 pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
471 in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
473 4. There were a number of related bugs in the code for matching backrefences
474 caselessly in UTF-8 mode when codes for the characters concerned were
475 different numbers of bytes. For example, U+023A and U+2C65 are an upper
476 and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
477 (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
478 code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
479 2-byte code at the end of the subject (it thought there wasn't enough data
480 left).
482 5. Comprehensive information about what went wrong is now returned by
483 pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
484 as the output vector has at least 2 elements. The offset of the start of
485 the failing character and a reason code are placed in the vector.
487 6. When the UTF-8 string check fails for pcre_compile(), the offset that is
488 now returned is for the first byte of the failing character, instead of the
489 last byte inspected. This is an incompatible change, but I hope it is small
490 enough not to be a problem. It makes the returned offset consistent with
491 pcre_exec() and pcre_dfa_exec().
493 7. pcretest now gives a text phrase as well as the error number when
494 pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
495 failure, the offset and reason code are output.
497 8. When \R was used with a maximizing quantifier it failed to skip backwards
498 over a \r\n pair if the subsequent match failed. Instead, it just skipped
499 back over a single character (\n). This seems wrong (because it treated the
500 two characters as a single entity when going forwards), conflicts with the
501 documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
502 behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
503 has been changed.
505 9. Some internal refactoring has changed the processing so that the handling
506 of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
507 time (the PCRE_DOTALL option was changed this way some time ago: version
508 7.7 change 16). This has made it possible to abolish the OP_OPT op code,
509 which was always a bit of a fudge. It also means that there is one less
510 argument for the match() function, which reduces its stack requirements
511 slightly. This change also fixes an incompatibility with Perl: the pattern
512 (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
514 10. More internal refactoring has drastically reduced the number of recursive
515 calls to match() for possessively repeated groups such as (abc)++ when
516 using pcre_exec().
518 11. While implementing 10, a number of bugs in the handling of groups were
519 discovered and fixed:
521 (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
522 (a|)*(?1) gave a compile-time internal error.
523 ((a|)+)+ did not notice that the outer group could match an empty string.
524 (^a|^)+ was not marked as anchored.
525 (.*a|.*)+ was not marked as matching at start or after a newline.
527 12. Yet more internal refactoring has removed another argument from the match()
528 function. Special calls to this function are now indicated by setting a
529 value in a variable in the "match data" data block.
531 13. Be more explicit in pcre_study() instead of relying on "default" for
532 opcodes that mean there is no starting character; this means that when new
533 ones are added and accidentally left out of pcre_study(), testing should
534 pick them up.
536 14. The -s option of pcretest has been documented for ages as being an old
537 synonym of -m (show memory usage). I have changed it to mean "force study
538 for every regex", that is, assume /S for every regex. This is similar to -i
539 and -d etc. It's slightly incompatible, but I'm hoping nobody is still
540 using it. It makes it easier to run collections of tests with and without
541 study enabled, and thereby test pcre_study() more easily. All the standard
542 tests are now run with and without -s (but some patterns can be marked as
543 "never study" - see 20 below).
545 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
546 restoration of the capturing data to the outer values was not happening
547 correctly.
549 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
550 empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
551 pattern had matched an empty string, and so incorrectly returned a no
552 match.
554 17. There was optimizing code for the last branch of non-capturing parentheses,
555 and also for the obeyed branch of a conditional subexpression, which used
556 tail recursion to cut down on stack usage. Unfortunately, now that there is
557 the possibility of (*THEN) occurring in these branches, tail recursion is
558 no longer possible because the return has to be checked for (*THEN). These
559 two optimizations have therefore been removed. [But see 8.20/11 above.]
561 18. If a pattern containing \R was studied, it was assumed that \R always
562 matched two bytes, thus causing the minimum subject length to be
563 incorrectly computed because \R can also match just one byte.
565 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
566 was incorrectly computed.
568 20. If /S is present twice on a test pattern in pcretest input, it now
569 *disables* studying, thereby overriding the use of -s on the command line
570 (see 14 above). This is necessary for one or two tests to keep the output
571 identical in both cases.
573 21. When (*ACCEPT) was used in an assertion that matched an empty string and
574 PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
576 22. When an atomic group that contained a capturing parenthesis was
577 successfully matched, but the branch in which it appeared failed, the
578 capturing was not being forgotten if a higher numbered group was later
579 captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
580 group 1 to "a", when in fact it should be unset. This applied to multi-
581 branched capturing and non-capturing groups, repeated or not, and also to
582 positive assertions (capturing in negative assertions does not happen
583 in PCRE) and also to nested atomic groups.
585 23. Add the ++ qualifier feature to pcretest, to show the remainder of the
586 subject after a captured substring, to make it easier to tell which of a
587 number of identical substrings has been captured.
589 24. The way atomic groups are processed by pcre_exec() has been changed so that
590 if they are repeated, backtracking one repetition now resets captured
591 values correctly. For example, if ((?>(a+)b)+aabab) is matched against
592 "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
593 "aaa". Previously, it would have been "a". As part of this code
594 refactoring, the way recursive calls are handled has also been changed.
596 25. If an assertion condition captured any substrings, they were not passed
597 back unless some other capturing happened later. For example, if
598 (?(?=(a))a) was matched against "a", no capturing was returned.
600 26. When studying a pattern that contained subroutine calls or assertions,
601 the code for finding the minimum length of a possible match was handling
602 direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
603 group 1 called group 2 while simultaneously a separate group 2 called group
604 1). A stack overflow occurred in this case. I have fixed this by limiting
605 the recursion depth to 10.
607 27. Updated RunTest.bat in the distribution to the version supplied by Tom
608 Fortmann. This supports explicit test numbers on the command line, and has
609 argument validation and error reporting.
611 28. An instance of \X with an unlimited repeat could fail if at any point the
612 first character it looked at was a mark character.
614 29. Some minor code refactoring concerning Unicode properties and scripts
615 should reduce the stack requirement of match() slightly.
617 30. Added the '=' option to pcretest to check the setting of unused capturing
618 slots at the end of the pattern, which are documented as being -1, but are
619 not included in the return count.
621 31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
622 compiled something random. Now it gives a compile-time error (as does
623 Perl).
625 32. A *MARK encountered during the processing of a positive assertion is now
626 recorded and passed back (compatible with Perl).
628 33. If --only-matching or --colour was set on a pcregrep call whose pattern
629 had alternative anchored branches, the search for a second match in a line
630 was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
631 matched the line "0102" twice. The same bug affected patterns that started
632 with a backwards assertion. For example /\b01|\b02/ also matched "0102"
633 twice.
635 34. Previously, PCRE did not allow quantification of assertions. However, Perl
636 does, and because of capturing effects, quantifying parenthesized
637 assertions may at times be useful. Quantifiers are now allowed for
638 parenthesized assertions.
640 35. A minor code tidy in pcre_compile() when checking options for \R usage.
642 36. \g was being checked for fancy things in a character class, when it should
643 just be a literal "g".
645 37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
646 appearance of a nested POSIX class supersedes an apparent external class.
647 For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
648 unescaped square brackets may also appear as part of class names. For
649 example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
650 more like Perl. (But see 8.20/1 above.)
652 38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
653 was because it thought it was \N{name}, which is not supported).
655 39. Add minix to OS list not supporting the -S option in pcretest.
657 40. PCRE tries to detect cases of infinite recursion at compile time, but it
658 cannot analyze patterns in sufficient detail to catch mutual recursions
659 such as ((?1))((?2)). There is now a runtime test that gives an error if a
660 subgroup is called recursively as a subpattern for a second time at the
661 same position in the subject string. In previous releases this might have
662 been caught by the recursion limit, or it might have run out of stack.
664 41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
665 happen only once. PCRE was, however incorrectly giving a compile time error
666 "recursive call could loop indefinitely" because it cannot analyze the
667 pattern in sufficient detail. The compile time test no longer happens when
668 PCRE is compiling a conditional subpattern, but actual runaway loops are
669 now caught at runtime (see 40 above).
671 42. It seems that Perl allows any characters other than a closing parenthesis
672 to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
673 has been changed to be the same.
675 43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
676 as not to get warnings when autogen.sh is called. Also changed
677 AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
679 44. To help people who use pcregrep to scan files containing exceedingly long
680 lines, the following changes have been made:
682 (a) The default value of the buffer size parameter has been increased from
683 8K to 20K. (The actual buffer used is three times this size.)
685 (b) The default can be changed by ./configure --with-pcregrep-bufsize when
686 PCRE is built.
688 (c) A --buffer-size=n option has been added to pcregrep, to allow the size
689 to be set at run time.
691 (d) Numerical values in pcregrep options can be followed by K or M, for
692 example --buffer-size=50K.
694 (e) If a line being scanned overflows pcregrep's buffer, an error is now
695 given and the return code is set to 2.
697 45. Add a pointer to the latest mark to the callout data block.
699 46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
700 partial match of an empty string instead of no match. This was specific to
701 the use of ".".
703 47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
704 complete match instead of a partial match. This bug was dependent on both
705 the PCRE_UTF8 and PCRE_DOTALL options being set.
707 48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
708 starting byte set, because \b was not being ignored.
711 Version 8.12 15-Jan-2011
712 ------------------------
714 1. Fixed some typos in the markup of the man pages, and wrote a script that
715 checks for such things as part of the documentation building process.
717 2. On a big-endian 64-bit system, pcregrep did not correctly process the
718 --match-limit and --recursion-limit options (added for 8.11). In
719 particular, this made one of the standard tests fail. (The integer value
720 went into the wrong half of a long int.)
722 3. If the --colour option was given to pcregrep with -v (invert match), it
723 did strange things, either producing crazy output, or crashing. It should,
724 of course, ignore a request for colour when reporting lines that do not
725 match.
727 4. Another pcregrep bug caused similar problems if --colour was specified with
728 -M (multiline) and the pattern match finished with a line ending.
730 5. In pcregrep, when a pattern that ended with a literal newline sequence was
731 matched in multiline mode, the following line was shown as part of the
732 match. This seems wrong, so I have changed it.
734 6. Another pcregrep bug in multiline mode, when --colour was specified, caused
735 the check for further matches in the same line (so they could be coloured)
736 to overrun the end of the current line. If another match was found, it was
737 incorrectly shown (and then shown again when found in the next line).
739 7. If pcregrep was compiled under Windows, there was a reference to the
740 function pcregrep_exit() before it was defined. I am assuming this was
741 the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
742 reported by a user. I've moved the definition above the reference.
745 Version 8.11 10-Dec-2010
746 ------------------------
748 1. (*THEN) was not working properly if there were untried alternatives prior
749 to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
750 backtracked to try for "b" instead of moving to the next alternative branch
751 at the same level (in this case, to look for "c"). The Perl documentation
752 is clear that when (*THEN) is backtracked onto, it goes to the "next
753 alternative in the innermost enclosing group".
755 2. (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
756 such as (A(*COMMIT)B(*THEN)C|D) any failure after matching A should
757 result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
758 (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
759 (*THEN).
761 3. If \s appeared in a character class, it removed the VT character from
762 the class, even if it had been included by some previous item, for example
763 in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
764 of \s, but is part of the POSIX "space" class.)
766 4. A partial match never returns an empty string (because you can always
767 match an empty string at the end of the subject); however the checking for
768 an empty string was starting at the "start of match" point. This has been
769 changed to the "earliest inspected character" point, because the returned
770 data for a partial match starts at this character. This means that, for
771 example, /(?<=abc)def/ gives a partial match for the subject "abc"
772 (previously it gave "no match").
774 5. Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
775 of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
776 previously a full match would be given. However, setting PCRE_PARTIAL_HARD
777 has an implication that the given string is incomplete (because a partial
778 match is preferred over a full match). For this reason, these items now
779 give a partial match in this situation. [Aside: previously, the one case
780 /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
781 match rather than a full match, which was wrong by the old rules, but is
782 now correct.]
784 6. There was a bug in the handling of #-introduced comments, recognized when
785 PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
786 If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
787 UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
788 scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
789 but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
790 places in pcre_compile().
792 7. Related to (6) above, when pcre_compile() was skipping #-introduced
793 comments when looking ahead for named forward references to subpatterns,
794 the only newline sequence it recognized was NL. It now handles newlines
795 according to the set newline convention.
797 8. SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
798 former, but used strtoul(), whereas pcretest avoided strtoul() but did not
799 cater for a lack of strerror(). These oversights have been fixed.
801 9. Added --match-limit and --recursion-limit to pcregrep.
803 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
805 11. When the -o option was used, pcregrep was setting a return code of 1, even
806 when matches were found, and --line-buffered was not being honoured.
808 12. Added an optional parentheses number to the -o and --only-matching options
809 of pcregrep.
811 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
812 can match an empty string. The code to do it in pcretest and pcredemo
813 needed fixing:
815 (a) When the newline convention was "crlf", pcretest got it wrong, skipping
816 only one byte after an empty string match just before CRLF (this case
817 just got forgotten; "any" and "anycrlf" were OK).
819 (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
820 mode when an empty string match preceded an ASCII character followed by
821 a non-ASCII character. (The code for advancing by one character rather
822 than one byte was nonsense.)
824 (c) The pcredemo.c sample program did not have any code at all to handle
825 the cases when CRLF is a valid newline sequence.
827 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
828 as a starting offset was within the subject string. There is now a new
829 error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
830 negative or greater than the length of the string. In order to test this,
831 pcretest is extended to allow the setting of negative starting offsets.
833 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
834 starting offset points to the beginning of a UTF-8 character was
835 unnecessarily clumsy. I tidied it up.
837 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
838 bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
840 17. Nobody had reported that the --include_dir option, which was added in
841 release 7.7 should have been called --include-dir (hyphen, not underscore)
842 for compatibility with GNU grep. I have changed it to --include-dir, but
843 left --include_dir as an undocumented synonym, and the same for
844 --exclude-dir, though that is not available in GNU grep, at least as of
845 release 2.5.4.
847 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
848 characters from a string of bytes) have been redefined so as not to use
849 loops, in order to improve performance in some environments. At the same
850 time, I abstracted some of the common code into auxiliary macros to save
851 repetition (this should not affect the compiled code).
853 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
854 compile-time error is now given if \c is not followed by an ASCII
855 character, that is, a byte less than 128. (In EBCDIC mode, the code is
856 different, and any byte value is allowed.)
858 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
859 START_OPTIMIZE option, which is now allowed at compile time - but just
860 passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
861 to pcregrep and other applications that have no direct access to PCRE
862 options. The new /Y option in pcretest sets this option when calling
863 pcre_compile().
865 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
866 back references. Groups containing recursive back references were forced to
867 be atomic by that change, but in the case of named groups, the amount of
868 memory required was incorrectly computed, leading to "Failed: internal
869 error: code overflow". This has been fixed.
871 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
872 pcretest.c, to avoid build problems in some Borland environments.
875 Version 8.10 25-Jun-2010
876 ------------------------
878 1. Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
879 THEN.
881 2. (*ACCEPT) was not working when inside an atomic group.
883 3. Inside a character class, \B is treated as a literal by default, but
884 faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
885 causes the error). The code is unchanged, but I tidied the documentation.
887 4. Inside a character class, PCRE always treated \R and \X as literals,
888 whereas Perl faults them if its -w option is set. I have changed PCRE so
889 that it faults them when PCRE_EXTRA is set.
891 5. Added support for \N, which always matches any character other than
892 newline. (It is the same as "." when PCRE_DOTALL is not set.)
894 6. When compiling pcregrep with newer versions of gcc which may have
895 FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
896 declared with attribute warn_unused_result" were given. Just casting the
897 result to (void) does not stop the warnings; a more elaborate fudge is
898 needed. I've used a macro to implement this.
900 7. Minor change to pcretest.c to avoid a compiler warning.
902 8. Added four artifical Unicode properties to help with an option to make
903 \s etc use properties (see next item). The new properties are: Xan
904 (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
906 9. Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
907 use Unicode properties. (*UCP) at the start of a pattern can be used to set
908 this option. Modified pcretest to add /W to test this facility. Added
909 REG_UCP to make it available via the POSIX interface.
911 10. Added --line-buffered to pcregrep.
913 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
914 studied, and the match started with a letter with a code point greater than
915 127 whose first byte was different to the first byte of the other case of
916 the letter, the other case of this starting letter was not recognized
917 (#976).
919 12. If a pattern that was studied started with a repeated Unicode property
920 test, for example, \p{Nd}+, there was the theoretical possibility of
921 setting up an incorrect bitmap of starting bytes, but fortunately it could
922 not have actually happened in practice until change 8 above was made (it
923 added property types that matched character-matching opcodes).
925 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
926 possible starting bytes for non-anchored patterns.
928 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
929 \R, and also a number of cases that involve Unicode properties, both
930 explicit and implicit when PCRE_UCP is set.
932 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
933 input, it could crash or give wrong results if characters with values
934 greater than 0xc0 were present in the subject string. (Detail: it assumed
935 UTF-8 input when processing these items.)
937 16. Added a lot of (int) casts to avoid compiler warnings in systems where
938 size_t is 64-bit (#991).
940 17. Added a check for running out of memory when PCRE is compiled with
941 --disable-stack-for-recursion (#990).
943 18. If the last data line in a file for pcretest does not have a newline on
944 the end, a newline was missing in the output.
946 19. The default pcre_chartables.c file recognizes only ASCII characters (values
947 less than 128) in its various bitmaps. However, there is a facility for
948 generating tables according to the current locale when PCRE is compiled. It
949 turns out that in some environments, 0x85 and 0xa0, which are Unicode space
950 characters, are recognized by isspace() and therefore were getting set in
951 these tables, and indeed these tables seem to approximate to ISO 8859. This
952 caused a problem in UTF-8 mode when pcre_study() was used to create a list
953 of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
954 which of course cannot start UTF-8 characters. I have changed the code so
955 that only real ASCII characters (less than 128) and the correct starting
956 bytes for UTF-8 encodings are set for characters greater than 127 when in
957 UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
958 altogether.)
960 20. Added the /T option to pcretest so as to be able to run tests with non-
961 standard character tables, thus making it possible to include the tests
962 used for 19 above in the standard set of tests.
964 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
965 reference to a subpattern the other side of a comment that contains an
966 opening parenthesis caused either an internal compiling error, or a
967 reference to the wrong subpattern.
970 Version 8.02 19-Mar-2010
971 ------------------------
973 1. The Unicode data tables have been updated to Unicode 5.2.0.
975 2. Added the option --libs-cpp to pcre-config, but only when C++ support is
976 configured.
978 3. Updated the licensing terms in the pcregexp.pas file, as agreed with the
979 original author of that file, following a query about its status.
981 4. On systems that do not have stdint.h (e.g. Solaris), check for and include
982 inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
984 5. A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
985 quantifier applied to a forward-referencing subroutine call, could compile
986 incorrect code or give the error "internal error: previously-checked
987 referenced subpattern not found".
989 6. Both MS Visual Studio and Symbian OS have problems with initializing
990 variables to point to external functions. For these systems, therefore,
991 pcre_malloc etc. are now initialized to local functions that call the
992 relevant global functions.
994 7. There were two entries missing in the vectors called coptable and poptable
995 in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
996 I've fixed the data, and added a kludgy way of testing at compile time that
997 the lengths are correct (equal to the number of opcodes).
999 8. Following on from 7, I added a similar kludge to check the length of the
1000 eint vector in pcreposix.c.
1002 9. Error texts for pcre_compile() are held as one long string to avoid too
1003 much relocation at load time. To find a text, the string is searched,
1004 counting zeros. There was no check for running off the end of the string,
1005 which could happen if a new error number was added without updating the
1006 string.
1008 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
1010 11. \K was not working if it appeared in an atomic group or in a group that
1011 was called as a "subroutine", or in an assertion. Perl 5.11 documents that
1012 \K is "not well defined" if used in an assertion. PCRE now accepts it if
1013 the assertion is positive, but not if it is negative.
1015 12. Change 11 fortuitously reduced the size of the stack frame used in the
1016 "match()" function of pcre_exec.c by one pointer. Forthcoming
1017 implementation of support for (*MARK) will need an extra pointer on the
1018 stack; I have reserved it now, so that the stack frame size does not
1019 decrease.
1021 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
1022 item in branch that calls a recursion is a subroutine call - as in the
1023 second branch in the above example - was incorrectly given the compile-
1024 time error "recursive call could loop indefinitely" because pcre_compile()
1025 was not correctly checking the subroutine for matching a non-empty string.
1027 14. The checks for overrunning compiling workspace could trigger after an
1028 overrun had occurred. This is a "should never occur" error, but it can be
1029 triggered by pathological patterns such as hundreds of nested parentheses.
1030 The checks now trigger 100 bytes before the end of the workspace.
1032 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
1035 Version 8.01 19-Jan-2010
1036 ------------------------
1038 1. If a pattern contained a conditional subpattern with only one branch (in
1039 particular, this includes all (*DEFINE) patterns), a call to pcre_study()
1040 computed the wrong minimum data length (which is of course zero for such
1041 subpatterns). This could cause incorrect "no match" results.
1043 2. For patterns such as (?i)a(?-i)b|c where an option setting at the start of
1044 the pattern is reset in the first branch, pcre_compile() failed with
1045 "internal error: code overflow at offset...". This happened only when
1046 the reset was to the original external option setting. (An optimization
1047 abstracts leading options settings into an external setting, which was the
1048 cause of this.)
1050 3. A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
1051 of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
1052 assertion pattern did not match (meaning that the assertion was true), it
1053 was incorrectly treated as false if the SKIP had been reached during the
1054 matching. This also applied to assertions used as conditions.
1056 4. If an item that is not supported by pcre_dfa_exec() was encountered in an
1057 assertion subpattern, including such a pattern used as a condition,
1058 unpredictable results occurred, instead of the error return
1061 5. The C++ GlobalReplace function was not working like Perl for the special
1062 situation when an empty string is matched. It now does the fancy magic
1063 stuff that is necessary.
1065 6. In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
1066 removed. (These were left over from very, very early versions of PCRE.)
1068 7. Some cosmetic changes to the code to make life easier when compiling it
1069 as part of something else:
1071 (a) Change DEBUG to PCRE_DEBUG.
1073 (b) In pcre_compile(), rename the member of the "branch_chain" structure
1074 called "current" as "current_branch", to prevent a collision with the
1075 Linux macro when compiled as a kernel module.
1077 (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
1078 prevent a collision with the Linux macro when compiled as a kernel
1079 module.
1081 8. In pcre_compile() there are some checks for integer overflows that used to
1082 cast potentially large values to (double). This has been changed to that
1083 when building, a check for int64_t is made, and if it is found, it is used
1084 instead, thus avoiding the use of floating point arithmetic. (There is no
1085 other use of FP in PCRE.) If int64_t is not found, the fallback is to
1086 double.
1088 9. Added two casts to avoid signed/unsigned warnings from VS Studio Express
1089 2005 (difference between two addresses compared to an unsigned value).
1091 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
1092 custom one, because of the following reported problem in Windows:
1094 - libbz2 uses the Pascal calling convention (WINAPI) for the functions
1095 under Win32.
1096 - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
1097 therefore missing the function definition.
1098 - The compiler thus generates a "C" signature for the test function.
1099 - The linker fails to find the "C" function.
1100 - PCRE fails to configure if asked to do so against libbz2.
1102 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
1103 messages were output:
1105 Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
1106 rerunning libtoolize, to keep the correct libtool macros in-tree.
1107 Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
1109 I have done both of these things.
1111 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
1112 most of the time, it *can* run out if it is given a pattern that contains a
1113 runaway infinite recursion. I updated the discussion in the pcrestack man
1114 page.
1116 13. Now that we have gone to the x.xx style of version numbers, the minor
1117 version may start with zero. Using 08 or 09 is a bad idea because users
1118 might check the value of PCRE_MINOR in their code, and 08 or 09 may be
1119 interpreted as invalid octal numbers. I've updated the previous comment in
1120 configure.ac, and also added a check that gives an error if 08 or 09 are
1121 used.
1123 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
1124 causing partial matching to fail when the end of the subject matched \W
1125 in a UTF-8 pattern where \W was quantified with a minimum of 3.
1127 15. There were some discrepancies between the declarations in pcre_internal.h
1128 of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
1129 their definitions. The declarations used "const uschar *" and the
1130 definitions used USPTR. Even though USPTR is normally defined as "const
1131 unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
1132 reported that: "This difference in casting confuses some C++ compilers, for
1133 example, SunCC recognizes above declarations as different functions and
1134 generates broken code for hbpcre." I have changed the declarations to use
1135 USPTR.
1137 16. GNU libtool is named differently on some systems. The autogen.sh script now
1138 tries several variants such as glibtoolize (MacOSX) and libtoolize1x
1139 (FreeBSD).
1141 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
1142 (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
1143 comment: "Figure out how to create a longlong from a string: strtoll and
1144 equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
1145 instance, but it only takes 2 args instead of 3!"
1147 18. A subtle bug concerned with back references has been fixed by a change of
1148 specification, with a corresponding code fix. A pattern such as
1149 ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
1150 refers, was giving matches when it shouldn't. For example, xa=xaaa would
1151 match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
1152 same bug. Such groups have to be quantified to be useful, or contained
1153 inside another quantified group. (If there's no repetition, the reference
1154 can never match.) The problem arises because, having left the group and
1155 moved on to the rest of the pattern, a later failure that backtracks into
1156 the group uses the captured value from the final iteration of the group
1157 rather than the correct earlier one. I have fixed this in PCRE by forcing
1158 any group that contains a reference to itself to be an atomic group; that
1159 is, there cannot be any backtracking into it once it has completed. This is
1160 similar to recursive and subroutine calls.
1163 Version 8.00 19-Oct-09
1164 ----------------------
1166 1. The table for translating pcre_compile() error codes into POSIX error codes
1167 was out-of-date, and there was no check on the pcre_compile() error code
1168 being within the table. This could lead to an OK return being given in
1169 error.
1171 2. Changed the call to open a subject file in pcregrep from fopen(pathname,
1172 "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
1173 in a Windows environment.
1175 3. The pcregrep --count option prints the count for each file even when it is
1176 zero, as does GNU grep. However, pcregrep was also printing all files when
1177 --files-with-matches was added. Now, when both options are given, it prints
1178 counts only for those files that have at least one match. (GNU grep just
1179 prints the file name in this circumstance, but including the count seems
1180 more useful - otherwise, why use --count?) Also ensured that the
1181 combination -clh just lists non-zero counts, with no names.
1183 4. The long form of the pcregrep -F option was incorrectly implemented as
1184 --fixed_strings instead of --fixed-strings. This is an incompatible change,
1185 but it seems right to fix it, and I didn't think it was worth preserving
1186 the old behaviour.
1188 5. The command line items --regex=pattern and --regexp=pattern were not
1189 recognized by pcregrep, which required --regex pattern or --regexp pattern
1190 (with a space rather than an '='). The man page documented the '=' forms,
1191 which are compatible with GNU grep; these now work.
1193 6. No libpcreposix.pc file was created for pkg-config; there was just
1194 libpcre.pc and libpcrecpp.pc. The omission has been rectified.
1196 7. Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
1197 when UCP support is not needed, by modifying the Python script that
1198 generates it from Unicode data files. This should not matter if the module
1199 is correctly used as a library, but I received one complaint about 50K of
1200 unwanted data. My guess is that the person linked everything into his
1201 program rather than using a library. Anyway, it does no harm.
1203 8. A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
1204 was a minimum greater than 1 for a wide character in a possessive
1205 repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
1206 which had an unlimited repeat of a nested, fixed maximum repeat of a wide
1207 character. Chaos in the form of incorrect output or a compiling loop could
1208 result.
1210 9. The restrictions on what a pattern can contain when partial matching is
1211 requested for pcre_exec() have been removed. All patterns can now be
1212 partially matched by this function. In addition, if there are at least two
1213 slots in the offset vector, the offset of the earliest inspected character
1214 for the match and the offset of the end of the subject are set in them when
1215 PCRE_ERROR_PARTIAL is returned.
1217 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
1218 synonymous with PCRE_PARTIAL, for backwards compatibility, and
1219 PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
1220 and may be more useful for multi-segment matching.
1222 11. Partial matching with pcre_exec() is now more intuitive. A partial match
1223 used to be given if ever the end of the subject was reached; now it is
1224 given only if matching could not proceed because another character was
1225 needed. This makes a difference in some odd cases such as Z(*FAIL) with the
1226 string "Z", which now yields "no match" instead of "partial match". In the
1227 case of pcre_dfa_exec(), "no match" is given if every matching path for the
1228 final character ended with (*FAIL).
1230 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
1231 if the pattern had a "must contain" character that was already found in the
1232 earlier partial match, unless partial matching was again requested. For
1233 example, with the pattern /dog.(body)?/, the "must contain" character is
1234 "g". If the first part-match was for the string "dog", restarting with
1235 "sbody" failed. This bug has been fixed.
1237 13. The string returned by pcre_dfa_exec() after a partial match has been
1238 changed so that it starts at the first inspected character rather than the
1239 first character of the match. This makes a difference only if the pattern
1240 starts with a lookbehind assertion or \b or \B (\K is not supported by
1241 pcre_dfa_exec()). It's an incompatible change, but it makes the two
1242 matching functions compatible, and I think it's the right thing to do.
1244 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
1245 so that the demonstration program is easily available in environments where
1246 PCRE has not been installed from source.
1248 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
1249 libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
1250 library.
1252 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
1253 It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
1254 is not the first non-POSIX option to be added. Clearly some people find
1255 these options useful.
1257 17. If a caller to the POSIX matching function regexec() passes a non-zero
1258 value for nmatch with a NULL value for pmatch, the value of
1259 nmatch is forced to zero.
1261 18. RunGrepTest did not have a test for the availability of the -u option of
1262 the diff command, as RunTest does. It now checks in the same way as
1263 RunTest, and also checks for the -b option.
1265 19. If an odd number of negated classes containing just a single character
1266 interposed, within parentheses, between a forward reference to a named
1267 subpattern and the definition of the subpattern, compilation crashed with
1268 an internal error, complaining that it could not find the referenced
1269 subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
1270 [The bug was that it was starting one character too far in when skipping
1271 over the character class, thus treating the ] as data rather than
1272 terminating the class. This meant it could skip too much.]
1274 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
1275 /g option in pcretest when the pattern contains \K, which makes it possible
1276 to have an empty string match not at the start, even when the pattern is
1277 anchored. Updated pcretest and pcredemo to use this option.
1279 21. If the maximum number of capturing subpatterns in a recursion was greater
1280 than the maximum at the outer level, the higher number was returned, but
1281 with unset values at the outer level. The correct (outer level) value is
1282 now given.
1284 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
1285 PCRE did not set those parentheses (unlike Perl). I have now found a way to
1286 make it do so. The string so far is captured, making this feature
1287 compatible with Perl.
1289 23. The tests have been re-organized, adding tests 11 and 12, to make it
1290 possible to check the Perl 5.10 features against Perl 5.10.
1292 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
1293 pattern matches a fixed length string. PCRE did not allow this; now it
1294 does. Neither allows recursion.
1296 25. I finally figured out how to implement a request to provide the minimum
1297 length of subject string that was needed in order to match a given pattern.
1298 (It was back references and recursion that I had previously got hung up
1299 on.) This code has now been added to pcre_study(); it finds a lower bound
1300 to the length of subject needed. It is not necessarily the greatest lower
1301 bound, but using it to avoid searching strings that are too short does give
1302 some useful speed-ups. The value is available to calling programs via
1303 pcre_fullinfo().
1305 26. While implementing 25, I discovered to my embarrassment that pcretest had
1306 not been passing the result of pcre_study() to pcre_dfa_exec(), so the
1307 study optimizations had never been tested with that matching function.
1308 Oops. What is worse, even when it was passed study data, there was a bug in
1309 pcre_dfa_exec() that meant it never actually used it. Double oops. There
1310 were also very few tests of studied patterns with pcre_dfa_exec().
1312 27. If (?| is used to create subpatterns with duplicate numbers, they are now
1313 allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
1314 on the other side of the coin, they are no longer allowed to have different
1315 names, because these cannot be distinguished in PCRE, and this has caused
1316 confusion. (This is a difference from Perl.)
1318 28. When duplicate subpattern names are present (necessarily with different
1319 numbers, as required by 27 above), and a test is made by name in a
1320 conditional pattern, either for a subpattern having been matched, or for
1321 recursion in such a pattern, all the associated numbered subpatterns are
1322 tested, and the overall condition is true if the condition is true for any
1323 one of them. This is the way Perl works, and is also more like the way
1324 testing by number works.
1327 Version 7.9 11-Apr-09
1328 ---------------------
1330 1. When building with support for bzlib/zlib (pcregrep) and/or readline
1331 (pcretest), all targets were linked against these libraries. This included
1332 libpcre, libpcreposix, and libpcrecpp, even though they do not use these
1333 libraries. This caused unwanted dependencies to be created. This problem
1334 has been fixed, and now only pcregrep is linked with bzlib/zlib and only
1335 pcretest is linked with readline.
1337 2. The "typedef int BOOL" in pcre_internal.h that was included inside the
1338 "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
1339 moved outside it again, because FALSE and TRUE are already defined in AIX,
1340 but BOOL is not.
1342 3. The pcre_config() function was treating the PCRE_MATCH_LIMIT and
1343 PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
1345 4. The pcregrep documentation said spaces were inserted as well as colons (or
1346 hyphens) following file names and line numbers when outputting matching
1347 lines. This is not true; no spaces are inserted. I have also clarified the
1348 wording for the --colour (or --color) option.
1350 5. In pcregrep, when --colour was used with -o, the list of matching strings
1351 was not coloured; this is different to GNU grep, so I have changed it to be
1352 the same.
1354 6. When --colo(u)r was used in pcregrep, only the first matching substring in
1355 each matching line was coloured. Now it goes on to look for further matches
1356 of any of the test patterns, which is the same behaviour as GNU grep.
1358 7. A pattern that could match an empty string could cause pcregrep to loop; it
1359 doesn't make sense to accept an empty string match in pcregrep, so I have
1360 locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
1361 seems to be how GNU grep behaves.
1363 8. The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
1364 start or after a newline", because the conditional assertion was not being
1365 correctly handled. The rule now is that both the assertion and what follows
1366 in the first alternative must satisfy the test.
1368 9. If auto-callout was enabled in a pattern with a conditional group whose
1369 condition was an assertion, PCRE could crash during matching, both with
1370 pcre_exec() and pcre_dfa_exec().
1372 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
1373 used for matching.
1375 11. Unicode property support in character classes was not working for
1376 characters (bytes) greater than 127 when not in UTF-8 mode.
1378 12. Added the -M command line option to pcretest.
1380 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
1382 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
1384 16. Added comments and documentation about mis-use of no_arg in the C++
1385 wrapper.
1387 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
1388 from Martin Jerabek that uses macro names for all relevant character and
1389 string constants.
1391 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
1392 SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
1393 SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
1394 these, but not everybody uses configure.
1396 19. A conditional group that had only one branch was not being correctly
1397 recognized as an item that could match an empty string. This meant that an
1398 enclosing group might also not be so recognized, causing infinite looping
1399 (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
1400 with the subject "ab", where knowledge that the repeated group can match
1401 nothing is needed in order to break the loop.
1403 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
1404 exec(), but without supplying a callout function, matching went wrong.
1406 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
1407 leak if the size of the offset vector was greater than 30. When the vector
1408 is smaller, the saved offsets during recursion go onto a local stack
1409 vector, but for larger vectors malloc() is used. It was failing to free
1410 when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
1411 error, in fact).
1413 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
1414 heapframe that is used only when UTF-8 support is enabled. This caused no
1415 problem, but was untidy.
1417 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
1418 CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
1419 included within another project.
1421 24. Steven Van Ingelgem's patches to add more options to the CMake support,
1422 slightly modified by me:
1424 (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
1425 not building pcregrep.
1427 (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
1428 if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
1430 25. Forward references, both numeric and by name, in patterns that made use of
1431 duplicate group numbers, could behave incorrectly or give incorrect errors,
1432 because when scanning forward to find the reference group, PCRE was not
1433 taking into account the duplicate group numbers. A pattern such as
1434 ^X(?3)(a)(?|(b)|(q))(Y) is an example.
1436 26. Changed a few more instances of "const unsigned char *" to USPTR, making
1437 the feature of a custom pointer more persuasive (as requested by a user).
1439 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
1440 pcretest.c, inside #ifndefs, because it seems they are sometimes already
1441 pre-defined.
1443 28. Added support for (*UTF8) at the start of a pattern.
1445 29. Arrange for flags added by the "release type" setting in CMake to be shown
1446 in the configuration summary.
1449 Version 7.8 05-Sep-08
1450 ---------------------
1452 1. Replaced UCP searching code with optimized version as implemented for Ad
1453 Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
1454 stage table and inline lookup instead of a function, giving speed ups of 2
1455 to 5 times on some simple patterns that I tested. Permission was given to
1456 distribute the MultiStage2.py script that generates the tables (it's not in
1457 the tarball, but is in the Subversion repository).
1459 2. Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
1460 scripts.
1462 3. Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
1463 a group with a zero qualifier. The result of the study could be incorrect,
1464 or the function might crash, depending on the pattern.
1466 4. Caseless matching was not working for non-ASCII characters in back
1467 references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
1468 It now works when Unicode Property Support is available.
1470 5. In pcretest, an escape such as \x{de} in the data was always generating
1471 a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
1472 non-UTF-8 mode. If the value is greater than 255, it gives a warning about
1473 truncation.
1475 6. Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
1477 7. Added two (int) casts to pcregrep when printing the difference of two
1478 pointers, in case they are 64-bit values.
1480 8. Added comments about Mac OS X stack usage to the pcrestack man page and to
1481 test 2 if it fails.
1483 9. Added PCRE_CALL_CONVENTION just before the names of all exported functions,
1484 and a #define of that name to empty if it is not externally set. This is to
1485 allow users of MSVC to set it if necessary.
1487 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
1488 the convenience functions in the pcre_get.c source file.
1490 11. An option change at the start of a pattern that had top-level alternatives
1491 could cause overwriting and/or a crash. This command provoked a crash in
1492 some environments:
1494 printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
1496 This potential security problem was recorded as CVE-2008-2371.
1498 12. For a pattern where the match had to start at the beginning or immediately
1499 after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
1500 pcre_dfa_exec() could read past the end of the passed subject if there was
1501 no match. To help with detecting such bugs (e.g. with valgrind), I modified
1502 pcretest so that it places the subject at the end of its malloc-ed buffer.
1504 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
1505 exec() might read past the end of the data buffer in UTF-8 mode.
1507 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
1508 the data contained the byte 0x85 as part of a UTF-8 character within its
1509 first line. This applied both to normal and DFA matching.
1511 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
1512 /^[^d]*?$/8 failed to match "abc".
1514 16. Added a missing copyright notice to pcrecpp_internal.h.
1516 17. Make it more clear in the documentation that values returned from
1517 pcre_exec() in ovector are byte offsets, not character counts.
1519 18. Tidied a few places to stop certain compilers from issuing warnings.
1521 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
1522 supplied by Stefan Weber. I made a further small update for 7.8 because
1523 there is a change of source arrangements: the pcre_searchfuncs.c module is
1524 replaced by pcre_ucd.c.
1527 Version 7.7 07-May-08
1528 ---------------------
1530 1. Applied Craig's patch to sort out a long long problem: "If we can't convert
1531 a string to a long long, pretend we don't even have a long long." This is
1532 done by checking for the strtoq, strtoll, and _strtoi64 functions.
1534 2. Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
1535 pre-7.6 versions, which defined a global no_arg variable instead of putting
1536 it in the RE class. (See also #8 below.)
1538 3. Remove a line of dead code, identified by coverity and reported by Nuno
1539 Lopes.
1541 4. Fixed two related pcregrep bugs involving -r with --include or --exclude:
1543 (1) The include/exclude patterns were being applied to the whole pathnames
1544 of files, instead of just to the final components.
1546 (2) If there was more than one level of directory, the subdirectories were
1547 skipped unless they satisfied the include/exclude conditions. This is
1548 inconsistent with GNU grep (and could even be seen as contrary to the
1549 pcregrep specification - which I improved to make it absolutely clear).
1550 The action now is always to scan all levels of directory, and just
1551 apply the include/exclude patterns to regular files.
1553 5. Added the --include_dir and --exclude_dir patterns to pcregrep, and used
1554 --exclude_dir in the tests to avoid scanning .svn directories.
1556 6. Applied Craig's patch to the QuoteMeta function so that it escapes the
1557 NUL character as backslash + 0 rather than backslash + NUL, because PCRE
1558 doesn't support NULs in patterns.
1560 7. Added some missing "const"s to declarations of static tables in
1561 pcre_compile.c and pcre_dfa_exec.c.
1563 8. Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
1564 caused by fix #2 above. (Subsequently also a second patch to fix the
1565 first patch. And a third patch - this was a messy problem.)
1567 9. Applied Craig's patch to remove the use of push_back().
1569 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
1570 matching function regexec().
1572 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
1573 which, however, unlike Perl's \g{...}, are subroutine calls, not back
1574 references. PCRE supports relative numbers with this syntax (I don't think
1575 Oniguruma does).
1577 12. Previously, a group with a zero repeat such as (...){0} was completely
1578 omitted from the compiled regex. However, this means that if the group
1579 was called as a subroutine from elsewhere in the pattern, things went wrong
1580 (an internal error was given). Such groups are now left in the compiled
1581 pattern, with a new opcode that causes them to be skipped at execution
1582 time.
1584 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
1585 to the way PCRE behaves:
1587 (a) A lone ] character is dis-allowed (Perl treats it as data).
1589 (b) A back reference to an unmatched subpattern matches an empty string
1590 (Perl fails the current match path).
1592 (c) A data ] in a character class must be notated as \] because if the
1593 first data character in a class is ], it defines an empty class. (In
1594 Perl it is not possible to have an empty class.) The empty class []
1595 never matches; it forces failure and is equivalent to (*FAIL) or (?!).
1596 The negative empty class [^] matches any one character, independently
1597 of the DOTALL setting.
1599 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
1600 non-existent subpattern following a character class starting with ']' and
1601 containing () gave an internal compiling error instead of "reference to
1602 non-existent subpattern". Fortunately, when the pattern did exist, the
1603 compiled code was correct. (When scanning forwards to check for the
1604 existencd of the subpattern, it was treating the data ']' as terminating
1605 the class, so got the count wrong. When actually compiling, the reference
1606 was subsequently set up correctly.)
1608 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
1609 it was being rejected as not supported by pcre_dfa_exec(), even though
1610 other assertions are supported. I have made pcre_dfa_exec() support
1611 (*FAIL).
1613 16. The implementation of 13c above involved the invention of a new opcode,
1614 OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
1615 cannot be changed at match time, I realized I could make a small
1616 improvement to matching performance by compiling OP_ALLANY instead of
1617 OP_ANY for "." when DOTALL was set, and then removing the runtime tests
1618 on the OP_ANY path.
1620 17. Compiling pcretest on Windows with readline support failed without the
1621 following two fixes: (1) Make the unistd.h include conditional on
1622 HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
1624 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
1625 ncurses library to be included for pcretest when ReadLine support is
1626 requested, but also to allow for it to be overridden. This patch came from
1627 Daniel Bergström.
1629 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
1630 as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
1631 any errors with the current Unicode tables. Thanks to Peter Kankowski for
1632 spotting this.
1635 Version 7.6 28-Jan-08
1636 ---------------------
1638 1. A character class containing a very large number of characters with
1639 codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
1640 overflow.
1642 2. Patch to cut out the "long long" test in pcrecpp_unittest when
1643 HAVE_LONG_LONG is not defined.
1645 3. Applied Christian Ehrlicher's patch to update the CMake build files to
1646 bring them up to date and include new features. This patch includes:
1648 - Fixed PH's badly added libz and libbz2 support.
1649 - Fixed a problem with static linking.
1650 - Added pcredemo. [But later removed - see 7 below.]
1651 - Fixed dftables problem and added an option.
1652 - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
1654 - Added readline support for pcretest.
1655 - Added an listing of the option settings after cmake has run.
1657 4. A user submitted a patch to Makefile that makes it easy to create
1658 "pcre.dll" under mingw when using Configure/Make. I added stuff to
1659 Makefile.am that cause it to include this special target, without
1660 affecting anything else. Note that the same mingw target plus all
1661 the other distribution libraries and programs are now supported
1662 when configuring with CMake (see 6 below) instead of with
1663 Configure/Make.
1665 5. Applied Craig's patch that moves no_arg into the RE class in the C++ code.
1666 This is an attempt to solve the reported problem "pcrecpp::no_arg is not
1667 exported in the Windows port". It has not yet been confirmed that the patch
1668 solves the problem, but it does no harm.
1670 6. Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
1671 NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
1672 with CMake, and also correct the comment about stack recursion.
1674 7. Remove the automatic building of pcredemo from the ./configure system and
1675 from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
1676 of a program that users should build themselves after PCRE is installed, so
1677 building it automatically is not really right. What is more, it gave
1678 trouble in some build environments.
1680 8. Further tidies to CMakeLists.txt from Sheri and Christian.
1683 Version 7.5 10-Jan-08
1684 ---------------------
1686 1. Applied a patch from Craig: "This patch makes it possible to 'ignore'
1687 values in parens when parsing an RE using the C++ wrapper."
1689 2. Negative specials like \S did not work in character classes in UTF-8 mode.
1690 Characters greater than 255 were excluded from the class instead of being
1691 included.
1693 3. The same bug as (2) above applied to negated POSIX classes such as
1694 [:^space:].
1696 4. PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
1697 defined or documented. It seems to have been a typo for PCRE_STATIC, so
1698 I have changed it.
1700 5. The construct (?&) was not diagnosed as a syntax error (it referenced the
1701 first named subpattern) and a construct such as (?&a) would reference the
1702 first named subpattern whose name started with "a" (in other words, the
1703 length check was missing). Both these problems are fixed. "Subpattern name
1704 expected" is now given for (?&) (a zero-length name), and this patch also
1705 makes it give the same error for \k'' (previously it complained that that
1706 was a reference to a non-existent subpattern).
1708 6. The erroneous patterns (?+-a) and (?-+a) give different error messages;
1709 this is right because (?- can be followed by option settings as well as by
1710 digits. I have, however, made the messages clearer.
1712 7. Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
1713 than the number used in the conditional) now cause a compile-time error.
1714 This is actually not compatible with Perl, which accepts such patterns, but
1715 treats the conditional as always being FALSE (as PCRE used to), but it
1716 seems to me that giving a diagnostic is better.
1718 8. Change "alphameric" to the more common word "alphanumeric" in comments
1719 and messages.
1721 9. Fix two occurrences of "backslash" in comments that should have been
1722 "backspace".
1724 10. Remove two redundant lines of code that can never be obeyed (their function
1725 was moved elsewhere).
1727 11. The program that makes PCRE's Unicode character property table had a bug
1728 which caused it to generate incorrect table entries for sequences of
1729 characters that have the same character type, but are in different scripts.
1730 It amalgamated them into a single range, with the script of the first of
1731 them. In other words, some characters were in the wrong script. There were
1732 thirteen such cases, affecting characters in the following ranges:
1734 U+002b0 - U+002c1
1735 U+0060c - U+0060d
1736 U+0061e - U+00612
1737 U+0064b - U+0065e
1738 U+0074d - U+0076d
1739 U+01800 - U+01805
1740 U+01d00 - U+01d77
1741 U+01d9b - U+01dbf
1742 U+0200b - U+0200f
1743 U+030fc - U+030fe
1744 U+03260 - U+0327f
1745 U+0fb46 - U+0fbb1
1746 U+10450 - U+1049d
1748 12. The -o option (show only the matching part of a line) for pcregrep was not
1749 compatible with GNU grep in that, if there was more than one match in a
1750 line, it showed only the first of them. It now behaves in the same way as
1751 GNU grep.
1753 13. If the -o and -v options were combined for pcregrep, it printed a blank
1754 line for every non-matching line. GNU grep prints nothing, and pcregrep now
1755 does the same. The return code can be used to tell if there were any
1756 non-matching lines.
1758 14. Added --file-offsets and --line-offsets to pcregrep.
1760 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
1761 infinitely looping recursion. The bug was that positive lookaheads were not
1762 being skipped when checking for a possible empty match (negative lookaheads
1763 and both kinds of lookbehind were skipped).
1765 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
1766 inclusion of <windows.h> to before rather than after the definition of
1767 INVALID_FILE_ATTRIBUTES (patch from David Byron).
1769 17. Specifying a possessive quantifier with a specific limit for a Unicode
1770 character property caused pcre_compile() to compile bad code, which led at
1771 runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
1772 are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
1773 caused the error; without that there was no problem.
1775 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
1777 19. Added --enable-pcretest-libreadline.
1779 20. In pcrecpp.cc, the variable 'count' was incremented twice in
1780 RE::GlobalReplace(). As a result, the number of replacements returned was
1781 double what it should be. I removed one of the increments, but Craig sent a
1782 later patch that removed the other one (the right fix) and added unit tests
1783 that check the return values (which was not done before).
1785 21. Several CMake things:
1787 (1) Arranged that, when cmake is used on Unix, the libraries end up with
1788 the names libpcre and libpcreposix, not just pcre and pcreposix.
1790 (2) The above change means that pcretest and pcregrep are now correctly
1791 linked with the newly-built libraries, not previously installed ones.
1795 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
1796 crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
1797 UTF-8 newline character). The key issue is that the pattern starts .*;
1798 this means that the match must be either at the beginning, or after a
1799 newline. The bug was in the code for advancing after a failed match and
1800 checking that the new position followed a newline. It was not taking
1801 account of UTF-8 characters correctly.
1803 23. PCRE was behaving differently from Perl in the way it recognized POSIX
1804 character classes. PCRE was not treating the sequence [:...:] as a
1805 character class unless the ... were all letters. Perl, however, seems to
1806 allow any characters between [: and :], though of course it rejects as
1807 unknown any "names" that contain non-letters, because all the known class
1808 names consist only of letters. Thus, Perl gives an error for [[:1234:]],
1809 for example, whereas PCRE did not - it did not recognize a POSIX character
1810 class. This seemed a bit dangerous, so the code has been changed to be
1811 closer to Perl. The behaviour is not identical to Perl, because PCRE will
1812 diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
1813 treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
1814 Perl does, and where it didn't before.
1816 24. Rewrite so as to remove the single use of %n from pcregrep because in some
1817 Windows environments %n is disabled by default.
1820 Version 7.4 21-Sep-07
1821 ---------------------
1823 1. Change 7.3/28 was implemented for classes by looking at the bitmap. This
1824 means that a class such as [\s] counted as "explicit reference to CR or
1825 LF". That isn't really right - the whole point of the change was to try to
1826 help when there was an actual mention of one of the two characters. So now
1827 the change happens only if \r or \n (or a literal CR or LF) character is
1828 encountered.
1830 2. The 32-bit options word was also used for 6 internal flags, but the numbers
1831 of both had grown to the point where there were only 3 bits left.
1832 Fortunately, there was spare space in the data structure, and so I have
1833 moved the internal flags into a new 16-bit field to free up more option
1834 bits.
1836 3. The appearance of (?J) at the start of a pattern set the DUPNAMES option,
1837 but did not set the internal JCHANGED flag - either of these is enough to
1838 control the way the "get" function works - but the PCRE_INFO_JCHANGED
1839 facility is supposed to tell if (?J) was ever used, so now (?J) at the
1840 start sets both bits.
1842 4. Added options (at build time, compile time, exec time) to change \R from
1843 matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
1845 5. doc/pcresyntax.html was missing from the distribution.
1847 6. Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
1848 compatibility, even though it is no longer used.
1850 7. Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
1851 strtoull to pcrecpp.cc to select the available functions in WIN32 when the
1852 windows.h file is present (where different names are used). [This was
1853 reversed later after testing - see 16 below.]
1855 8. Changed all #include <config.h> to #include "config.h". There were also
1856 some further <pcre.h> cases that I changed to "pcre.h".
1858 9. When pcregrep was used with the --colour option, it missed the line ending
1859 sequence off the lines that it output.
1861 10. It was pointed out to me that arrays of string pointers cause lots of
1862 relocations when a shared library is dynamically loaded. A technique of
1863 using a single long string with a table of offsets can drastically reduce
1864 these. I have refactored PCRE in four places to do this. The result is
1865 dramatic:
1867 Originally: 290
1868 After changing UCP table: 187
1869 After changing error message table: 43
1870 After changing table of "verbs" 36
1871 After changing table of Posix names 22
1873 Thanks to the folks working on Gregex for glib for this insight.
1875 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
1876 unicode-properties was also set.
1878 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
1880 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
1881 checked only for CRLF.
1883 14. Added casts to pcretest.c to avoid compiler warnings.
1885 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
1887 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
1888 and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
1889 entirely. This removes changes made in 7 above.
1891 17. The CMake files have been updated, and there is now more information about
1892 building with CMake in the NON-UNIX-USE document.
1895 Version 7.3 28-Aug-07
1896 ---------------------
1898 1. In the rejigging of the build system that eventually resulted in 7.1, the
1899 line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
1900 brackets there is not right, since it causes compilers to look for an
1901 installed pcre.h, not the version that is in the source that is being
1902 compiled (which of course may be different). I have changed it back to:
1904 #include "pcre.h"
1906 I have a vague recollection that the change was concerned with compiling in
1907 different directories, but in the new build system, that is taken care of
1908 by the VPATH setting the Makefile.
1910 2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
1911 when the subject happened to end in the byte 0x85 (e.g. if the last
1912 character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
1913 characters but of course it shouldn't be taken as a newline when it is part
1914 of another character. The bug was that, for an unlimited repeat of . in
1915 not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
1916 characters when looking for a newline.
1918 3. A small performance improvement in the DOTALL UTF-8 mode .* case.
1920 4. Debugging: adjusted the names of opcodes for different kinds of parentheses
1921 in debug output.
1923 5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
1924 long printing in the pcrecpp unittest when running under MinGW.
1926 6. ESC_K was left out of the EBCDIC table.
1928 7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
1929 parentheses; I made it 1000, which seemed large enough. Unfortunately, the
1930 limit also applies to "virtual nesting" when a pattern is recursive, and in
1931 this case 1000 isn't so big. I have been able to remove this limit at the
1932 expense of backing off one optimization in certain circumstances. Normally,
1933 when pcre_exec() would call its internal match() function recursively and
1934 immediately return the result unconditionally, it uses a "tail recursion"
1935 feature to save stack. However, when a subpattern that can match an empty
1936 string has an unlimited repetition quantifier, it no longer makes this
1937 optimization. That gives it a stack frame in which to save the data for
1938 checking that an empty string has been matched. Previously this was taken
1939 from the 1000-entry workspace that had been reserved. So now there is no
1940 explicit limit, but more stack is used.
1942 8. Applied Daniel's patches to solve problems with the import/export magic
1943 syntax that is required for Windows, and which was going wrong for the
1944 pcreposix and pcrecpp parts of the library. These were overlooked when this
1945 problem was solved for the main library.
1947 9. There were some crude static tests to avoid integer overflow when computing
1948 the size of patterns that contain repeated groups with explicit upper
1949 limits. As the maximum quantifier is 65535, the maximum group length was
1950 set at 30,000 so that the product of these two numbers did not overflow a
1951 32-bit integer. However, it turns out that people want to use groups that
1952 are longer than 30,000 bytes (though not repeat them that many times).
1953 Change 7.0/17 (the refactoring of the way the pattern size is computed) has
1954 made it possible to implement the integer overflow checks in a much more
1955 dynamic way, which I have now done. The artificial limitation on group
1956 length has been removed - we now have only the limit on the total length of
1957 the compiled pattern, which depends on the LINK_SIZE setting.
1959 10. Fixed a bug in the documentation for get/copy named substring when
1960 duplicate names are permitted. If none of the named substrings are set, the
1961 functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
1962 empty string.
1964 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
1965 instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
1966 because the ] is interpreted as the first data character and the
1967 terminating ] is not found. PCRE has been made compatible with Perl in this
1968 regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
1969 cause memory overwriting.
1971 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
1972 string has been matched (to stop an infinite loop). It was not recognizing
1973 a conditional subpattern that could match an empty string if that
1974 subpattern was within another subpattern. For example, it looped when
1975 trying to match (((?(1)X|))*) but it was OK with ((?(1)X|)*) where the
1976 condition was not nested. This bug has been fixed.
1978 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
1979 past the start of the subject in the presence of bytes with the top bit
1980 set, for example "\x8aBCD".
1982 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
1983 (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
1985 14. Optimized (?!) to (*FAIL).
1987 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
1988 This restricts code points to be within the range 0 to 0x10FFFF, excluding
1989 the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
1990 full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
1991 does: it's just the validity check that is more restrictive.
1993 16. Inserted checks for integer overflows during escape sequence (backslash)
1994 processing, and also fixed erroneous offset values for syntax errors during
1995 backslash processing.
1997 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
1998 for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
2000 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
2001 caused an overrun.
2003 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
2004 something other than just ASCII characters) inside a group that had an
2005 unlimited repeat caused a loop at compile time (while checking to see
2006 whether the group could match an empty string).
2008 20. Debugging a pattern containing \p or \P could cause a crash. For example,
2009 [\P{Any}] did so. (Error in the code for printing property names.)
2011 21. An orphan \E inside a character class could cause a crash.
2013 22. A repeated capturing bracket such as (A)? could cause a wild memory
2014 reference during compilation.
2016 23. There are several functions in pcre_compile() that scan along a compiled
2017 expression for various reasons (e.g. to see if it's fixed length for look
2018 behind). There were bugs in these functions when a repeated \p or \P was
2019 present in the pattern. These operators have additional parameters compared
2020 with \d, etc, and these were not being taken into account when moving along
2021 the compiled data. Specifically:
2023 (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
2024 length.
2026 (b) An item such as \pL+ within a repeated group could cause crashes or
2027 loops.
2029 (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
2030 "reference to non-existent subpattern" error.
2032 (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
2034 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
2035 characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
2037 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
2039 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
2040 character were causing crashes (broken optimization).
2042 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
2043 \p or \P) caused a compile-time loop.
2045 28. More problems have arisen in unanchored patterns when CRLF is a valid line
2046 break. For example, the unstudied pattern [\r\n]A does not match the string
2047 "\r\nA" because change 7.0/46 below moves the current point on by two
2048 characters after failing to match at the start. However, the pattern \nA
2049 *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
2050 the same is true. There doesn't seem any very clean way out of this, but
2051 what I have chosen to do makes the common cases work: PCRE now takes note
2052 of whether there can be an explicit match for \r or \n anywhere in the
2053 pattern, and if so, 7.0/46 no longer applies. As part of this change,
2054 there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
2055 pattern has explicit CR or LF references.
2057 29. Added (*CR) etc for changing newline setting at start of pattern.
2060 Version 7.2 19-Jun-07
2061 ---------------------
2063 1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
2064 which is apparently normally available under Windows.
2066 2. Re-jig the pcregrep tests with different newline settings in an attempt
2067 to make them independent of the local environment's newline setting.
2069 3. Add code to configure.ac to remove -g from the CFLAGS default settings.
2071 4. Some of the "internals" tests were previously cut out when the link size
2072 was not 2, because the output contained actual offsets. The recent new
2073 "Z" feature of pcretest means that these can be cut out, making the tests
2074 usable with all link sizes.
2076 5. Implemented Stan Switzer's goto replacement for longjmp() when not using
2077 stack recursion. This gives a massive performance boost under BSD, but just
2078 a small improvement under Linux. However, it saves one field in the frame
2079 in all cases.
2081 6. Added more features from the forthcoming Perl 5.10:
2083 (a) (?-n) (where n is a string of digits) is a relative subroutine or
2084 recursion call. It refers to the nth most recently opened parentheses.
2086 (b) (?+n) is also a relative subroutine call; it refers to the nth next
2087 to be opened parentheses.
2089 (c) Conditions that refer to capturing parentheses can be specified
2090 relatively, for example, (?(-2)... or (?(+3)...
2092 (d) \K resets the start of the current match so that everything before
2093 is not part of it.
2095 (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
2097 (f) \g{name} is another synonym - part of Perl 5.10's unification of
2098 reference syntax.
2100 (g) (?| introduces a group in which the numbering of parentheses in each
2101 alternative starts with the same number.
2103 (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
2105 7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
2108 8. A pattern such as (.*(.)?)* caused pcre_exec() to fail by either not
2109 terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
2110 for detecting groups that can match an empty string.
2112 9. A pattern with a very large number of alternatives (more than several
2113 hundred) was running out of internal workspace during the pre-compile
2114 phase, where pcre_compile() figures out how much memory will be needed. A
2115 bit of new cunning has reduced the workspace needed for groups with
2116 alternatives. The 1000-alternative test pattern now uses 12 bytes of
2117 workspace instead of running out of the 4096 that are available.
2119 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
2121 11. Applied patch from Google to remove an optimization that didn't quite work.
2122 The report of the bug said:
2124 pcrecpp::RE("a*").FullMatch("aaa") matches, while
2125 pcrecpp::RE("a*?").FullMatch("aaa") does not, and
2126 pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
2128 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
2129 it matched the wrong number of bytes.
2132 Version 7.1 24-Apr-07
2133 ---------------------
2135 1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
2136 that is more "standard", making use of automake and other Autotools. There
2137 is some re-arrangement of the files and adjustment of comments consequent
2138 on this.
2140 2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
2141 for recursive directory scanning broke on some systems because the files
2142 are not scanned in any specific order and on different systems the order
2143 was different. A call to "sort" has been inserted into RunGrepTest for the
2144 approprate test as a short-term fix. In the longer term there may be an
2145 alternative.
2147 3. I had an email from Eric Raymond about problems translating some of PCRE's
2148 man pages to HTML (despite the fact that I distribute HTML pages, some
2149 people do their own conversions for various reasons). The problems
2150 concerned the use of low-level troff macros .br and .in. I have therefore
2151 removed all such uses from the man pages (some were redundant, some could
2152 be replaced by .nf/.fi pairs). The 132html script that I use to generate
2153 HTML has been updated to handle .nf/.fi and to complain if it encounters
2154 .br or .in.
2156 4. Updated comments in configure.ac that get placed in config.h.in and also
2157 arranged for config.h to be included in the distribution, with the name
2158 config.h.generic, for the benefit of those who have to compile without
2159 Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
2161 5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
2162 Weber: (1) pcre_internal.h was missing some function renames; (2) updated
2163 makevp.bat for the current PCRE, using the additional files
2164 makevp_c.txt, makevp_l.txt, and pcregexp.pas.
2166 6. A Windows user reported a minor discrepancy with test 2, which turned out
2167 to be caused by a trailing space on an input line that had got lost in his
2168 copy. The trailing space was an accident, so I've just removed it.
2170 7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
2171 that is needed.
2173 8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
2174 as "const" (a) because they are and (b) because it helps the PHP
2175 maintainers who have recently made a script to detect big data structures
2176 in the php code that should be moved to the .rodata section. I remembered
2177 to update Builducptable as well, so it won't revert if ucptable.h is ever
2178 re-created.
2180 9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
2181 pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
2182 order to be able to cut out the UTF-8 tables in the latter when UTF-8
2183 support is not required. This saves 1.5-2K of code, which is important in
2184 some applications.
2186 Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
2187 so as not to refer to the tables, even though these functions will never be
2188 called when UTF-8 support is disabled. Otherwise there are problems with a
2189 shared library.
2191 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
2193 (a) It was defining its arguments as char * instead of void *.
2195 (b) It was assuming that all moves were upwards in memory; this was true
2196 a long time ago when I wrote it, but is no longer the case.
2198 The emulated memove() is provided for those environments that have neither
2199 memmove() nor bcopy(). I didn't think anyone used it these days, but that
2200 is clearly not the case, as these two bugs were recently reported.
2202 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
2203 and Detrail to create the HTML documentation, the .txt form of the man
2204 pages, and it removes trailing spaces from listed files. It also creates
2205 pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
2206 case, it wraps all the #defines with #ifndefs. This script should be run
2207 before "make dist".
2209 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
2210 with Unicode property support.
2212 (a) For a maximizing quantifier, if the two different cases of the
2213 character were of different lengths in their UTF-8 codings (there are
2214 some cases like this - I found 11), and the matching function had to
2215 back up over a mixture of the two cases, it incorrectly assumed they
2216 were both the same length.
2218 (b) When PCRE was configured to use the heap rather than the stack for
2219 recursion during matching, it was not correctly preserving the data for
2220 the other case of a UTF-8 character when checking ahead for a match
2221 while processing a minimizing repeat. If the check also involved
2222 matching a wide character, but failed, corruption could cause an
2223 erroneous result when trying to check for a repeat of the original
2224 character.
2226 13. Some tidying changes to the testing mechanism:
2228 (a) The RunTest script now detects the internal link size and whether there
2229 is UTF-8 and UCP support by running ./pcretest -C instead of relying on
2230 values substituted by "configure". (The RunGrepTest script already did
2231 this for UTF-8.) The configure.ac script no longer substitutes the
2232 relevant variables.
2234 (b) The debugging options /B and /D in pcretest show the compiled bytecode
2235 with length and offset values. This means that the output is different
2236 for different internal link sizes. Test 2 is skipped for link sizes
2237 other than 2 because of this, bypassing the problem. Unfortunately,
2238 there was also a test in test 3 (the locale tests) that used /B and
2239 failed for link sizes other than 2. Rather than cut the whole test out,
2240 I have added a new /Z option to pcretest that replaces the length and
2241 offset values with spaces. This is now used to make test 3 independent
2242 of link size. (Test 2 will be tidied up later.)
2244 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
2245 segmentation fault instead of returning the appropriate error message.
2247 15. In multiline mode when the newline sequence was set to "any", the pattern
2248 ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
2249 This doesn't seem right; it now treats the CRLF combination as the line
2250 ending, and so does not match in that case. It's only a pattern such as ^$
2251 that would hit this one: something like ^ABC$ would have failed after \r
2252 and then tried again after \r\n.
2254 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
2255 in an attempt to make files that differ only in their line terminators
2256 compare equal. This works on Linux.
2258 17. Under certain error circumstances pcregrep might try to free random memory
2259 as it exited. This is now fixed, thanks to valgrind.
2261 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
2262 "abc\r\n\r\n", it found an unwanted second match after the second \r. This
2263 was because its rules for how to advance for /g after matching an empty
2264 string at the end of a line did not allow for this case. They now check for
2265 it specially.
2267 20. pcretest is supposed to handle patterns and data of any length, by
2268 extending its buffers when necessary. It was getting this wrong when the
2269 buffer for a data line had to be extended.
2271 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
2272 CRLF as a newline sequence.
2274 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
2275 out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
2276 I have nevertheless tidied it up.
2278 23. Added some casts to kill warnings from HP-UX ia64 compiler.
2280 24. Added a man page for pcre-config.
2283 Version 7.0 19-Dec-06
2284 ---------------------
2286 1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
2287 moving to gcc 4.1.1.
2289 2. The -S option for pcretest uses setrlimit(); I had omitted to #include
2290 sys/time.h, which is documented as needed for this function. It doesn't
2291 seem to matter on Linux, but it showed up on some releases of OS X.
2293 3. It seems that there are systems where bytes whose values are greater than
2294 127 match isprint() in the "C" locale. The "C" locale should be the
2295 default when a C program starts up. In most systems, only ASCII printing
2296 characters match isprint(). This difference caused the output from pcretest
2297 to vary, making some of the tests fail. I have changed pcretest so that:
2299 (a) When it is outputting text in the compiled version of a pattern, bytes
2300 other than 32-126 are always shown as hex escapes.
2302 (b) When it is outputting text that is a matched part of a subject string,
2303 it does the same, unless a different locale has been set for the match
2304 (using the /L modifier). In this case, it uses isprint() to decide.
2306 4. Fixed a major bug that caused incorrect computation of the amount of memory
2307 required for a compiled pattern when options that changed within the
2308 pattern affected the logic of the preliminary scan that determines the
2309 length. The relevant options are -x, and -i in UTF-8 mode. The result was
2310 that the computed length was too small. The symptoms of this bug were
2311 either the PCRE error "internal error: code overflow" from pcre_compile(),
2312 or a glibc crash with a message such as "pcretest: free(): invalid next
2313 size (fast)". Examples of patterns that provoked this bug (shown in
2314 pcretest format) are:
2316 /(?-x: )/x
2317 /(?x)(?-x: \s*#\s*)/
2318 /((?i)[\x{c0}])/8
2319 /(?i:[\x{c0}])/8
2321 HOWEVER: Change 17 below makes this fix obsolete as the memory computation
2322 is now done differently.
2324 5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
2325 wrapper classes; (b) implement a new function in the C++ scanner that is
2326 more efficient than the old way of doing things because it avoids levels of
2327 recursion in the regex matching; (c) add a paragraph to the documentation
2328 for the FullMatch() function.
2330 6. The escape sequence \n was being treated as whatever was defined as
2331 "newline". Not only was this contrary to the documentation, which states
2332 that \n is character 10 (hex 0A), but it also went horribly wrong when
2333 "newline" was defined as CRLF. This has been fixed.
2335 7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
2336 was being set to -1 for the "end of line" case (supposedly a value that no
2337 character can have). Though this value is never used (the check for end of
2338 line is "zero bytes in current character"), it caused compiler complaints.
2339 I've changed it to 0xffffffff.
2341 8. In pcre_version.c, the version string was being built by a sequence of
2342 C macros that, in the event of PCRE_PRERELEASE being defined as an empty
2343 string (as it is for production releases) called a macro with an empty
2344 argument. The C standard says the result of this is undefined. The gcc
2345 compiler treats it as an empty string (which was what was wanted) but it is
2346 reported that Visual C gives an error. The source has been hacked around to
2347 avoid this problem.
2349 9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
2350 builds of pcretest, and changed the call to _setmode() to use _O_BINARY
2351 instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
2352 of them did).
2354 10. Originally, pcretest opened its input and output without "b"; then I was
2355 told that "b" was needed in some environments, so it was added for release
2356 5.0 to both the input and output. (It makes no difference on Unix-like
2357 systems.) Later I was told that it is wrong for the input on Windows. I've
2358 now abstracted the modes into two macros, to make it easier to fiddle with
2359 them, and removed "b" from the input mode under Windows.
2361 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
2363 12. Added -help and --help to pcretest as an official way of being reminded
2364 of the options.
2366 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
2367 and pcrecpp.cc because they annoy compilers at high warning levels.
2369 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
2371 15. Fixed an occurrence of == in configure.ac that should have been = (shell
2372 scripts are not C programs :-) and which was not noticed because it works
2373 on Linux.
2375 16. pcretest is supposed to handle any length of pattern and data line (as one
2376 line or as a continued sequence of lines) by extending its input buffer if
2377 necessary. This feature was broken for very long pattern lines, leading to
2378 a string of junk being passed to pcre_compile() if the pattern was longer
2379 than about 50K.
2381 17. I have done a major re-factoring of the way pcre_compile() computes the
2382 amount of memory needed for a compiled pattern. Previously, there was code
2383 that made a preliminary scan of the pattern in order to do this. That was
2384 OK when PCRE was new, but as the facilities have expanded, it has become
2385 harder and harder to keep it in step with the real compile phase, and there
2386 have been a number of bugs (see for example, 4 above). I have now found a
2387 cunning way of running the real compile function in a "fake" mode that
2388 enables it to compute how much memory it would need, while actually only
2389 ever using a few hundred bytes of working memory and without too many
2390 tests of the mode. This should make future maintenance and development
2391 easier. A side effect of this work is that the limit of 200 on the nesting
2392 depth of parentheses has been removed (though this was never a serious
2393 limitation, I suspect). However, there is a downside: pcre_compile() now
2394 runs more slowly than before (30% or more, depending on the pattern). I
2395 hope this isn't a big issue. There is no effect on runtime performance.
2397 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
2398 newline (only possible for the last line of a file) and it was a
2399 pattern that set a locale (followed by /Lsomething), pcretest crashed.
2401 19. Added additional timing features to pcretest. (1) The -tm option now times
2402 matching only, not compiling. (2) Both -t and -tm can be followed, as a
2403 separate command line item, by a number that specifies the number of
2404 repeats to use when timing. The default is 50000; this gives better
2405 precision, but takes uncomfortably long for very large patterns.
2407 20. Extended pcre_study() to be more clever in cases where a branch of a
2408 subpattern has no definite first character. For example, (a*|b*)[cd] would
2409 previously give no result from pcre_study(). Now it recognizes that the
2410 first character must be a, b, c, or d.
2412 21. There was an incorrect error "recursive call could loop indefinitely" if
2413 a subpattern (or the entire pattern) that was being tested for matching an
2414 empty string contained only one non-empty item after a nested subpattern.
2415 For example, the pattern (?>\x{100}*)\d(?R) provoked this error
2416 incorrectly, because the \d was being skipped in the check.
2418 22. The pcretest program now has a new pattern option /B and a command line
2419 option -b, which is equivalent to adding /B to every pattern. This causes
2420 it to show the compiled bytecode, without the additional information that
2421 -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
2422 is the same as /B/I).
2424 23. A new optimization is now able automatically to treat some sequences such
2425 as a*b as a*+b. More specifically, if something simple (such as a character
2426 or a simple class like \d) has an unlimited quantifier, and is followed by
2427 something that cannot possibly match the quantified thing, the quantifier
2428 is automatically "possessified".
2430 24. A recursive reference to a subpattern whose number was greater than 39
2431 went wrong under certain circumstances in UTF-8 mode. This bug could also
2432 have affected the operation of pcre_study().
2434 25. Realized that a little bit of performance could be had by replacing
2435 (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
2437 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
2439 27. Possessive quantifiers such as a++ were previously implemented by turning
2440 them into atomic groups such as ($>a+). Now they have their own opcodes,
2441 which improves performance. This includes the automatically created ones
2442 from 23 above.
2444 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
2445 lookahead was broken if it was not anchored. PCRE was mistakenly expecting
2446 the first matched character to be a colon. This applied both to named and
2447 numbered groups.
2449 29. The ucpinternal.h header file was missing its idempotency #ifdef.
2451 30. I was sent a "project" file called libpcre.a.dev which I understand makes
2452 building PCRE on Windows easier, so I have included it in the distribution.
2454 31. There is now a check in pcretest against a ridiculously large number being
2455 returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
2456 loop, the loop is abandoned.
2458 32. Forward references to subpatterns in conditions such as (?(2)...) where
2459 subpattern 2 is defined later cause pcre_compile() to search forwards in
2460 the pattern for the relevant set of parentheses. This search went wrong
2461 when there were unescaped parentheses in a character class, parentheses
2462 escaped with \Q...\E, or parentheses in a #-comment in /x mode.
2464 33. "Subroutine" calls and backreferences were previously restricted to
2465 referencing subpatterns earlier in the regex. This restriction has now
2466 been removed.
2468 34. Added a number of extra features that are going to be in Perl 5.10. On the
2469 whole, these are just syntactic alternatives for features that PCRE had
2470 previously implemented using the Python syntax or my own invention. The
2471 other formats are all retained for compatibility.
2473 (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
2474 as (?P<name>...). The new forms, as well as being in Perl 5.10, are
2475 also .NET compatible.
2477 (b) A recursion or subroutine call to a named group can now be defined as
2478 (?&name) as well as (?P>name).
2480 (c) A backreference to a named group can now be defined as \k<name> or
2481 \k'name' as well as (?P=name). The new forms, as well as being in Perl
2482 5.10, are also .NET compatible.
2484 (d) A conditional reference to a named group can now use the syntax
2485 (?(<name>) or (?('name') as well as (?(name).
2487 (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
2488 groups (named and numbered) that are never evaluated inline, but can be
2489 called as "subroutines" from elsewhere. In effect, the DEFINE condition
2490 is always false. There may be only one alternative in such a group.
2492 (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
2493 as the simple (?(R). The condition is true only if the most recent
2494 recursion is that of the given number or name. It does not search out
2495 through the entire recursion stack.
2497 (g) The escape \gN or \g{N} has been added, where N is a positive or
2498 negative number, specifying an absolute or relative reference.
2500 35. Tidied to get rid of some further signed/unsigned compiler warnings and
2501 some "unreachable code" warnings.
2503 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
2504 things, this adds five new scripts.
2506 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
2507 There were also incompatibilities regarding the handling of \Q..\E inside
2508 character classes, for example with patterns like [\Qa\E-\Qz\E] where the
2509 hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
2511 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
2512 matches an empty string, and forcibly breaks the loop. There were bugs in
2513 this code in non-simple cases. For a pattern such as ^(a()*)* matched
2514 against aaaa the result was just "a" rather than "aaaa", for example. Two
2515 separate and independent bugs (that affected different cases) have been
2516 fixed.
2518 39. Refactored the code to abolish the use of different opcodes for small
2519 capturing bracket numbers. This is a tidy that I avoided doing when I
2520 removed the limit on the number of capturing brackets for 3.5 back in 2001.
2521 The new approach is not only tidier, it makes it possible to reduce the
2522 memory needed to fix the previous bug (38).
2524 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
2525 sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
2526 processing dot, circumflex, or dollar metacharacters, or #-comments in /x
2527 mode.
2529 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
2530 report.
2532 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
2533 copy construction and assignment in the C++ wrapper.
2535 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
2536 couple of bugs that could have given wrong results in the "--newline=crlf"
2537 case.
2539 44. Added a number of casts and did some reorganization of signed/unsigned int
2540 variables following suggestions from Dair Grant. Also renamed the variable
2541 "this" as "item" because it is a C++ keyword.
2543 45. Arranged for dftables to add
2545 #include "pcre_internal.h"
2547 to pcre_chartables.c because without it, gcc 4.x may remove the array
2548 definition from the final binary if PCRE is built into a static library and
2549 dead code stripping is activated.
2551 46. For an unanchored pattern, if a match attempt fails at the start of a
2552 newline sequence, and the newline setting is CRLF or ANY, and the next two
2553 characters are CRLF, advance by two characters instead of one.
2556 Version 6.7 04-Jul-06
2557 ---------------------
2559 1. In order to handle tests when input lines are enormously long, pcretest has
2560 been re-factored so that it automatically extends its buffers when
2561 necessary. The code is crude, but this _is_ just a test program. The
2562 default size has been increased from 32K to 50K.
2564 2. The code in pcre_study() was using the value of the re argument before
2565 testing it for NULL. (Of course, in any sensible call of the function, it
2566 won't be NULL.)
2568 3. The memmove() emulation function in pcre_internal.h, which is used on
2569 systems that lack both memmove() and bcopy() - that is, hardly ever -
2570 was missing a "static" storage class specifier.
2572 4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
2573 containing an extended class (one that cannot be represented by a bitmap
2574 because it contains high-valued characters or Unicode property items, e.g.
2575 [\pZ]). Almost always one would set UTF-8 mode when processing such a
2576 pattern, but PCRE should not loop if you do not (it no longer does).
2577 [Detail: two cases were found: (a) a repeated subpattern containing an
2578 extended class; (b) a recursive reference to a subpattern that followed a
2579 previous extended class. It wasn't skipping over the extended class
2580 correctly when UTF-8 mode was not set.]
2582 5. A negated single-character class was not being recognized as fixed-length
2583 in lookbehind assertions such as (?<=[^f]), leading to an incorrect
2584 compile error "lookbehind assertion is not fixed length".
2586 6. The RunPerlTest auxiliary script was showing an unexpected difference
2587 between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
2588 write a Perl script that can interpret lines of an input file either as
2589 byte characters or as UTF-8, which is what "perltest" was being required to
2590 do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
2591 can't do is switch easily at run time between having the "use utf8;" pragma
2592 or not. In the end, I fudged it by using the RunPerlTest script to insert
2593 "use utf8;" explicitly for the UTF-8 tests.
2595 7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
2596 the end of the subject string, contrary to the documentation and to what
2597 Perl does. This was true of both matching functions. Now it matches only at
2598 the start of the subject and immediately after *internal* newlines.
2600 8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
2601 a pointer to an int instead of a pointer to an unsigned long int. This
2602 caused problems on 64-bit systems.
2604 9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
2605 instance of the 'standard' template library not being so standard".
2607 10. There was no check on the number of named subpatterns nor the maximum
2608 length of a subpattern name. The product of these values is used to compute
2609 the size of the memory block for a compiled pattern. By supplying a very
2610 long subpattern name and a large number of named subpatterns, the size
2611 computation could be caused to overflow. This is now prevented by limiting
2612 the length of names to 32 characters, and the number of named subpatterns
2613 to 10,000.
2615 11. Subpatterns that are repeated with specific counts have to be replicated in
2616 the compiled pattern. The size of memory for this was computed from the
2617 length of the subpattern and the repeat count. The latter is limited to
2618 65535, but there was no limit on the former, meaning that integer overflow
2619 could in principle occur. The compiled length of a repeated subpattern is
2620 now limited to 30,000 bytes in order to prevent this.
2622 12. Added the optional facility to have named substrings with the same name.
2624 13. Added the ability to use a named substring as a condition, using the
2625 Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
2626 are numbers (not recommended). Forward references are permitted.
2628 14. Added forward references in named backreferences (if you see what I mean).
2630 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
2631 pattern could run off the end of the subject. For example, the pattern
2632 "(?s)(.{1,5})"8 did this with the subject "ab".
2634 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
2635 PCRE_CASELESS was set when matching characters that were quantified with ?
2636 or *.
2638 17. A character class other than a single negated character that had a minimum
2639 but no maximum quantifier - for example [ab]{6,} - was not handled
2640 correctly by pce_dfa_exec(). It would match only one character.
2642 18. A valid (though odd) pattern that looked like a POSIX character
2643 class but used an invalid character after [ (for example [[,abc,]]) caused
2644 pcre_compile() to give the error "Failed: internal error: code overflow" or
2645 in some cases to crash with a glibc free() error. This could even happen if
2646 the pattern terminated after [[ but there just happened to be a sequence of
2647 letters, a binary zero, and a closing ] in the memory that followed.
2649 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
2650 over the years. Originally (before any Unicode support), just the bottom 8
2651 bits were taken. Thus, for example, \500 really meant \100. Nowadays the
2652 output from "man perlunicode" includes this:
2654 The regular expression compiler produces polymorphic opcodes. That
2655 is, the pattern adapts to the data and automatically switches to
2656 the Unicode character scheme when presented with Unicode data--or
2657 instead uses a traditional byte scheme when presented with byte
2658 data.
2660 Sadly, a wide octal escape does not cause a switch, and in a string with
2661 no other multibyte characters, these octal escapes are treated as before.
2662 Thus, in Perl, the pattern /\500/ actually matches \100 but the pattern
2663 /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
2664 Unicode string.
2666 I have not perpetrated such confusion in PCRE. Up till now, it took just
2667 the bottom 8 bits, as in old Perl. I have now made octal escapes with
2668 values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
2669 translate to the appropriate multibyte character.
2671 29. Applied some refactoring to reduce the number of warnings from Microsoft
2672 and Borland compilers. This has included removing the fudge introduced
2673 seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
2674 a warning about an unused variable.
2676 21. PCRE has not included VT (character 0x0b) in the set of whitespace
2677 characters since release 4.0, because Perl (from release 5.004) does not.
2678 [Or at least, is documented not to: some releases seem to be in conflict
2679 with the documentation.] However, when a pattern was studied with
2680 pcre_study() and all its branches started with \s, PCRE still included VT
2681 as a possible starting character. Of course, this did no harm; it just
2682 caused an unnecessary match attempt.
2684 22. Removed a now-redundant internal flag bit that recorded the fact that case
2685 dependency changed within the pattern. This was once needed for "required
2686 byte" processing, but is no longer used. This recovers a now-scarce options
2687 bit. Also moved the least significant internal flag bit to the most-
2688 significant bit of the word, which was not previously used (hangover from
2689 the days when it was an int rather than a uint) to free up another bit for
2690 the future.
2692 23. Added support for CRLF line endings as well as CR and LF. As well as the
2693 default being selectable at build time, it can now be changed at runtime
2694 via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
2695 specify that it is scanning data with non-default line endings.
2697 24. Changed the definition of CXXLINK to make it agree with the definition of
2698 LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
2700 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
2701 recursions. This makes a big different to stack usage for some patterns.
2703 26. If a subpattern containing a named recursion or subroutine reference such
2704 as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
2705 the space required for the compiled pattern went wrong and gave too small a
2706 value. Depending on the environment, this could lead to "Failed: internal
2707 error: code overflow at offset 49" or "glibc detected double free or
2708 corruption" errors.
2710 27. Applied patches from Google (a) to support the new newline modes and (b) to
2711 advance over multibyte UTF-8 characters in GlobalReplace.
2713 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
2714 difference for some implementation of PCRE in some Windows version.
2716 29. Added some extra testing facilities to pcretest:
2718 \q<number> in a data line sets the "match limit" value
2719 \Q<number> in a data line sets the "match recursion limt" value
2720 -S <number> sets the stack size, where <number> is in megabytes
2722 The -S option isn't available for Windows.
2725 Version 6.6 06-Feb-06
2726 ---------------------
2728 1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
2729 in pcreposix.h. I have copied the definition from pcre.h.
2731 2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
2732 because pcre.h is no longer a built file.
2734 3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
2735 not normally included in the compiled code.
2738 Version 6.5 01-Feb-06
2739 ---------------------
2741 1. When using the partial match feature with pcre_dfa_exec(), it was not
2742 anchoring the second and subsequent partial matches at the new starting
2743 point. This could lead to incorrect results. For example, with the pattern
2744 /1234/, partially matching against "123" and then "a4" gave a match.
2746 2. Changes to pcregrep:
2748 (a) All non-match returns from pcre_exec() were being treated as failures
2749 to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
2750 error message is output. Some extra information is given for the
2752 probably the only errors that are likely to be caused by users (by
2753 specifying a regex that has nested indefinite repeats, for instance).
2754 If there are more than 20 of these errors, pcregrep is abandoned.
2756 (b) A binary zero was treated as data while matching, but terminated the
2757 output line if it was written out. This has been fixed: binary zeroes
2758 are now no different to any other data bytes.
2760 (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
2761 used to set a locale for matching. The --locale=xxxx long option has
2762 been added (no short equivalent) to specify a locale explicitly on the
2763 pcregrep command, overriding the environment variables.
2765 (d) When -B was used with -n, some line numbers in the output were one less
2766 than they should have been.
2768 (e) Added the -o (--only-matching) option.
2770 (f) If -A or -C was used with -c (count only), some lines of context were
2771 accidentally printed for the final match.
2773 (g) Added the -H (--with-filename) option.
2775 (h) The combination of options -rh failed to suppress file names for files
2776 that were found from directory arguments.
2778 (i) Added the -D (--devices) and -d (--directories) options.
2780 (j) Added the -F (--fixed-strings) option.
2782 (k) Allow "-" to be used as a file name for -f as well as for a data file.
2784 (l) Added the --colo(u)r option.
2786 (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
2787 is not present by default.
2789 3. A nasty bug was discovered in the handling of recursive patterns, that is,
2790 items such as (?R) or (?1), when the recursion could match a number of
2791 alternatives. If it matched one of the alternatives, but subsequently,
2792 outside the recursion, there was a failure, the code tried to back up into
2793 the recursion. However, because of the way PCRE is implemented, this is not
2794 possible, and the result was an incorrect result from the match.
2796 In order to prevent this happening, the specification of recursion has
2797 been changed so that all such subpatterns are automatically treated as
2798 atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
2800 4. I had overlooked the fact that, in some locales, there are characters for
2801 which isalpha() is true but neither isupper() nor islower() are true. In
2802 the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
2803 and ordfeminine) are like this. This affected the treatment of \w and \W
2804 when they appeared in character classes, but not when they appeared outside
2805 a character class. The bit map for "word" characters is now created
2806 separately from the results of isalnum() instead of just taking it from the
2807 upper, lower, and digit maps. (Plus the underscore character, of course.)
2809 5. The above bug also affected the handling of POSIX character classes such as
2810 [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
2811 permanent tables. Instead, the bit maps for such a class were previously
2812 created as the appropriate unions of the upper, lower, and digit bitmaps.
2813 Now they are created by subtraction from the [[:word:]] class, which has
2814 its own bitmap.
2816 6. The [[:blank:]] character class matches horizontal, but not vertical space.
2817 It is created by subtracting the vertical space characters (\x09, \x0a,
2818 \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
2819 subtraction was done in the overall bitmap for a character class, meaning
2820 that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
2821 be recognized. This bug has been fixed.
2823 7. Patches from the folks at Google:
2825 (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
2826 real life, but is still worth protecting against".
2828 (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
2829 regular expressions".
2831 (c) pcre_scanner.cc: avoid use of std::count() because not all systems
2832 have it.
2834 (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
2835 "configure" and the latter not, in order to fix a problem somebody had
2836 with compiling the Arg class on HP-UX.
2838 (e) Improve the error-handling of the C++ wrapper a little bit.
2840 (f) New tests for checking recursion limiting.
2842 8. The pcre_memmove() function, which is used only if the environment does not
2843 have a standard memmove() function (and is therefore rarely compiled),
2844 contained two bugs: (a) use of int instead of size_t, and (b) it was not
2845 returning a result (though PCRE never actually uses the result).
2847 9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
2848 large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
2849 returned instead of calling malloc() with an overflowing number that would
2850 most likely cause subsequent chaos.
2852 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
2854 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
2855 with this option is matched, the nmatch and pmatch options of regexec() are
2856 ignored.
2858 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
2859 provided in case anyone wants to the the POSIX interface with UTF-8
2860 strings.
2862 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
2863 C++ linking (needed for some HP-UX environments).
2865 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
2866 (unused parameter) and in the pcre_printint() function (omitted "default"
2867 switch label when the default is to do nothing).
2869 15. Added some code to make it possible, when PCRE is compiled as a C++
2870 library, to replace subject pointers for pcre_exec() with a smart pointer
2871 class, thus making it possible to process discontinuous strings.
2873 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
2874 much the same function. They were added by different people who were trying
2875 to make PCRE easy to compile on non-Unix systems. It has been suggested
2876 that PCRE_EXPORT be abolished now that there is more automatic apparatus
2877 for compiling on Windows systems. I have therefore replaced it with
2878 PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
2879 defaults to "extern" for C or "extern C" for C++, which works fine on
2880 Unix-like systems. It is now possible to override the value of PCRE_DATA_
2881 SCOPE with something explicit in config.h. In addition:
2883 (a) pcreposix.h still had just "extern" instead of either of these macros;
2884 I have replaced it with PCRE_DATA_SCOPE.
2886 (b) Functions such as _pcre_xclass(), which are internal to the library,
2887 but external in the C sense, all had PCRE_EXPORT in their definitions.
2888 This is apparently wrong for the Windows case, so I have removed it.
2889 (It makes no difference on Unix-like systems.)
2891 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
2892 of recursive calls to match(). This is different to MATCH_LIMIT because
2893 that limits the total number of calls to match(), not all of which increase
2894 the depth of recursion. Limiting the recursion depth limits the amount of
2895 stack (or heap if NO_RECURSE is set) that is used. The default can be set
2896 when PCRE is compiled, and changed at run time. A patch from Google adds
2897 this functionality to the C++ interface.
2899 18. Changes to the handling of Unicode character properties:
2901 (a) Updated the table to Unicode 4.1.0.
2903 (b) Recognize characters that are not in the table as "Cn" (undefined).
2905 (c) I revised the way the table is implemented to a much improved format
2906 which includes recognition of ranges. It now supports the ranges that
2907 are defined in UnicodeData.txt, and it also amalgamates other
2908 characters into ranges. This has reduced the number of entries in the
2909 table from around 16,000 to around 3,000, thus reducing its size
2910 considerably. I realized I did not need to use a tree structure after
2911 all - a binary chop search is just as efficient. Having reduced the
2912 number of entries, I extended their size from 6 bytes to 8 bytes to
2913 allow for more data.
2915 (d) Added support for Unicode script names via properties such as \p{Han}.
2917 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
2918 matching that character.
2920 20. When matching a repeated Unicode property with a minimum greater than zero,
2921 (for example \pL{2,}), PCRE could look past the end of the subject if it
2922 reached it while seeking the minimum number of characters. This could
2923 happen only if some of the characters were more than one byte long, because
2924 there is a check for at least the minimum number of bytes.
2926 21. Refactored the implementation of \p and \P so as to be more general, to
2927 allow for more different types of property in future. This has changed the
2928 compiled form incompatibly. Anybody with saved compiled patterns that use
2929 \p or \P will have to recompile them.
2931 22. Added "Any" and "L&" to the supported property types.
2933 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
2934 but give a compile time error if the value is greater than 0xff.
2936 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
2937 accidentally not being installed or uninstalled.
2939 25. The pcre.h file was built from pcre.h.in, but the only changes that were
2940 made were to insert the current release number. This seemed silly, because
2941 it made things harder for people building PCRE on systems that don't run
2942 "configure". I have turned pcre.h into a distributed file, no longer built
2943 by "configure", with the version identification directly included. There is
2944 no longer a pcre.h.in file.
2946 However, this change necessitated a change to the pcre-config script as
2947 well. It is built from pcre-config.in, and one of the substitutions was the
2948 release number. I have updated configure.ac so that ./configure now finds
2949 the release number by grepping pcre.h.
2951 26. Added the ability to run the tests under valgrind.
2954 Version 6.4 05-Sep-05
2955 ---------------------
2957 1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
2958 "--" to be printed when multiple files were scanned, even when none of the
2959 -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
2960 consider it to be a bug, and have restored the previous behaviour.
2962 2. A couple of code tidies to get rid of compiler warnings.
2964 3. The pcretest program used to cheat by referring to symbols in the library
2965 whose names begin with _pcre_. These are internal symbols that are not
2966 really supposed to be visible externally, and in some environments it is
2967 possible to suppress them. The cheating is now confined to including
2968 certain files from the library's source, which is a bit cleaner.
2970 4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
2971 file's purpose clearer.
2973 5. Reorganized pcre_ucp_findchar().
2976 Version 6.3 15-Aug-05
2977 ---------------------
2979 1. The file libpcre.pc.in did not have general read permission in the tarball.
2981 2. There were some problems when building without C++ support:
2983 (a) If C++ support was not built, "make install" and "make test" still
2984 tried to test it.
2986 (b) There were problems when the value of CXX was explicitly set. Some
2987 changes have been made to try to fix these, and ...
2989 (c) --disable-cpp can now be used to explicitly disable C++ support.
2991 (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
2992 backslash in a target when C++ was disabled. This confuses some
2993 versions of "make", apparently. Using an intermediate variable solves
2994 this. (Same for CPP_LOBJ.)
2996 3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
2997 (non-Windows) now includes $(CFLAGS) because these flags are sometimes
2998 necessary on certain architectures.
3000 4. Added a setting of -export-symbols-regex to the link command to remove
3001 those symbols that are exported in the C sense, but actually are local
3002 within the library, and not documented. Their names all begin with
3003 "_pcre_". This is not a perfect job, because (a) we have to except some
3004 symbols that pcretest ("illegally") uses, and (b) the facility isn't always
3005 available (and never for static libraries). I have made a note to try to
3006 find a way round (a) in the future.
3009 Version 6.2 01-Aug-05
3010 ---------------------
3012 1. There was no test for integer overflow of quantifier values. A construction
3013 such as {1111111111111111} would give undefined results. What is worse, if
3014 a minimum quantifier for a parenthesized subpattern overflowed and became
3015 negative, the calculation of the memory size went wrong. This could have
3016 led to memory overwriting.
3018 2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
3020 3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
3021 operating environments where this matters.
3023 4. Applied Giuseppe Maxia's patch to add additional features for controlling
3024 PCRE options from within the C++ wrapper.
3026 5. Named capturing subpatterns were not being correctly counted when a pattern
3027 was compiled. This caused two problems: (a) If there were more than 100
3028 such subpatterns, the calculation of the memory needed for the whole
3029 compiled pattern went wrong, leading to an overflow error. (b) Numerical
3030 back references of the form \12, where the number was greater than 9, were
3031 not recognized as back references, even though there were sufficient
3032 previous subpatterns.
3034 6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
3035 versions of gcc, e.g. 2.95.4.
3038 Version 6.1 21-Jun-05
3039 ---------------------
3041 1. There was one reference to the variable "posix" in pcretest.c that was not
3042 surrounded by "#if !defined NOPOSIX".
3044 2. Make it possible to compile pcretest without DFA support, UTF8 support, or
3045 the cross-check on the old pcre_info() function, for the benefit of the
3046 cut-down version of PCRE that is currently imported into Exim.
3048 3. A (silly) pattern starting with (?i)(?-i) caused an internal space
3049 allocation error. I've done the easy fix, which wastes 2 bytes for sensible
3050 patterns that start (?i) but I don't think that matters. The use of (?i) is
3051 just an example; this all applies to the other options as well.
3053 4. Since libtool seems to echo the compile commands it is issuing, the output
3054 from "make" can be reduced a bit by putting "@" in front of each libtool
3055 compile command.
3057 5. Patch from the folks at Google for configure.in to be a bit more thorough
3058 in checking for a suitable C++ installation before trying to compile the
3059 C++ stuff. This should fix a reported problem when a compiler was present,
3060 but no suitable headers.
3062 6. The man pages all had just "PCRE" as their title. I have changed them to
3063 be the relevant file name. I have also arranged that these names are
3064 retained in the file doc/pcre.txt, which is a concatenation in text format
3065 of all the man pages except the little individual ones for each function.
3067 7. The NON-UNIX-USE file had not been updated for the different set of source
3068 files that come with release 6. I also added a few comments about the C++
3069 wrapper.
3072 Version 6.0 07-Jun-05
3073 ---------------------
3075 1. Some minor internal re-organization to help with my DFA experiments.
3077 2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
3078 didn't matter for the library itself when fully configured, but did matter
3079 when compiling without UCP support, or within Exim, where the ucp files are
3080 not imported.
3082 3. Refactoring of the library code to split up the various functions into
3083 different source modules. The addition of the new DFA matching code (see
3084 below) to a single monolithic source would have made it really too
3085 unwieldy, quite apart from causing all the code to be include in a
3086 statically linked application, when only some functions are used. This is
3087 relevant even without the DFA addition now that patterns can be compiled in
3088 one application and matched in another.
3090 The downside of splitting up is that there have to be some external
3091 functions and data tables that are used internally in different modules of
3092 the library but which are not part of the API. These have all had their
3093 names changed to start with "_pcre_" so that they are unlikely to clash
3094 with other external names.
3096 4. Added an alternate matching function, pcre_dfa_exec(), which matches using
3097 a different (DFA) algorithm. Although it is slower than the original
3098 function, it does have some advantages for certain types of matching
3099 problem.
3101 5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
3102 including restarting after a partial match.
3104 6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
3105 defined when compiling for Windows was sent to me. I have put it into the
3106 code, though I have no means of testing or verifying it.
3108 7. Added the pcre_refcount() auxiliary function.
3110 8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
3111 match before or at the first newline in the subject string. In pcretest,
3112 the /f option on a pattern can be used to set this.
3114 9. A repeated \w when used in UTF-8 mode with characters greater than 256
3115 would behave wrongly. This has been present in PCRE since release 4.0.
3117 10. A number of changes to the pcregrep command:
3119 (a) Refactored how -x works; insert ^(...)$ instead of setting
3120 PCRE_ANCHORED and checking the length, in preparation for adding
3121 something similar for -w.
3123 (b) Added the -w (match as a word) option.
3125 (c) Refactored the way lines are read and buffered so as to have more
3126 than one at a time available.
3128 (d) Implemented a pcregrep test script.
3130 (e) Added the -M (multiline match) option. This allows patterns to match
3131 over several lines of the subject. The buffering ensures that at least
3132 8K, or the rest of the document (whichever is the shorter) is available
3133 for matching (and similarly the previous 8K for lookbehind assertions).
3135 (f) Changed the --help output so that it now says
3137 -w, --word-regex(p)
3139 instead of two lines, one with "regex" and the other with "regexp"
3140 because that confused at least one person since the short forms are the
3141 same. (This required a bit of code, as the output is generated
3142 automatically from a table. It wasn't just a text change.)
3144 (g) -- can be used to terminate pcregrep options if the next thing isn't an
3145 option but starts with a hyphen. Could be a pattern or a path name
3146 starting with a hyphen, for instance.
3148 (h) "-" can be given as a file name to represent stdin.
3150 (i) When file names are being printed, "(standard input)" is used for
3151 the standard input, for compatibility with GNU grep. Previously
3152 "<stdin>" was used.
3154 (j) The option --label=xxx can be used to supply a name to be used for
3155 stdin when file names are being printed. There is no short form.
3157 (k) Re-factored the options decoding logic because we are going to add
3158 two more options that take data. Such options can now be given in four
3159 different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
3161 (l) Added the -A, -B, and -C options for requesting that lines of context
3162 around matches be printed.
3164 (m) Added the -L option to print the names of files that do not contain
3165 any matching lines, that is, the complement of -l.
3167 (n) The return code is 2 if any file cannot be opened, but pcregrep does
3168 continue to scan other files.
3170 (o) The -s option was incorrectly implemented. For compatibility with other
3171 greps, it now suppresses the error message for a non-existent or non-
3172 accessible file (but not the return code). There is a new option called
3173 -q that suppresses the output of matching lines, which was what -s was
3174 previously doing.
3176 (p) Added --include and --exclude options to specify files for inclusion
3177 and exclusion when recursing.
3179 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
3180 Hopefully, it now does.
3182 12. Missing cast in pcre_study().
3184 13. Added an "uninstall" target to the makefile.
3186 14. Replaced "extern" in the function prototypes in Makefile.in with
3187 "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
3188 world, but is set differently for Windows.
3190 15. Added a second compiling function called pcre_compile2(). The only
3191 difference is that it has an extra argument, which is a pointer to an
3192 integer error code. When there is a compile-time failure, this is set
3193 non-zero, in addition to the error test pointer being set to point to an
3194 error message. The new argument may be NULL if no error number is required
3195 (but then you may as well call pcre_compile(), which is now just a
3196 wrapper). This facility is provided because some applications need a
3197 numeric error indication, but it has also enabled me to tidy up the way
3198 compile-time errors are handled in the POSIX wrapper.
3200 16. Added VPATH=.libs to the makefile; this should help when building with one
3201 prefix path and installing with another. (Or so I'm told by someone who
3202 knows more about this stuff than I do.)
3204 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
3205 passes PCRE_DOTALL to the pcre_compile() function, making the "." character
3206 match everything, including newlines. This is not POSIX-compatible, but
3207 somebody wanted the feature. From pcretest it can be activated by using
3208 both the P and the s flags.
3210 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
3212 19. libpcre.pc was being incorrectly installed as executable.
3214 20. A couple of places in pcretest check for end-of-line by looking for '\n';
3215 it now also looks for '\r' so that it will work unmodified on Windows.
3217 21. Added Google's contributed C++ wrapper to the distribution.
3219 22. Added some untidy missing memory free() calls in pcretest, to keep
3220 Electric Fence happy when testing.
3224 Version 5.0 13-Sep-04
3225 ---------------------
3227 1. Internal change: literal characters are no longer packed up into items
3228 containing multiple characters in a single byte-string. Each character
3229 is now matched using a separate opcode. However, there may be more than one
3230 byte in the character in UTF-8 mode.
3232 2. The pcre_callout_block structure has two new fields: pattern_position and
3233 next_item_length. These contain the offset in the pattern to the next match
3234 item, and its length, respectively.
3236 3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
3237 insertion of callouts before each pattern item. Added the /C option to
3238 pcretest to make use of this.
3240 4. On the advice of a Windows user, the lines
3242 #if defined(_WIN32) || defined(WIN32)
3243 _setmode( _fileno( stdout ), 0x8000 );
3244 #endif /* defined(_WIN32) || defined(WIN32) */
3246 have been added to the source of pcretest. This apparently does useful
3247 magic in relation to line terminators.
3249 5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
3250 for the benefit of those environments where the "b" makes a difference.
3252 6. The icc compiler has the same options as gcc, but "configure" doesn't seem
3253 to know about it. I have put a hack into configure.in that adds in code
3254 to set GCC=yes if CC=icc. This seems to end up at a point in the
3255 generated configure script that is early enough to affect the setting of
3256 compiler options, which is what is needed, but I have no means of testing
3257 whether it really works. (The user who reported this had patched the
3258 generated configure script, which of course I cannot do.)
3260 LATER: After change 22 below (new libtool files), the configure script
3261 seems to know about icc (and also ecc). Therefore, I have commented out
3262 this hack in configure.in.
3264 7. Added support for pkg-config (2 patches were sent in).
3266 8. Negated POSIX character classes that used a combination of internal tables
3267 were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
3268 [[:^ascii]]. Typically, they would match almost any characters. The other
3269 POSIX classes were not broken in this way.
3271 9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
3272 to find the match, as PCRE was deluded into thinking that the match had to
3273 start at the start point or following a newline. The same bug applied to
3274 patterns with negative forward assertions or any backward assertions
3275 preceding ".*" at the start, unless the pattern required a fixed first
3276 character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
3278 10. In UTF-8 mode, when moving forwards in the subject after a failed match
3279 starting at the last subject character, bytes beyond the end of the subject
3280 string were read.
3282 11. Renamed the variable "class" as "classbits" to make life easier for C++
3283 users. (Previously there was a macro definition, but it apparently wasn't
3284 enough.)
3286 12. Added the new field "tables" to the extra data so that tables can be passed
3287 in at exec time, or the internal tables can be re-selected. This allows
3288 a compiled regex to be saved and re-used at a later time by a different
3289 program that might have everything at different addresses.
3291 13. Modified the pcre-config script so that, when run on Solaris, it shows a
3292 -R library as well as a -L library.
3294 14. The debugging options of pcretest (-d on the command line or D on a
3295 pattern) showed incorrect output for anything following an extended class
3296 that contained multibyte characters and which was followed by a quantifier.
3298 15. Added optional support for general category Unicode character properties
3299 via the \p, \P, and \X escapes. Unicode property support implies UTF-8
3300 support. It adds about 90K to the size of the library. The meanings of the
3301 inbuilt class escapes such as \d and \s have NOT been changed.
3303 16. Updated pcredemo.c to include calls to free() to release the memory for the
3304 compiled pattern.
3306 17. The generated file chartables.c was being created in the source directory
3307 instead of in the building directory. This caused the build to fail if the
3308 source directory was different from the building directory, and was
3309 read-only.
3311 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
3312 file. No doubt somebody will tell me if they don't make sense... Also added
3313 Dan Mooney's comments about building on OpenVMS.
3315 19. Added support for partial matching via the PCRE_PARTIAL option for
3316 pcre_exec() and the \P data escape in pcretest.
3318 20. Extended pcretest with 3 new pattern features:
3320 (i) A pattern option of the form ">rest-of-line" causes pcretest to
3321 write the compiled pattern to the file whose name is "rest-of-line".
3322 This is a straight binary dump of the data, with the saved pointer to
3323 the character tables forced to be NULL. The study data, if any, is
3324 written too. After writing, pcretest reads a new pattern.
3326 (ii) If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
3327 compiled pattern from the given file. There must not be any
3328 occurrences of "<" in the file name (pretty unlikely); if there are,
3329 pcretest will instead treat the initial "<" as a pattern delimiter.
3330 After reading in the pattern, pcretest goes on to read data lines as
3331 usual.
3333 (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
3334 and 16-bit fields in a compiled pattern, to simulate a pattern that
3335 was compiled on a host of opposite endianness.
3337 21. The pcre-exec() function can now cope with patterns that were compiled on
3338 hosts of opposite endianness, with this restriction:
3340 As for any compiled expression that is saved and used later, the tables
3341 pointer field cannot be preserved; the extra_data field in the arguments
3342 to pcre_exec() should be used to pass in a tables address if a value
3343 other than the default internal tables were used at compile time.
3345 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
3346 now diagnosed as an error. Previously, most of the time, a negative number
3347 would have been treated as zero, but if in addition "ovector" was passed as
3348 NULL, a crash could occur.
3350 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
3351 new versions from the libtool 1.5 distribution (the last one is a copy of
3352 a file called libtool.m4). This seems to have fixed the need to patch
3353 "configure" to support Darwin 1.3 (which I used to do). However, I still
3354 had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
3355 workstation).
3357 24. Changed the PCRE licence to be the more standard "BSD" licence.
3360 Version 4.5 01-Dec-03
3361 ---------------------
3363 1. There has been some re-arrangement of the code for the match() function so
3364 that it can be compiled in a version that does not call itself recursively.
3365 Instead, it keeps those local variables that need separate instances for
3366 each "recursion" in a frame on the heap, and gets/frees frames whenever it
3367 needs to "recurse". Keeping track of where control must go is done by means
3368 of setjmp/longjmp. The whole thing is implemented by a set of macros that
3369 hide most of the details from the main code, and operates only if
3370 NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
3371 "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
3372 operating.
3374 To make it easier for callers to provide specially tailored get/free
3375 functions for this usage, two new functions, pcre_stack_malloc, and
3376 pcre_stack_free, are used. They are always called in strict stacking order,
3377 and the size of block requested is always the same.
3379 The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
3380 PCRE has been compiled to use the stack or the heap for recursion. The
3381 -C option of pcretest uses this to show which version is compiled.
3383 A new data escape \S, is added to pcretest; it causes the amounts of store
3384 obtained and freed by both kinds of malloc/free at match time to be added
3385 to the output.
3387 2. Changed the locale test to use "fr_FR" instead of "fr" because that's
3388 what's available on my current Linux desktop machine.
3390 3. When matching a UTF-8 string, the test for a valid string at the start has
3391 been extended. If start_offset is not zero, PCRE now checks that it points
3392 to a byte that is the start of a UTF-8 character. If not, it returns
3393 PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
3394 this is necessary because there may be backward assertions in the pattern.
3395 When matching the same subject several times, it may save resources to use
3396 PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
3398 4. The code for checking the validity of UTF-8 strings has been tightened so
3399 that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
3400 containing "overlong sequences".
3402 5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
3403 I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
3404 should have been "&", but it just so happened that all the cases this let
3405 through by mistake were picked up later in the function.
3407 6. I had used a variable called "isblank" - this is a C99 function, causing
3408 some compilers to warn. To avoid this, I renamed it (as "blankclass").
3410 7. Cosmetic: (a) only output another newline at the end of pcretest if it is
3411 prompting; (b) run "./pcretest /dev/null" at the start of the test script
3412 so the version is shown; (c) stop "make test" echoing "./RunTest".
3414 8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
3416 9. The prototype for memmove() for systems that don't have it was using
3417 size_t, but the inclusion of the header that defines size_t was later. I've
3418 moved the #includes for the C headers earlier to avoid this.
3420 10. Added some adjustments to the code to make it easier to compiler on certain
3421 special systems:
3423 (a) Some "const" qualifiers were missing.
3424 (b) Added the macro EXPORT before all exported functions; by default this
3425 is defined to be empty.
3426 (c) Changed the dftables auxiliary program (that builds chartables.c) so
3427 that it reads its output file name as an argument instead of writing
3428 to the standard output and assuming this can be redirected.
3430 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
3431 class containing characters with values greater than 255, PCRE compilation
3432 went into a loop.
3434 12. A recursive reference to a subpattern that was within another subpattern
3435 that had a minimum quantifier of zero caused PCRE to crash. For example,
3436 (x(y(?2))z)? provoked this bug with a subject that got as far as the
3437 recursion. If the recursively-called subpattern itself had a zero repeat,
3438 that was OK.
3440 13. In pcretest, the buffer for reading a data line was set at 30K, but the
3441 buffer into which it was copied (for escape processing) was still set at
3442 1024, so long lines caused crashes.
3444 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
3445 "internal error: code overflow...". This applied to any character class
3446 that was followed by a possessive quantifier.
3448 15. Modified the Makefile to add libpcre.la as a prerequisite for
3449 libpcreposix.la because I was told this is needed for a parallel build to
3450 work.
3452 16. If a pattern that contained .* following optional items at the start was
3453 studied, the wrong optimizing data was generated, leading to matching
3454 errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
3455 matching string must start with a or b or c. The correct conclusion for
3456 this pattern is that a match can start with any character.
3459 Version 4.4 13-Aug-03
3460 ---------------------
3462 1. In UTF-8 mode, a character class containing characters with values between
3463 127 and 255 was not handled correctly if the compiled pattern was studied.
3464 In fixing this, I have also improved the studying algorithm for such
3465 classes (slightly).
3467 2. Three internal functions had redundant arguments passed to them. Removal
3468 might give a very teeny performance improvement.
3470 3. Documentation bug: the value of the capture_top field in a callout is *one
3471 more than* the number of the hightest numbered captured substring.
3473 4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
3474 in incorrectly linking with a previously installed version. They now link
3475 explicitly with libpcre.la.
3477 5. configure.in no longer needs to recognize Cygwin specially.
3479 6. A problem in pcre.in for Windows platforms is fixed.
3481 7. If a pattern was successfully studied, and the -d (or /D) flag was given to
3482 pcretest, it used to include the size of the study block as part of its
3483 output. Unfortunately, the structure contains a field that has a different
3484 size on different hardware architectures. This meant that the tests that
3485 showed this size failed. As the block is currently always of a fixed size,
3486 this information isn't actually particularly useful in pcretest output, so
3487 I have just removed it.
3489 8. Three pre-processor statements accidentally did not start in column 1.
3490 Sadly, there are *still* compilers around that complain, even though
3491 standard C has not required this for well over a decade. Sigh.
3493 9. In pcretest, the code for checking callouts passed small integers in the
3494 callout_data field, which is a void * field. However, some picky compilers
3495 complained about the casts involved for this on 64-bit systems. Now
3496 pcretest passes the address of the small integer instead, which should get
3497 rid of the warnings.
3499 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
3500 both compile and run time, and gives an error if an invalid UTF-8 sequence
3501 is found. There is a option for disabling this check in cases where the
3502 string is known to be correct and/or the maximum performance is wanted.
3504 11. In response to a bug report, I changed one line in Makefile.in from
3506 -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
3507 to
3508 -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
3510 to look similar to other lines, but I have no way of telling whether this
3511 is the right thing to do, as I do not use Windows. No doubt I'll get told
3512 if it's wrong...
3515 Version 4.3 21-May-03
3516 ---------------------
3518 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
3519 Makefile.
3521 2. Some refactoring to improve the quality of the code:
3523 (i) The utf8_table... variables are now declared "const".
3525 (ii) The code for \cx, which used the "case flipping" table to upper case
3526 lower case letters, now just substracts 32. This is ASCII-specific,
3527 but the whole concept of \cx is ASCII-specific, so it seems
3528 reasonable.
3530 (iii) PCRE was using its character types table to recognize decimal and
3531 hexadecimal digits in the pattern. This is silly, because it handles
3532 only 0-9, a-f, and A-F, but the character types table is locale-
3533 specific, which means strange things might happen. A private
3534 table is now used for this - though it costs 256 bytes, a table is
3535 much faster than multiple explicit tests. Of course, the standard
3536 character types table is still used for matching digits in subject
3537 strings against \d.
3539 (iv) Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
3540 ending in _t are). So I've renamed it as ESC_tee.
3542 3. The first argument for regexec() in the POSIX wrapper should have been
3543 defined as "const".
3545 4. Changed pcretest to use malloc() for its buffers so that they can be
3546 Electric Fenced for debugging.
3548 5. There were several places in the code where, in UTF-8 mode, PCRE would try
3549 to read one or more bytes before the start of the subject string. Often this
3550 had no effect on PCRE's behaviour, but in some circumstances it could
3551 provoke a segmentation fault.
3553 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
3554 to try to read one or more bytes before the start of the subject string.
3556 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
3557 UTF-8 support could misbehave in various ways if the subject string
3558 contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
3559 area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
3560 back over UTF-8 characters.)
3563 Version 4.2 14-Apr-03
3564 ---------------------
3566 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
3568 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
3569 [ON_WINDOWS]: new variable, "#" on non-Windows platforms
3570 [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
3571 [WIN_PREFIX]: new variable, "cyg" for Cygwin
3572 * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
3574 Note: automatic setting of the BUILD variables is not yet working
3575 set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
3576 compile-time but not at link-time
3577 [LINK]: use for linking executables only
3578 make different versions for Windows and non-Windows
3579 [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
3580 libraries
3581 [LINK_FOR_BUILD]: new variable
3582 [OBJEXT]: use throughout
3583 [EXEEXT]: use throughout
3584 <winshared>: new target
3585 <wininstall>: new target
3586 <dftables.o>: use native compiler
3587 <dftables>: use native linker
3588 <install>: handle Windows platform correctly
3589 <clean>: ditto
3590 <check>: ditto
3591 copy DLL to top builddir before testing
3593 As part of these changes, -no-undefined was removed again. This was reported
3594 to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
3595 in any case.
3597 3. Some tidies to get rid of compiler warnings:
3599 . In the match_data structure, match_limit was an unsigned long int, whereas
3600 match_call_count was an int. I've made them both unsigned long ints.
3602 . In pcretest the fact that a const uschar * doesn't automatically cast to
3603 a void * provoked a warning.
3605 . Turning on some more compiler warnings threw up some "shadow" variables
3606 and a few more missing casts.
3608 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
3609 option, a class that contained a single character with a value between 128
3610 and 255 (e.g. /[\xFF]/) caused PCRE to crash.
3612 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
3613 option, a class that contained several characters, but with at least one
3614 whose value was between 128 and 255 caused PCRE to crash.
3617 Version 4.1 12-Mar-03
3618 ---------------------
3620 1. Compiling with gcc -pedantic found a couple of places where casts were
3621 needed, and a string in dftables.c that was longer than standard compilers are
3622 required to support.
3624 2. Compiling with Sun's compiler found a few more places where the code could
3625 be tidied up in order to avoid warnings.
3627 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
3628 first of these names is deprecated in the latest Autoconf in favour of the name
3629 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
3630 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
3631 analogy I have changed it to CFLAGS_FOR_BUILD.
3633 4. Added -no-undefined to the linking command in the Makefile, because this is
3634 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
3635 linking step for the pcreposix library.
3637 5. PCRE was failing to diagnose the case of two named groups with the same
3638 name.
3640 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
3641 literal character that is needed in the subject for a match, and scans along to
3642 ensure that it is present before embarking on the full matching process. This
3643 saves time in cases of nested unlimited repeats that are never going to match.
3644 Problem: the scan can take a lot of time if the subject is very long (e.g.
3645 megabytes), thus penalizing straightforward matches. It is now done only if the
3646 amount of subject to be scanned is less than 1000 bytes.
3648 7. A lesser problem with the same optimization is that it was recording the
3649 first character of an anchored pattern as "needed", thus provoking a search
3650 right along the subject, even when the first match of the pattern was going to
3651 fail. The "needed" character is now not set for anchored patterns, unless it
3652 follows something in the pattern that is of non-fixed length. Thus, it still
3653 fulfils its original purpose of finding quick non-matches in cases of nested
3654 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
3657 Version 4.0 17-Feb-03
3658 ---------------------
3660 1. If a comment in an extended regex that started immediately after a meta-item
3661 extended to the end of string, PCRE compiled incorrect data. This could lead to
3662 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
3664 2. Moved to autoconf 2.53 and libtool 1.4.2.
3666 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
3667 the special perltest8 script is no longer needed - all the tests can be run
3668 from a single perltest script.
3670 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
3671 by \s. It has now been removed in PCRE. This means it isn't recognized as
3672 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
3673 class [:space:] *does* include VT, thereby creating a mess.
3675 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
3676 space and tab.
3678 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
3679 its new features into the main test script, reducing the number of scripts.
3681 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
3682 were backward compatible, and made the (?i) apply to the whole pattern, as if
3683 /i were given. Now it behaves more logically, and applies the option setting
3684 only to what follows. PCRE has been changed to follow suit. However, if it
3685 finds options settings right at the start of the pattern, it extracts them into
3686 the global options, as before. Thus, they show up in the info data.
3688 8. Added support for the \Q...\E escape sequence. Characters in between are
3689 treated as literals. This is slightly different from Perl in that $ and @ are
3690 also handled as literals inside the quotes. In Perl, they will cause variable
3691 interpolation. Note the following examples:
3693 Pattern PCRE matches Perl matches
3695 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
3696 \Qabc\$xyz\E abc\$xyz abc\$xyz
3697 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
3699 For compatibility with Perl, \Q...\E sequences are recognized inside character
3700 classes as well as outside them.
3702 9. Re-organized 3 code statements in pcretest to avoid "overflow in
3703 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
3704 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
3705 signed/unsigned warnings.
3707 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
3708 option for pcretest, so I've replaced it by a simple function that does just
3709 that job.
3711 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
3712 "pcregrep -".
3714 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
3715 Java package. This provides some syntactic sugar for simple cases of what my
3716 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
3717 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
3718 item, you can use this simplified notation. Note that only makes sense with
3719 greedy quantifiers. Consequently, the use of the possessive quantifier forces
3720 greediness, whatever the setting of the PCRE_UNGREEDY option.
3722 13. A change of greediness default within a pattern was not taking effect at
3723 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
3724 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
3725 was abstracted outside.
3727 14. PCRE now supports the \G assertion. It is true when the current matching
3728 position is at the start point of the match. This differs from \A when the
3729 starting offset is non-zero. Used with the /g option of pcretest (or similar
3730 code), it works in the same way as it does for Perl's /g option. If all
3731 alternatives of a regex begin with \G, the expression is anchored to the start
3732 match position, and the "anchored" flag is set in the compiled expression.
3734 15. Some bugs concerning the handling of certain option changes within patterns
3735 have been fixed. These applied to options other than (?ims). For example,
3736 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
3737 been the other way round. Some of this was related to change 7 above.
3739 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
3740 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
3741 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
3742 POSIX classes only within a class (e.g. /[[:alpha:]]/).
3744 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
3745 mode. Unlike ".", it always matches newline, whatever the setting of
3746 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
3747 assertions. Perl allows it, but it doesn't (in general) work because it can't
3748 calculate the length of the lookbehind. At least, that's the case for Perl
3749 5.8.0 - I've been told they are going to document that it doesn't work in
3750 future.
3752 18. Added an error diagnosis for escapes that PCRE does not support: these are
3753 \L, \l, \N, \P, \p, \U, \u, and \X.
3755 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
3756 reading past the end of the pattern in cases such as /[abcd/.
3758 20. PCRE was getting more memory than necessary for patterns with classes that
3759 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
3761 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
3762 compiling PCRE for use with Virtual Pascal.
3764 22. Small fix to the Makefile to make it work properly if the build is done
3765 outside the source tree.
3767 23. Added a new extension: a condition to go with recursion. If a conditional
3768 subpattern starts with (?(R) the "true" branch is used if recursion has
3769 happened, whereas the "false" branch is used only at the top level.
3771 24. When there was a very long string of literal characters (over 255 bytes
3772 without UTF support, over 250 bytes with UTF support), the computation of how
3773 much memory was required could be incorrect, leading to segfaults or other
3774 strange effects.
3776 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
3777 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
3778 there was a subsequent back reference to those brackets. This meant that, for
3779 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
3780 possible to check for precisely this case. All we can do is abandon the
3781 optimization if .* occurs inside capturing brackets when there are any back
3782 references whatsoever. (See below for a better fix that came later.)
3784 26. The handling of the optimization for finding the first character of a
3785 non-anchored pattern, and for finding a character that is required later in the
3786 match were failing in some cases. This didn't break the matching; it just
3787 failed to optimize when it could. The way this is done has been re-implemented.
3789 27. Fixed typo in error message for invalid (?R item (it said "(?p").
3791 28. Added a new feature that provides some of the functionality that Perl
3792 provides with (?{...}). The facility is termed a "callout". The way it is done
3793 in PCRE is for the caller to provide an optional function, by setting
3794 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
3795 global variable. By default it is unset, which disables all calling out. To get
3796 the function called, the regex must include (?C) at appropriate points. This
3797 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
3798 This provides a means of identifying different callout points. When PCRE
3799 reaches such a point in the regex, if pcre_callout has been set, the external
3800 function is called. It is provided with data in a structure called
3801 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
3802 matching continues; if it returns a non-zero value, the match at the current
3803 point fails. However, backtracking will occur if possible. [This was changed
3804 later and other features added - see item 49 below.]
3806 29. pcretest is upgraded to test the callout functionality. It provides a
3807 callout function that displays information. By default, it shows the start of
3808 the match and the current position in the text. There are some new data escapes
3809 to vary what happens:
3811 \C+ in addition, show current contents of captured substrings
3812 \C- do not supply a callout function
3813 \C!n return 1 when callout number n is reached
3814 \C!n!m return 1 when callout number n is reached for the mth time
3816 30. If pcregrep was called with the -l option and just a single file name, it
3817 output "<stdin>" if a match was found, instead of the file name.
3819 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
3820 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
3821 pcre_exec(). This saves a malloc/free per call. The default value of
3822 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
3823 when configuring.
3825 32. The default maximum size of a compiled pattern is 64K. There have been a
3826 few cases of people hitting this limit. The code now uses macros to handle the
3827 storing of links as offsets within the compiled pattern. It defaults to 2-byte
3828 links, but this can be changed to 3 or 4 bytes by --with-link-size when
3829 configuring. Tests 2 and 5 work only with 2-byte links because they output
3830 debugging information about compiled patterns.
3832 33. Internal code re-arrangements:
3834 (a) Moved the debugging function for printing out a compiled regex into
3835 its own source file (printint.c) and used #include to pull it into
3836 pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
3837 separate copies.
3839 (b) Defined the list of op-code names for debugging as a macro in
3840 internal.h so that it is next to the definition of the opcodes.
3842 (c) Defined a table of op-code lengths for simpler skipping along compiled
3843 code. This is again a macro in internal.h so that it is next to the
3844 definition of the opcodes.
3846 34. Added support for recursive calls to individual subpatterns, along the
3847 lines of Robin Houston's patch (but implemented somewhat differently).
3849 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
3850 allow it to read and process whole directories in Win32. This code was
3851 contributed by Lionel Fourquaux; it has not been tested by me.
3853 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
3854 used to name a group. Names consist of alphanumerics and underscores, and must
3855 be unique. Back references use the syntax (?P=name) and recursive calls use
3856 (?P>name) which is a PCRE extension to the Python extension. Groups still have
3857 numbers. The function pcre_fullinfo() can be used after compilation to extract
3858 a name/number map. There are three relevant calls:
3860 PCRE_INFO_NAMEENTRYSIZE yields the size of each entry in the map
3861 PCRE_INFO_NAMECOUNT yields the number of entries
3862 PCRE_INFO_NAMETABLE yields a pointer to the map.
3864 The map is a vector of fixed-size entries. The size of each entry depends on
3865 the length of the longest name used. The first two bytes of each entry are the
3866 group number, most significant byte first. There follows the corresponding
3867 name, zero terminated. The names are in alphabetical order.
3869 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
3870 case instead of 255. Making it the same both with and without UTF-8 support
3871 means that the same test output works with both.
3873 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
3874 calling malloc() with a zero argument.
3876 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
3877 optimization. I've improved things by keeping a bitmap of backreferences with
3878 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
3879 fact referenced, the optimization can be applied. It is unlikely that a
3880 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
3881 the match to follow \n) will appear inside brackets with a number greater than
3882 31, but if it does, any back reference > 31 suppresses the optimization.
3884 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
3885 of disabling numbered capturing parentheses. Any opening parenthesis that is
3886 not followed by ? behaves as if it were followed by ?: but named parentheses
3887 can still be used for capturing (and they will acquire numbers in the usual
3888 way).
3890 41. Redesigned the return codes from the match() function into yes/no/error so
3891 that errors can be passed back from deep inside the nested calls. A malloc
3892 failure while inside a recursive subpattern call now causes the
3893 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
3895 42. It is now possible to set a limit on the number of times the match()
3896 function is called in a call to pcre_exec(). This facility makes it possible to
3897 limit the amount of recursion and backtracking, though not in a directly
3898 obvious way, because the match() function is used in a number of different
3899 circumstances. The count starts from zero for each position in the subject
3900 string (for non-anchored patterns). The default limit is, for compatibility, a
3901 large number, namely 10 000 000. You can change this in two ways:
3903 (a) When configuring PCRE before making, you can use --with-match-limit=n
3904 to set a default value for the compiled library.
3906 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
3907 a different value is set. See 45 below.
3909 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
3911 43. Added a new function pcre_config(int, void *) to enable run-time extraction
3912 of things that can be changed at compile time. The first argument specifies
3913 what is wanted and the second points to where the information is to be placed.
3914 The current list of available information is:
3918 The output is an integer that is set to one if UTF-8 support is available;
3919 otherwise it is set to zero.
3923 The output is an integer that it set to the value of the code that is used for
3924 newline. It is either LF (10) or CR (13).
3928 The output is an integer that contains the number of bytes used for internal
3929 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
3933 The output is an integer that contains the threshold above which the POSIX
3934 interface uses malloc() for output vectors. See item 31 above.
3938 The output is an unsigned integer that contains the default limit of the number
3939 of match() calls in a pcre_exec() execution. See 42 above.
3941 44. pcretest has been upgraded by the addition of the -C option. This causes it
3942 to extract all the available output from the new pcre_config() function, and to
3943 output it. The program then exits immediately.
3945 45. A need has arisen to pass over additional data with calls to pcre_exec() in
3946 order to support additional features. One way would have been to define
3947 pcre_exec2() (for example) with extra arguments, but this would not have been
3948 extensible, and would also have required all calls to the original function to
3949 be mapped to the new one. Instead, I have chosen to extend the mechanism that
3950 is used for passing in "extra" data from pcre_study().
3952 The pcre_extra structure is now exposed and defined in pcre.h. It currently
3953 contains the following fields:
3955 flags a bitmap indicating which of the following fields are set
3956 study_data opaque data from pcre_study()
3957 match_limit a way of specifying a limit on match() calls for a specific
3958 call to pcre_exec()
3959 callout_data data for callouts (see 49 below)
3961 The flag bits are also defined in pcre.h, and are
3967 The pcre_study() function now returns one of these new pcre_extra blocks, with
3968 the actual study data pointed to by the study_data field, and the
3969 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
3970 before. That is, this change is entirely upwards-compatible and requires no
3971 change to existing code.
3973 If you want to pass in additional data to pcre_exec(), you can either place it
3974 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
3975 block.
3977 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
3978 data string contains the escape sequence \M, pcretest calls pcre_exec() several
3979 times with different match limits, until it finds the minimum value needed for
3980 pcre_exec() to complete. The value is then output. This can be instructive; for
3981 most simple matches the number is quite small, but for pathological cases it
3982 gets very large very quickly.
3984 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
3985 returns the size of the data block pointed to by the study_data field in a
3986 pcre_extra block, that is, the value that was passed as the argument to
3987 pcre_malloc() when PCRE was getting memory in which to place the information
3988 created by pcre_study(). The fourth argument should point to a size_t variable.
3989 pcretest has been extended so that this information is shown after a successful
3990 pcre_study() call when information about the compiled regex is being displayed.
3992 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
3993 because what follows is always an absolute path. (Later: it turns out that this
3994 is more than cosmetic for MinGW, because it doesn't like empty path
3995 components.)
3997 49. Some changes have been made to the callout feature (see 28 above):
3999 (i) A callout function now has three choices for what it returns:
4001 0 => success, carry on matching
4002 > 0 => failure at this point, but backtrack if possible
4003 < 0 => serious error, return this value from pcre_exec()
4005 Negative values should normally be chosen from the set of PCRE_ERROR_xxx
4006 values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
4007 "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
4008 use by callout functions. It will never be used by PCRE itself.
4010 (ii) The pcre_extra structure (see 45 above) has a void * field called
4011 callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
4012 pcre_callout_block structure has a field of the same name. The contents of
4013 the field passed in the pcre_extra structure are passed to the callout
4014 function in the corresponding field in the callout block. This makes it
4015 easier to use the same callout-containing regex from multiple threads. For
4016 testing, the pcretest program has a new data escape
4018 \C*n pass the number n (may be negative) as callout_data
4020 If the callout function in pcretest receives a non-zero value as
4021 callout_data, it returns that value.
4023 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
4024 there were some redundant $(CFLAGS) in commands that are now specified as
4025 $(LINK), which already includes $(CFLAGS).
4027 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
4028 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
4029 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
4030 one-byte characters throughout. Note that case-insensitive matching applies
4031 only to characters whose values are less than 256. PCRE doesn't support the
4032 notion of cases for higher-valued characters.
4034 (i) A character class whose characters are all within 0-255 is handled as
4035 a bit map, and the map is inverted for negative classes. Previously, a
4036 character > 255 always failed to match such a class; however it should
4037 match if the class was a negative one (e.g. [^ab]). This has been fixed.
4039 (ii) A negated character class with a single character < 255 is coded as
4040 "not this character" (OP_NOT). This wasn't working properly when the test
4041 character was multibyte, either singly or repeated.
4043 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
4044 mode, for example: \x{100}{2,3}.
4046 (iv) The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
4047 singly or repeated) now correctly test multibyte characters. However,
4048 PCRE doesn't recognize any characters with values greater than 255 as
4049 digits, spaces, or word characters. Such characters always match \D, \S,
4050 and \W, and never match \d, \s, or \w.
4052 (v) Classes may now contain characters and character ranges with values
4053 greater than 255. For example: [ab\x{100}-\x{400}].
4055 (vi) pcregrep now has a --utf-8 option (synonym -u) which makes it call
4056 PCRE in UTF-8 mode.
4058 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
4059 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
4060 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
4061 value.)
4063 53. The single man page has become too large. I have therefore split it up into
4064 a number of separate man pages. These also give rise to individual HTML pages;
4065 these are now put in a separate directory, and there is an index.html page that
4066 lists them all. Some hyperlinking between the pages has been installed.
4068 54. Added convenience functions for handling named capturing parentheses.
4070 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
4071 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
4072 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
4073 are faulted.
4075 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
4076 calling configure. These values are used when compiling the dftables.c program
4077 which is run to generate the source of the default character tables. They
4078 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
4079 you will need to set these values.
4081 57. Updated the building process for Windows DLL, as provided by Fred Cox.
4084 Version 3.9 02-Jan-02
4085 ---------------------
4087 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
4089 2. If --disable-static was given, the building process failed when trying to
4090 build pcretest and pcregrep. (For some reason it was using libtool to compile
4091 them, which is not right, as they aren't part of the library.)
4094 Version 3.8 18-Dec-01
4095 ---------------------
4097 1. The experimental UTF-8 code was completely screwed up. It was packing the
4098 bytes in the wrong order. How dumb can you get?
4101 Version 3.7 29-Oct-01
4102 ---------------------
4104 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
4105 This caused pcretest, when used on the test data, to segfault. Unfortunately,
4106 this didn't happen under Solaris 8, where I normally test things.
4108 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
4109 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
4110 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
4111 this fix an hour or so after the initial 3.7 release.)
4114 Version 3.6 23-Oct-01
4115 ---------------------
4117 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
4118 offsets passed as NULL with zero offset count.
4120 2. The config.guess and config.sub files had not been updated when I moved to
4121 the latest autoconf.
4124 Version 3.5 15-Aug-01
4125 ---------------------
4127 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
4128 had been forgotten.
4130 2. By using declared but undefined structures, we can avoid using "void"
4131 definitions in pcre.h while keeping the internal definitions of the structures
4132 private.
4134 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
4135 user point of view, this means that both static and shared libraries are built
4136 by default, but this can be individually controlled. More of the work of
4137 handling this static/shared cases is now inside libtool instead of PCRE's make
4138 file.
4140 4. The pcretest utility is now installed along with pcregrep because it is
4141 useful for users (to test regexs) and by doing this, it automatically gets
4142 relinked by libtool. The documentation has been turned into a man page, so
4143 there are now .1, .txt, and .html versions in /doc.
4145 5. Upgrades to pcregrep:
4146 (i) Added long-form option names like gnu grep.
4147 (ii) Added --help to list all options with an explanatory phrase.
4148 (iii) Added -r, --recursive to recurse into sub-directories.
4149 (iv) Added -f, --file to read patterns from a file.
4151 6. pcre_exec() was referring to its "code" argument before testing that
4152 argument for NULL (and giving an error if it was NULL).
4154 7. Upgraded Makefile.in to allow for compiling in a different directory from
4155 the source directory.
4157 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
4158 options bits, the pointer it was passed was to an int instead of to an unsigned
4159 long int. This mattered only on 64-bit systems.
4161 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
4162 generated) instead of pcre.in, which it its source. Also made the same change
4163 in several of the .c files.
4165 10. A new release of gcc defines printf() as a macro, which broke pcretest
4166 because it had an ifdef in the middle of a string argument for printf(). Fixed
4167 by using separate calls to printf().
4169 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
4170 script, to force use of CR or LF instead of \n in the source. On non-Unix
4171 systems, the value can be set in config.h.
4173 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
4174 absolute limit. Changed the text of the error message to make this clear, and
4175 likewise updated the man page.
4177 13. The limit of 99 on the number of capturing subpatterns has been removed.
4178 The new limit is 65535, which I hope will not be a "real" limit.
4181 Version 3.4 22-Aug-00
4182 ---------------------
4184 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
4186 2. Diagnose condition (?(0) as an error instead of crashing on matching.
4189 Version 3.3 01-Aug-00
4190 ---------------------
4192 1. If an octal character was given, but the value was greater than \377, it
4193 was not getting masked to the least significant bits, as documented. This could
4194 lead to crashes in some systems.
4196 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
4197 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
4199 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
4200 These just pass their arguments on to (pcre_free)(), but they are provided
4201 because some uses of PCRE bind it to non-C systems that can call its functions,
4202 but cannot call free() or pcre_free() directly.
4204 4. Add "make test" as a synonym for "make check". Corrected some comments in
4205 the Makefile.
4207 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
4208 Makefile.
4210 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
4211 command called pgrep for grepping around the active processes.
4213 7. Added the beginnings of support for UTF-8 character strings.
4215 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
4216 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
4217 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
4218 out for the ar command.)
4221 Version 3.2 12-May-00
4222 ---------------------
4224 This is purely a bug fixing release.
4226 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
4227 of ZA. This was just one example of several cases that could provoke this bug,
4228 which was introduced by change 9 of version 2.00. The code for breaking
4229 infinite loops after an iteration that matches an empty string was't working
4230 correctly.
4232 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
4233 when matched against abbab (for example). After matching an empty string, it
4234 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
4235 caused it to match further down the string than it should.
4237 3. The code contained an inclusion of sys/types.h. It isn't clear why this
4238 was there because it doesn't seem to be needed, and it causes trouble on some
4239 systems, as it is not a Standard C header. It has been removed.
4241 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
4242 were reported on the Macintosh. The changes were from
4244 while ((c = *(++ptr)) != 0 && c != '\n');
4245 to
4246 while ((c = *(++ptr)) != 0 && c != '\n') ;
4248 Totally extraordinary, but if that's what it takes...
4250 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
4251 available. Added HAVE_BCOPY and an autoconf test for it; if neither
4252 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
4253 assumes the way PCRE uses memmove() (always moving upwards).
4255 6. PCRE is being used in one environment where strchr() is not available. There
4256 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
4257 faster code anyway.
4260 Version 3.1 09-Feb-00
4261 ---------------------
4263 The only change in this release is the fixing of some bugs in Makefile.in for
4264 the "install" target:
4266 (1) It was failing to install pcreposix.h.
4268 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
4271 Version 3.0 01-Feb-00
4272 ---------------------
4274 1. Add support for the /+ modifier to perltest (to output $` like it does in
4275 pcretest).
4277 2. Add support for the /g modifier to perltest.
4279 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
4280 matches null strings.
4282 4. Fix perltest so that it doesn't do unwanted things when fed an empty
4283 pattern. Perl treats empty patterns specially - it reuses the most recent
4284 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
4285 effect.
4287 5. The POSIX interface was broken in that it was just handing over the POSIX
4288 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
4289 required a bigger vector, with some working space on the end. This means that
4290 the POSIX wrapper now has to get and free some memory, and copy the results.
4292 6. Added some simple autoconf support, placing the test data and the
4293 documentation in separate directories, re-organizing some of the
4294 information files, and making it build pcre-config (a GNU standard). Also added
4295 libtool support for building PCRE as a shared library, which is now the
4296 default.
4298 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
4299 09 are not valid octal constants. Single digits will be used for minor values
4300 less than 10.
4302 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
4303 existing programs that set these in the POSIX interface can use PCRE without
4304 modification.
4306 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
4307 return all that pcre_info() returns, plus additional data. The pcre_info()
4308 function is retained for compatibility, but is considered to be obsolete.
4310 10. Added experimental recursion feature (?R) to handle one common case that
4311 Perl 5.6 will be able to do with (?p{...}).
4313 11. Added support for POSIX character classes like [:alpha:], which Perl is
4314 adopting.
4317 Version 2.08 31-Aug-99
4318 ----------------------
4320 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
4321 trying to match at the startoffset position, but instead was moving forward to
4322 the next newline as if a previous match had failed.
4324 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
4325 and could get into a loop if a null string was matched other than at the start
4326 of the subject.
4328 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
4329 be distinguished at compile time, and for completeness also added PCRE_DATE.
4331 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
4332 in GnuWin32 environments.
4335 Version 2.07 29-Jul-99
4336 ----------------------
4338 1. The documentation is now supplied in plain text form and HTML as well as in
4339 the form of man page sources.
4341 2. C++ compilers don't like assigning (void *) values to other pointer types.
4342 In particular this affects malloc(). Although there is no problem in Standard
4343 C, I've put in casts to keep C++ compilers happy.
4345 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
4346 should be (const char *).
4348 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
4349 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
4350 However, I haven't made this a standard facility. The documentation doesn't
4351 mention it, and the Makefile doesn't support it.
4353 5. The Makefile now contains an "install" target, with editable destinations at
4354 the top of the file. The pcretest program is not installed.
4356 6. pgrep -V now gives the PCRE version number and date.
4358 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
4359 causing the entire string to be ignored, instead of just the last character.
4361 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
4362 non-matching string, it can take a very, very long time, even for strings of
4363 quite modest length, because of the nested recursion. PCRE now does better in
4364 some of these cases. It does this by remembering the last required literal
4365 character in the pattern, and pre-searching the subject to ensure it is present
4366 before running the real match. In other words, it applies a heuristic to detect
4367 some types of certain failure quickly, and in the above example, if presented
4368 with a string that has no trailing " it gives "no match" very quickly.
4370 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
4371 other alternatives are tried instead.
4374 Version 2.06 09-Jun-99
4375 ----------------------
4377 1. Change pcretest's output for amount of store used to show just the code
4378 space, because the remainder (the data block) varies in size between 32-bit and
4379 64-bit systems.
4381 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
4382 start matching at. This allows lookbehinds to work when searching for multiple
4383 occurrences in a string.
4385 3. Added additional options to pcretest for testing multiple occurrences:
4387 /+ outputs the rest of the string that follows a match
4388 /g loops for multiple occurrences, using the new startoffset argument
4389 /G loops for multiple occurrences by passing an incremented pointer
4391 4. PCRE wasn't doing the "first character" optimization for patterns starting
4392 with \b or \B, though it was doing it for other lookbehind assertions. That is,
4393 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
4394 the letter 'x'. On long subject strings, this gives a significant speed-up.
4397 Version 2.05 21-Apr-99
4398 ----------------------
4400 1. Changed the type of magic_number from int to long int so that it works
4401 properly on 16-bit systems.
4403 2. Fixed a bug which caused patterns starting with .* not to work correctly
4404 when the subject string contained newline characters. PCRE was assuming
4405 anchoring for such patterns in all cases, which is not correct because .* will
4406 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
4407 DOTALL is set at top level; otherwise it knows that patterns starting with .*
4408 must be retried after every newline in the subject.
4411 Version 2.04 18-Feb-99
4412 ----------------------
4414 1. For parenthesized subpatterns with repeats whose minimum was zero, the
4415 computation of the store needed to hold the pattern was incorrect (too large).
4416 If such patterns were nested a few deep, this could multiply and become a real
4417 problem.
4419 2. Added /M option to pcretest to show the memory requirement of a specific
4420 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
4422 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
4423 compiled in such a way that the backtracking after subsequent failure was
4424 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
4425 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
4428 Version 2.03 02-Feb-99
4429 ----------------------
4431 1. Fixed typo and small mistake in man page.
4433 2. Added 4th condition (GPL supersedes if conflict) and created separate
4434 LICENCE file containing the conditions.
4436 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
4437 Perl, that is the internal \ allows the delimiter to be included in the
4438 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
4439 the final delimiter, add \ to the end of the pattern (to test the error).
4441 4. Added the convenience functions for extracting substrings after a successful
4442 match. Updated pcretest to make it able to test these functions.
4445 Version 2.02 14-Jan-99
4446 ----------------------
4448 1. Initialized the working variables associated with each extraction so that
4449 their saving and restoring doesn't refer to uninitialized store.
4451 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
4452 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
4453 fix the problem.
4455 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
4456 calls, and wasn't printing the correct value for compiling calls. Increased the
4457 default value of LOOPREPEAT, and the number of significant figures in the
4458 times.
4460 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
4462 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
4463 a building problem on Windows NT with a FAT file system.
4466 Version 2.01 21-Oct-98
4467 ----------------------
4469 1. Changed the API for pcre_compile() to allow for the provision of a pointer
4470 to character tables built by pcre_maketables() in the current locale. If NULL
4471 is passed, the default tables are used.
4474 Version 2.00 24-Sep-98
4475 ----------------------
4477 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
4478 it any more.
4480 2. Allow quantification of (?>) groups, and make it work correctly.
4482 3. The first character computation wasn't working for (?>) groups.
4484 4. Correct the implementation of \Z (it is permitted to match on the \n at the
4485 end of the subject) and add 5.005's \z, which really does match only at the
4486 very end of the subject.
4488 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
4490 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
4491 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
4492 localized options. All options to pcre_study() were also removed.
4494 7. Add other new features from 5.005:
4496 $(?<= positive lookbehind
4497 $(?<! negative lookbehind
4498 (?imsx-imsx) added the unsetting capability
4499 such a setting is global if at outer level; local otherwise
4500 (?imsx-imsx:) non-capturing groups with option setting
4501 (?(cond)re|re) conditional pattern matching
4503 A backreference to itself in a repeated group matches the previous
4504 captured string.
4506 8. General tidying up of studying (both automatic and via "study")
4507 consequential on the addition of new assertions.
4509 9. As in 5.005, unlimited repeated groups that could match an empty substring
4510 are no longer faulted at compile time. Instead, the loop is forcibly broken at
4511 runtime if any iteration does actually match an empty substring.
4513 10. Include the RunTest script in the distribution.
4515 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
4516 discrepancies, some of which were old and were also with respect to 5.004. They
4517 have now been fixed.
4520 Version 1.09 28-Apr-98
4521 ----------------------
4523 1. A negated single character class followed by a quantifier with a minimum
4524 value of one (e.g. [^x]{1,6} ) was not compiled correctly. This could lead to
4525 program crashes, or just wrong answers. This did not apply to negated classes
4526 containing more than one character, or to minima other than one.
4529 Version 1.08 27-Mar-98
4530 ----------------------
4532 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
4534 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
4535 latter must appear before anything that relies on it in the pattern.
4538 Version 1.07 16-Feb-98
4539 ----------------------
4541 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
4542 repeat of a potentially empty string).
4545 Version 1.06 23-Jan-98
4546 ----------------------
4548 1. Added Markus Oberhumer's little patches for C++.
4550 2. Literal strings longer than 255 characters were broken.
4553 Version 1.05 23-Dec-97
4554 ----------------------
4556 1. Negated character classes containing more than one character were failing if
4557 PCRE_CASELESS was set at run time.
4560 Version 1.04 19-Dec-97
4561 ----------------------
4563 1. Corrected the man page, where some "const" qualifiers had been omitted.
4565 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
4566 input syntax.
4568 3. Fixed memory leak which occurred when a regex with back references was
4569 matched with an offsets vector that wasn't big enough. The temporary memory
4570 that is used in this case wasn't being freed if the match failed.
4572 4. Tidied pcretest to ensure it frees memory that it gets.
4574 5. Temporary memory was being obtained in the case where the passed offsets
4575 vector was exactly big enough.
4577 6. Corrected definition of offsetof() from change 5 below.
4579 7. I had screwed up change 6 below and broken the rules for the use of
4580 setjmp(). Now fixed.
4583 Version 1.03 18-Dec-97
4584 ----------------------
4586 1. A erroneous regex with a missing opening parenthesis was correctly
4587 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
4588 on some systems.
4590 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
4591 it was reported that one broken compiler failed on the former because "code" is
4592 also an independent variable.
4594 3. The erroneous regex a[]b caused an array overrun reference.
4596 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
4597 fail on data ending with that character. (It was going on too far, and checking
4598 the next character, typically a binary zero.) This was specific to the
4599 optimized code for single-character negative classes.
4601 5. Added a contributed patch from the TIN world which does the following:
4603 + Add an undef for memmove, in case the the system defines a macro for it.
4605 + Add a definition of offsetof(), in case there isn't one. (I don't know
4606 the reason behind this - offsetof() is part of the ANSI standard - but
4607 it does no harm).
4609 + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
4610 most of the places where whitespace preceded '#'. I have given up and
4611 allowed the remaining 2 cases to be at the margin.
4613 + Rename some variables in pcre to eliminate shadowing. This seems very
4614 pedantic, but does no harm, of course.
4616 6. Moved the call to setjmp() into its own function, to get rid of warnings
4617 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
4619 7. Constructs such as \d{8,} were compiling into the equivalent of
4620 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
4621 outcome, but in this particular case used more store than had been allocated,
4622 which caused the bug to be discovered because it threw up an internal error.
4624 8. The debugging code in both pcre and pcretest for outputting the compiled
4625 form of a regex was going wrong in the case of back references followed by
4626 curly-bracketed repeats.
4629 Version 1.02 12-Dec-97
4630 ----------------------
4632 1. Typos in pcre.3 and comments in the source fixed.
4634 2. Applied a contributed patch to get rid of places where it used to remove
4635 'const' from variables, and fixed some signed/unsigned and uninitialized
4636 variable warnings.
4638 3. Added the "runtest" target to Makefile.
4640 4. Set default compiler flag to -O2 rather than just -O.
4643 Version 1.01 19-Nov-97
4644 ----------------------
4646 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
4647 like /([ab]*)*/, that is, for classes with more than one character in them.
4649 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
4650 as /((?>a*))*/ (a PCRE_EXTRA facility).
4653 Version 1.00 18-Nov-97
4654 ----------------------
4656 1. Added compile-time macros to support systems such as SunOS4 which don't have
4657 memmove() or strerror() but have other things that can be used instead.
4659 2. Arranged that "make clean" removes the executables.
4662 Version 0.99 27-Oct-97
4663 ----------------------
4665 1. Fixed bug in code for optimizing classes with only one character. It was
4666 initializing a 32-byte map regardless, which could cause it to run off the end
4667 of the memory it had got.
4669 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
4672 Version 0.98 22-Oct-97
4673 ----------------------
4675 1. Fixed bug in code for handling temporary memory usage when there are more
4676 back references than supplied space in the ovector. This could cause segfaults.
4679 Version 0.97 21-Oct-97
4680 ----------------------
4682 1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
4684 2. Optimized negated single characters not to use a bit map.
4686 3. Brought error texts together as macro definitions; clarified some of them;
4687 fixed one that was wrong - it said "range out of order" when it meant "invalid
4688 escape sequence".
4690 4. Changed some char * arguments to const char *.
4692 5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
4694 6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
4695 pcretest.
4698 Version 0.96 16-Oct-97
4699 ----------------------
4701 1. Added a simple "pgrep" utility to the distribution.
4703 2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
4704 unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
4705 where "ddd" means "one or more decimal digits".
4707 3. Fixed serious bug. If a pattern had a back reference, but the call to
4708 pcre_exec() didn't supply a large enough ovector to record the related
4709 identifying subpattern, the match always failed. PCRE now remembers the number
4710 of the largest back reference, and gets some temporary memory in which to save
4711 the offsets during matching if necessary, in order to ensure that
4712 backreferences always work.
4714 4. Increased the compatibility with Perl in a number of ways:
4716 (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
4717 to request this handling. The option can be set at compile or exec time.
4719 (b) $ matches before a terminating newline by default; an option
4720 PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
4721 mode). The option can be set at compile or exec time.
4723 (c) The handling of \ followed by a digit other than 0 is now supposed to be
4724 the same as Perl's. If the decimal number it represents is less than 10
4725 or there aren't that many previous left capturing parentheses, an octal
4726 escape is read. Inside a character class, it's always an octal escape,
4727 even if it is a single digit.
4729 (d) An escaped but undefined alphabetic character is taken as a literal,
4730 unless PCRE_EXTRA is set. Currently this just reserves the remaining
4731 escapes.
4733 (e) {0} is now permitted. (The previous item is removed from the compiled
4734 pattern).
4736 5. Changed all the names of code files so that the basic parts are no longer
4737 than 10 characters, and abolished the teeny "globals.c" file.
4739 6. Changed the handling of character classes; they are now done with a 32-byte
4740 bit map always.
4742 7. Added the -d and /D options to pcretest to make it possible to look at the
4743 internals of compilation without having to recompile pcre.
4746 Version 0.95 23-Sep-97
4747 ----------------------
4749 1. Fixed bug in pre-pass concerning escaped "normal" characters such as \x5c or
4750 \x20 at the start of a run of normal characters. These were being treated as
4751 real characters, instead of the source characters being re-checked.
4754 Version 0.94 18-Sep-97
4755 ----------------------
4757 1. The functions are now thread-safe, with the caveat that the global variables
4758 containing pointers to malloc() and free() or alternative functions are the
4759 same for all threads.
4761 2. Get pcre_study() to generate a bitmap of initial characters for non-
4762 anchored patterns when this is possible, and use it if passed to pcre_exec().
4765 Version 0.93 15-Sep-97
4766 ----------------------
4768 1. /(b)|(:+)/ was computing an incorrect first character.
4770 2. Add pcre_study() to the API and the passing of pcre_extra to pcre_exec(),
4771 but not actually doing anything yet.
4773 3. Treat "-" characters in classes that cannot be part of ranges as literals,
4774 as Perl does (e.g. [-az] or [az-]).
4776 4. Set the anchored flag if a branch starts with .* or .*? because that tests
4777 all possible positions.
4779 5. Split up into different modules to avoid including unneeded functions in a
4780 compiled binary. However, compile and exec are still in one module. The "study"
4781 function is split off.
4783 6. The character tables are now in a separate module whose source is generated
4784 by an auxiliary program - but can then be edited by hand if required. There are
4785 now no calls to isalnum(), isspace(), isdigit(), isxdigit(), tolower() or
4786 toupper() in the code.
4788 7. Turn the malloc/free funtions variables into pcre_malloc and pcre_free and
4789 make them global. Abolish the function for setting them, as the caller can now
4790 set them directly.
4793 Version 0.92 11-Sep-97
4794 ----------------------
4796 1. A repeat with a fixed maximum and a minimum of 1 for an ordinary character
4797 (e.g. /a{1,3}/) was broken (I mis-optimized it).
4799 2. Caseless matching was not working in character classes if the characters in
4800 the pattern were in upper case.
4802 3. Make ranges like [W-c] work in the same way as Perl for caseless matching.
4804 4. Make PCRE_ANCHORED public and accept as a compile option.
4806 5. Add an options word to pcre_exec() and accept PCRE_ANCHORED and
4807 PCRE_CASELESS at run time. Add escapes \A and \I to pcretest to cause it to
4808 pass them.
4810 6. Give an error if bad option bits passed at compile or run time.
4812 7. Add PCRE_MULTILINE at compile and exec time, and (?m) as well. Add \M to
4813 pcretest to cause it to pass that flag.
4815 8. Add pcre_info(), to get the number of identifying subpatterns, the stored
4816 options, and the first character, if set.
4818 9. Recognize C+ or C{n,m} where n >= 1 as providing a fixed starting character.
4821 Version 0.91 10-Sep-97
4822 ----------------------
4824 1. PCRE was failing to diagnose unlimited repeats of subpatterns that could
4825 match the empty string as in /(a*)*/. It was looping and ultimately crashing.
4827 2. PCRE was looping on encountering an indefinitely repeated back reference to
4828 a subpattern that had matched an empty string, e.g. /(a|)\1*/. It now does what
4829 Perl does - treats the match as successful.
4831 ****


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

  ViewVC Help
Powered by ViewVC 1.1.5