ViewVC logotype

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log

Revision 911 - (show annotations)
Fri Feb 10 08:05:30 2012 UTC (8 years, 1 month ago) by zherczeg
File size: 226029 byte(s)
Fixing a wrong JIT test case
1 ChangeLog for PCRE
2 ------------------
4 Version 8.31
5 -----------------------------
6 1. Fixing a wrong JIT test case and some compiler warnings.
9 Version 8.30 04-February-2012
10 -----------------------------
12 1. Renamed "isnumber" as "is_a_number" because in some Mac environments this
13 name is defined in ctype.h.
15 2. Fixed a bug in fixed-length calculation for lookbehinds that would show up
16 only in quite long subpatterns.
18 3. Removed the function pcre_info(), which has been obsolete and deprecated
19 since it was replaced by pcre_fullinfo() in February 2000.
21 4. For a non-anchored pattern, if (*SKIP) was given with a name that did not
22 match a (*MARK), and the match failed at the start of the subject, a
23 reference to memory before the start of the subject could occur. This bug
24 was introduced by fix 17 of release 8.21.
26 5. A reference to an unset group with zero minimum repetition was giving
27 totally wrong answers (in non-JavaScript-compatibility mode). For example,
28 /(another)?(\1?)test/ matched against "hello world test". This bug was
29 introduced in release 8.13.
31 6. Add support for 16-bit character strings (a large amount of work involving
32 many changes and refactorings).
34 7. RunGrepTest failed on msys because \r\n was replaced by whitespace when the
35 command "pattern=`printf 'xxx\r\njkl'`" was run. The pattern is now taken
36 from a file.
38 8. Ovector size of 2 is also supported by JIT based pcre_exec (the ovector size
39 rounding is not applied in this particular case).
41 9. The invalid Unicode surrogate codepoints U+D800 to U+DFFF are now rejected
42 if they appear, or are escaped, in patterns.
44 10. Get rid of a number of -Wunused-but-set-variable warnings.
46 11. The pattern /(?=(*:x))(q|)/ matches an empty string, and returns the mark
47 "x". The similar pattern /(?=(*:x))((*:y)q|)/ did not return a mark at all.
48 Oddly, Perl behaves the same way. PCRE has been fixed so that this pattern
49 also returns the mark "x". This bug applied to capturing parentheses,
50 non-capturing parentheses, and atomic parentheses. It also applied to some
51 assertions.
53 12. Stephen Kelly's patch to CMakeLists.txt allows it to parse the version
54 information out of configure.ac instead of relying on pcre.h.generic, which
55 is not stored in the repository.
57 13. Applied Dmitry V. Levin's patch for a more portable method for linking with
58 -lreadline.
60 14. ZH added PCRE_CONFIG_JITTARGET; added its output to pcretest -C.
62 15. Applied Graycode's patch to put the top-level frame on the stack rather
63 than the heap when not using the stack for recursion. This gives a
64 performance improvement in many cases when recursion is not deep.
66 16. Experimental code added to "pcretest -C" to output the stack frame size.
69 Version 8.21 12-Dec-2011
70 ------------------------
72 1. Updating the JIT compiler.
74 2. JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
75 are added as well.
77 3. Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
78 PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
79 calling _pcre_jit_exec. Some extra comments are added.
81 4. (*MARK) settings inside atomic groups that do not contain any capturing
82 parentheses, for example, (?>a(*:m)), were not being passed out. This bug
83 was introduced by change 18 for 8.20.
85 5. Supporting of \x, \U and \u in JavaScript compatibility mode based on the
86 ECMA-262 standard.
88 6. Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
89 erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
90 This bug was probably introduced by change 9 of 8.13.
92 7. While fixing 6 above, I noticed that a number of other items were being
93 incorrectly rejected as "not fixed length". This arose partly because newer
94 opcodes had not been added to the fixed-length checking code. I have (a)
95 corrected the bug and added tests for these items, and (b) arranged for an
96 error to occur if an unknown opcode is encountered while checking for fixed
97 length instead of just assuming "not fixed length". The items that were
98 rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
99 (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
100 repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
102 8. A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
103 being incorrectly compiled and would have given unpredicatble results.
105 9. A possessively repeated subpattern with minimum repeat count greater than
106 one behaved incorrectly. For example, (A){2,}+ behaved as if it was
107 (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
108 the first (A) could occur when it should not.
110 10. Add a cast and remove a redundant test from the code.
112 11. JIT should use pcre_malloc/pcre_free for allocation.
114 12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
115 best practice nowadays, and helps with cross-compiling. (If the exec_prefix
116 is anything other than /usr, -L is still shown).
118 13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
120 14. Perl does not support \N without a following name in a [] class; PCRE now
121 also gives an error.
123 15. If a forward reference was repeated with an upper limit of around 2000,
124 it caused the error "internal error: overran compiling workspace". The
125 maximum number of forward references (including repeats) was limited by the
126 internal workspace, and dependent on the LINK_SIZE. The code has been
127 rewritten so that the workspace expands (via pcre_malloc) if necessary, and
128 the default depends on LINK_SIZE. There is a new upper limit (for safety)
129 of around 200,000 forward references. While doing this, I also speeded up
130 the filling in of repeated forward references.
132 16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
133 incorrectly expecting the subject to contain another "a" after the start.
135 17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
136 in the match, the SKIP should be ignored. This was not happening; instead
137 the SKIP was being treated as NOMATCH. For patterns such as
138 /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
139 tested.
141 18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
142 now much more compatible with Perl, in particular in cases where the result
143 is a non-match for a non-anchored pattern. For example, if
144 /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
145 "m", where previously it did not return a name. A side effect of this
146 change is that for partial matches, the last encountered mark name is
147 returned, as for non matches. A number of tests that were previously not
148 Perl-compatible have been moved into the Perl-compatible test files. The
149 refactoring has had the pleasing side effect of removing one argument from
150 the match() function, thus reducing its stack requirements.
152 19. If the /S+ option was used in pcretest to study a pattern using JIT,
153 subsequent uses of /S (without +) incorrectly behaved like /S+.
155 21. Retrieve executable code size support for the JIT compiler and fixing
156 some warnings.
158 22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
159 not work when the shorter character appeared right at the end of the
160 subject string.
162 23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
163 systems.
165 24. Added PCRE_INFO_JITSIZE to pass on the value from (21) above, and also
166 output it when the /M option is used in pcretest.
168 25. The CheckMan script was not being included in the distribution. Also, added
169 an explicit "perl" to run Perl scripts from the PrepareRelease script
170 because this is reportedly needed in Windows.
172 26. If study data was being save in a file and studying had not found a set of
173 "starts with" bytes for the pattern, the data written to the file (though
174 never used) was taken from uninitialized memory and so caused valgrind to
175 complain.
177 27. Updated RunTest.bat as provided by Sheri Pierce.
179 28. Fixed a possible uninitialized memory bug in pcre_jit_compile.c.
181 29. Computation of memory usage for the table of capturing group names was
182 giving an unnecessarily large value.
185 Version 8.20 21-Oct-2011
186 ------------------------
188 1. Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
189 a POSIX class. After further experiments with Perl, which convinced me that
190 Perl has bugs and confusions, a closing square bracket is no longer allowed
191 in a POSIX name. This bug also affected patterns with classes that started
192 with full stops.
194 2. If a pattern such as /(a)b|ac/ is matched against "ac", there is no
195 captured substring, but while checking the failing first alternative,
196 substring 1 is temporarily captured. If the output vector supplied to
197 pcre_exec() was not big enough for this capture, the yield of the function
198 was still zero ("insufficient space for captured substrings"). This cannot
199 be totally fixed without adding another stack variable, which seems a lot
200 of expense for a edge case. However, I have improved the situation in cases
201 such as /(a)(b)x|abc/ matched against "abc", where the return code
202 indicates that fewer than the maximum number of slots in the ovector have
203 been set.
205 3. Related to (2) above: when there are more back references in a pattern than
206 slots in the output vector, pcre_exec() uses temporary memory during
207 matching, and copies in the captures as far as possible afterwards. It was
208 using the entire output vector, but this conflicts with the specification
209 that only 2/3 is used for passing back captured substrings. Now it uses
210 only the first 2/3, for compatibility. This is, of course, another edge
211 case.
213 4. Zoltan Herczeg's just-in-time compiler support has been integrated into the
214 main code base, and can be used by building with --enable-jit. When this is
215 done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
216 runtime --no-jit option is given.
218 5. When the number of matches in a pcre_dfa_exec() run exactly filled the
219 ovector, the return from the function was zero, implying that there were
220 other matches that did not fit. The correct "exactly full" value is now
221 returned.
223 6. If a subpattern that was called recursively or as a subroutine contained
224 (*PRUNE) or any other control that caused it to give a non-standard return,
225 invalid errors such as "Error -26 (nested recursion at the same subject
226 position)" or even infinite loops could occur.
228 7. If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
229 computing the minimum length on reaching *ACCEPT, and so ended up with the
230 wrong value of 1 rather than 0. Further investigation indicates that
231 computing a minimum subject length in the presence of *ACCEPT is difficult
232 (think back references, subroutine calls), and so I have changed the code
233 so that no minimum is registered for a pattern that contains *ACCEPT.
235 8. If (*THEN) was present in the first (true) branch of a conditional group,
236 it was not handled as intended. [But see 16 below.]
238 9. Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
239 Sheri Pierce.
241 10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
242 the first byte in a match must be "a".
244 11. Change 17 for 8.13 increased the recursion depth for patterns like
245 /a(?:.)*?a/ drastically. I've improved things by remembering whether a
246 pattern contains any instances of (*THEN). If it does not, the old
247 optimizations are restored. It would be nice to do this on a per-group
248 basis, but at the moment that is not feasible.
250 12. In some environments, the output of pcretest -C is CRLF terminated. This
251 broke RunTest's code that checks for the link size. A single white space
252 character after the value is now allowed for.
254 13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
255 For "fr", it uses the Windows-specific input and output files.
257 14. If (*THEN) appeared in a group that was called recursively or as a
258 subroutine, it did not work as intended. [But see next item.]
260 15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
261 pattern fragments (but not containing any | characters). If A and B are
262 matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
263 was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
264 D. In other words, Perl considers parentheses that do not contain any |
265 characters to be part of a surrounding alternative, whereas PCRE was
266 treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
267 differently. PCRE now behaves in the same way as Perl, except in the case
268 of subroutine/recursion calls such as (?1) which have in any case always
269 been different (but PCRE had them first :-).
271 16. Related to 15 above: Perl does not treat the | in a conditional group as
272 creating alternatives. Such a group is treated in the same way as an
273 ordinary group without any | characters when processing (*THEN). PCRE has
274 been changed to match Perl's behaviour.
276 17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
277 RunGrepTest script failed.
279 18. Change 22 for version 13 caused atomic groups to use more stack. This is
280 inevitable for groups that contain captures, but it can lead to a lot of
281 stack use in large patterns. The old behaviour has been restored for atomic
282 groups that do not contain any capturing parentheses.
284 19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
285 suppress the check for a minimum subject length at run time. (If it was
286 given to pcre_exec() or pcre_dfa_exec() it did work.)
288 20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
289 fail to work when decoding hex characters in data strings in EBCDIC
290 environments.
292 21. It appears that in at least one Mac OS environment, the isxdigit() function
293 is implemented as a macro that evaluates to its argument more than once,
294 contravening the C 90 Standard (I haven't checked a later standard). There
295 was an instance in pcretest which caused it to go wrong when processing
296 \x{...} escapes in subject strings. The has been rewritten to avoid using
297 things like p++ in the argument of isxdigit().
300 Version 8.13 16-Aug-2011
301 ------------------------
303 1. The Unicode data tables have been updated to Unicode 6.0.0.
305 2. Two minor typos in pcre_internal.h have been fixed.
307 3. Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
308 pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
309 in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
311 4. There were a number of related bugs in the code for matching backrefences
312 caselessly in UTF-8 mode when codes for the characters concerned were
313 different numbers of bytes. For example, U+023A and U+2C65 are an upper
314 and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
315 (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
316 code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
317 2-byte code at the end of the subject (it thought there wasn't enough data
318 left).
320 5. Comprehensive information about what went wrong is now returned by
321 pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
322 as the output vector has at least 2 elements. The offset of the start of
323 the failing character and a reason code are placed in the vector.
325 6. When the UTF-8 string check fails for pcre_compile(), the offset that is
326 now returned is for the first byte of the failing character, instead of the
327 last byte inspected. This is an incompatible change, but I hope it is small
328 enough not to be a problem. It makes the returned offset consistent with
329 pcre_exec() and pcre_dfa_exec().
331 7. pcretest now gives a text phrase as well as the error number when
332 pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
333 failure, the offset and reason code are output.
335 8. When \R was used with a maximizing quantifier it failed to skip backwards
336 over a \r\n pair if the subsequent match failed. Instead, it just skipped
337 back over a single character (\n). This seems wrong (because it treated the
338 two characters as a single entity when going forwards), conflicts with the
339 documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
340 behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
341 has been changed.
343 9. Some internal refactoring has changed the processing so that the handling
344 of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
345 time (the PCRE_DOTALL option was changed this way some time ago: version
346 7.7 change 16). This has made it possible to abolish the OP_OPT op code,
347 which was always a bit of a fudge. It also means that there is one less
348 argument for the match() function, which reduces its stack requirements
349 slightly. This change also fixes an incompatibility with Perl: the pattern
350 (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
352 10. More internal refactoring has drastically reduced the number of recursive
353 calls to match() for possessively repeated groups such as (abc)++ when
354 using pcre_exec().
356 11. While implementing 10, a number of bugs in the handling of groups were
357 discovered and fixed:
359 (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
360 (a|)*(?1) gave a compile-time internal error.
361 ((a|)+)+ did not notice that the outer group could match an empty string.
362 (^a|^)+ was not marked as anchored.
363 (.*a|.*)+ was not marked as matching at start or after a newline.
365 12. Yet more internal refactoring has removed another argument from the match()
366 function. Special calls to this function are now indicated by setting a
367 value in a variable in the "match data" data block.
369 13. Be more explicit in pcre_study() instead of relying on "default" for
370 opcodes that mean there is no starting character; this means that when new
371 ones are added and accidentally left out of pcre_study(), testing should
372 pick them up.
374 14. The -s option of pcretest has been documented for ages as being an old
375 synonym of -m (show memory usage). I have changed it to mean "force study
376 for every regex", that is, assume /S for every regex. This is similar to -i
377 and -d etc. It's slightly incompatible, but I'm hoping nobody is still
378 using it. It makes it easier to run collections of tests with and without
379 study enabled, and thereby test pcre_study() more easily. All the standard
380 tests are now run with and without -s (but some patterns can be marked as
381 "never study" - see 20 below).
383 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
384 restoration of the capturing data to the outer values was not happening
385 correctly.
387 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
388 empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
389 pattern had matched an empty string, and so incorrectly returned a no
390 match.
392 17. There was optimizing code for the last branch of non-capturing parentheses,
393 and also for the obeyed branch of a conditional subexpression, which used
394 tail recursion to cut down on stack usage. Unfortunately, now that there is
395 the possibility of (*THEN) occurring in these branches, tail recursion is
396 no longer possible because the return has to be checked for (*THEN). These
397 two optimizations have therefore been removed. [But see 8.20/11 above.]
399 18. If a pattern containing \R was studied, it was assumed that \R always
400 matched two bytes, thus causing the minimum subject length to be
401 incorrectly computed because \R can also match just one byte.
403 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
404 was incorrectly computed.
406 20. If /S is present twice on a test pattern in pcretest input, it now
407 *disables* studying, thereby overriding the use of -s on the command line
408 (see 14 above). This is necessary for one or two tests to keep the output
409 identical in both cases.
411 21. When (*ACCEPT) was used in an assertion that matched an empty string and
412 PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
414 22. When an atomic group that contained a capturing parenthesis was
415 successfully matched, but the branch in which it appeared failed, the
416 capturing was not being forgotten if a higher numbered group was later
417 captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
418 group 1 to "a", when in fact it should be unset. This applied to multi-
419 branched capturing and non-capturing groups, repeated or not, and also to
420 positive assertions (capturing in negative assertions does not happen
421 in PCRE) and also to nested atomic groups.
423 23. Add the ++ qualifier feature to pcretest, to show the remainder of the
424 subject after a captured substring, to make it easier to tell which of a
425 number of identical substrings has been captured.
427 24. The way atomic groups are processed by pcre_exec() has been changed so that
428 if they are repeated, backtracking one repetition now resets captured
429 values correctly. For example, if ((?>(a+)b)+aabab) is matched against
430 "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
431 "aaa". Previously, it would have been "a". As part of this code
432 refactoring, the way recursive calls are handled has also been changed.
434 25. If an assertion condition captured any substrings, they were not passed
435 back unless some other capturing happened later. For example, if
436 (?(?=(a))a) was matched against "a", no capturing was returned.
438 26. When studying a pattern that contained subroutine calls or assertions,
439 the code for finding the minimum length of a possible match was handling
440 direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
441 group 1 called group 2 while simultaneously a separate group 2 called group
442 1). A stack overflow occurred in this case. I have fixed this by limiting
443 the recursion depth to 10.
445 27. Updated RunTest.bat in the distribution to the version supplied by Tom
446 Fortmann. This supports explicit test numbers on the command line, and has
447 argument validation and error reporting.
449 28. An instance of \X with an unlimited repeat could fail if at any point the
450 first character it looked at was a mark character.
452 29. Some minor code refactoring concerning Unicode properties and scripts
453 should reduce the stack requirement of match() slightly.
455 30. Added the '=' option to pcretest to check the setting of unused capturing
456 slots at the end of the pattern, which are documented as being -1, but are
457 not included in the return count.
459 31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
460 compiled something random. Now it gives a compile-time error (as does
461 Perl).
463 32. A *MARK encountered during the processing of a positive assertion is now
464 recorded and passed back (compatible with Perl).
466 33. If --only-matching or --colour was set on a pcregrep call whose pattern
467 had alternative anchored branches, the search for a second match in a line
468 was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
469 matched the line "0102" twice. The same bug affected patterns that started
470 with a backwards assertion. For example /\b01|\b02/ also matched "0102"
471 twice.
473 34. Previously, PCRE did not allow quantification of assertions. However, Perl
474 does, and because of capturing effects, quantifying parenthesized
475 assertions may at times be useful. Quantifiers are now allowed for
476 parenthesized assertions.
478 35. A minor code tidy in pcre_compile() when checking options for \R usage.
480 36. \g was being checked for fancy things in a character class, when it should
481 just be a literal "g".
483 37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
484 appearance of a nested POSIX class supersedes an apparent external class.
485 For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
486 unescaped square brackets may also appear as part of class names. For
487 example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
488 more like Perl. (But see 8.20/1 above.)
490 38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
491 was because it thought it was \N{name}, which is not supported).
493 39. Add minix to OS list not supporting the -S option in pcretest.
495 40. PCRE tries to detect cases of infinite recursion at compile time, but it
496 cannot analyze patterns in sufficient detail to catch mutual recursions
497 such as ((?1))((?2)). There is now a runtime test that gives an error if a
498 subgroup is called recursively as a subpattern for a second time at the
499 same position in the subject string. In previous releases this might have
500 been caught by the recursion limit, or it might have run out of stack.
502 41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
503 happen only once. PCRE was, however incorrectly giving a compile time error
504 "recursive call could loop indefinitely" because it cannot analyze the
505 pattern in sufficient detail. The compile time test no longer happens when
506 PCRE is compiling a conditional subpattern, but actual runaway loops are
507 now caught at runtime (see 40 above).
509 42. It seems that Perl allows any characters other than a closing parenthesis
510 to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
511 has been changed to be the same.
513 43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
514 as not to get warnings when autogen.sh is called. Also changed
515 AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
517 44. To help people who use pcregrep to scan files containing exceedingly long
518 lines, the following changes have been made:
520 (a) The default value of the buffer size parameter has been increased from
521 8K to 20K. (The actual buffer used is three times this size.)
523 (b) The default can be changed by ./configure --with-pcregrep-bufsize when
524 PCRE is built.
526 (c) A --buffer-size=n option has been added to pcregrep, to allow the size
527 to be set at run time.
529 (d) Numerical values in pcregrep options can be followed by K or M, for
530 example --buffer-size=50K.
532 (e) If a line being scanned overflows pcregrep's buffer, an error is now
533 given and the return code is set to 2.
535 45. Add a pointer to the latest mark to the callout data block.
537 46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
538 partial match of an empty string instead of no match. This was specific to
539 the use of ".".
541 47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
542 complete match instead of a partial match. This bug was dependent on both
543 the PCRE_UTF8 and PCRE_DOTALL options being set.
545 48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
546 starting byte set, because \b was not being ignored.
549 Version 8.12 15-Jan-2011
550 ------------------------
552 1. Fixed some typos in the markup of the man pages, and wrote a script that
553 checks for such things as part of the documentation building process.
555 2. On a big-endian 64-bit system, pcregrep did not correctly process the
556 --match-limit and --recursion-limit options (added for 8.11). In
557 particular, this made one of the standard tests fail. (The integer value
558 went into the wrong half of a long int.)
560 3. If the --colour option was given to pcregrep with -v (invert match), it
561 did strange things, either producing crazy output, or crashing. It should,
562 of course, ignore a request for colour when reporting lines that do not
563 match.
565 4. Another pcregrep bug caused similar problems if --colour was specified with
566 -M (multiline) and the pattern match finished with a line ending.
568 5. In pcregrep, when a pattern that ended with a literal newline sequence was
569 matched in multiline mode, the following line was shown as part of the
570 match. This seems wrong, so I have changed it.
572 6. Another pcregrep bug in multiline mode, when --colour was specified, caused
573 the check for further matches in the same line (so they could be coloured)
574 to overrun the end of the current line. If another match was found, it was
575 incorrectly shown (and then shown again when found in the next line).
577 7. If pcregrep was compiled under Windows, there was a reference to the
578 function pcregrep_exit() before it was defined. I am assuming this was
579 the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
580 reported by a user. I've moved the definition above the reference.
583 Version 8.11 10-Dec-2010
584 ------------------------
586 1. (*THEN) was not working properly if there were untried alternatives prior
587 to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
588 backtracked to try for "b" instead of moving to the next alternative branch
589 at the same level (in this case, to look for "c"). The Perl documentation
590 is clear that when (*THEN) is backtracked onto, it goes to the "next
591 alternative in the innermost enclosing group".
593 2. (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
594 such as (A(*COMMIT)B(*THEN)C|D) any failure after matching A should
595 result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
596 (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
597 (*THEN).
599 3. If \s appeared in a character class, it removed the VT character from
600 the class, even if it had been included by some previous item, for example
601 in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
602 of \s, but is part of the POSIX "space" class.)
604 4. A partial match never returns an empty string (because you can always
605 match an empty string at the end of the subject); however the checking for
606 an empty string was starting at the "start of match" point. This has been
607 changed to the "earliest inspected character" point, because the returned
608 data for a partial match starts at this character. This means that, for
609 example, /(?<=abc)def/ gives a partial match for the subject "abc"
610 (previously it gave "no match").
612 5. Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
613 of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
614 previously a full match would be given. However, setting PCRE_PARTIAL_HARD
615 has an implication that the given string is incomplete (because a partial
616 match is preferred over a full match). For this reason, these items now
617 give a partial match in this situation. [Aside: previously, the one case
618 /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
619 match rather than a full match, which was wrong by the old rules, but is
620 now correct.]
622 6. There was a bug in the handling of #-introduced comments, recognized when
623 PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
624 If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
625 UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
626 scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
627 but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
628 places in pcre_compile().
630 7. Related to (6) above, when pcre_compile() was skipping #-introduced
631 comments when looking ahead for named forward references to subpatterns,
632 the only newline sequence it recognized was NL. It now handles newlines
633 according to the set newline convention.
635 8. SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
636 former, but used strtoul(), whereas pcretest avoided strtoul() but did not
637 cater for a lack of strerror(). These oversights have been fixed.
639 9. Added --match-limit and --recursion-limit to pcregrep.
641 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
643 11. When the -o option was used, pcregrep was setting a return code of 1, even
644 when matches were found, and --line-buffered was not being honoured.
646 12. Added an optional parentheses number to the -o and --only-matching options
647 of pcregrep.
649 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
650 can match an empty string. The code to do it in pcretest and pcredemo
651 needed fixing:
653 (a) When the newline convention was "crlf", pcretest got it wrong, skipping
654 only one byte after an empty string match just before CRLF (this case
655 just got forgotten; "any" and "anycrlf" were OK).
657 (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
658 mode when an empty string match preceded an ASCII character followed by
659 a non-ASCII character. (The code for advancing by one character rather
660 than one byte was nonsense.)
662 (c) The pcredemo.c sample program did not have any code at all to handle
663 the cases when CRLF is a valid newline sequence.
665 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
666 as a starting offset was within the subject string. There is now a new
667 error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
668 negative or greater than the length of the string. In order to test this,
669 pcretest is extended to allow the setting of negative starting offsets.
671 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
672 starting offset points to the beginning of a UTF-8 character was
673 unnecessarily clumsy. I tidied it up.
675 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
676 bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
678 17. Nobody had reported that the --include_dir option, which was added in
679 release 7.7 should have been called --include-dir (hyphen, not underscore)
680 for compatibility with GNU grep. I have changed it to --include-dir, but
681 left --include_dir as an undocumented synonym, and the same for
682 --exclude-dir, though that is not available in GNU grep, at least as of
683 release 2.5.4.
685 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
686 characters from a string of bytes) have been redefined so as not to use
687 loops, in order to improve performance in some environments. At the same
688 time, I abstracted some of the common code into auxiliary macros to save
689 repetition (this should not affect the compiled code).
691 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
692 compile-time error is now given if \c is not followed by an ASCII
693 character, that is, a byte less than 128. (In EBCDIC mode, the code is
694 different, and any byte value is allowed.)
696 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
697 START_OPTIMIZE option, which is now allowed at compile time - but just
698 passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
699 to pcregrep and other applications that have no direct access to PCRE
700 options. The new /Y option in pcretest sets this option when calling
701 pcre_compile().
703 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
704 back references. Groups containing recursive back references were forced to
705 be atomic by that change, but in the case of named groups, the amount of
706 memory required was incorrectly computed, leading to "Failed: internal
707 error: code overflow". This has been fixed.
709 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
710 pcretest.c, to avoid build problems in some Borland environments.
713 Version 8.10 25-Jun-2010
714 ------------------------
716 1. Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
717 THEN.
719 2. (*ACCEPT) was not working when inside an atomic group.
721 3. Inside a character class, \B is treated as a literal by default, but
722 faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
723 causes the error). The code is unchanged, but I tidied the documentation.
725 4. Inside a character class, PCRE always treated \R and \X as literals,
726 whereas Perl faults them if its -w option is set. I have changed PCRE so
727 that it faults them when PCRE_EXTRA is set.
729 5. Added support for \N, which always matches any character other than
730 newline. (It is the same as "." when PCRE_DOTALL is not set.)
732 6. When compiling pcregrep with newer versions of gcc which may have
733 FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
734 declared with attribute warn_unused_result" were given. Just casting the
735 result to (void) does not stop the warnings; a more elaborate fudge is
736 needed. I've used a macro to implement this.
738 7. Minor change to pcretest.c to avoid a compiler warning.
740 8. Added four artifical Unicode properties to help with an option to make
741 \s etc use properties (see next item). The new properties are: Xan
742 (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
744 9. Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
745 use Unicode properties. (*UCP) at the start of a pattern can be used to set
746 this option. Modified pcretest to add /W to test this facility. Added
747 REG_UCP to make it available via the POSIX interface.
749 10. Added --line-buffered to pcregrep.
751 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
752 studied, and the match started with a letter with a code point greater than
753 127 whose first byte was different to the first byte of the other case of
754 the letter, the other case of this starting letter was not recognized
755 (#976).
757 12. If a pattern that was studied started with a repeated Unicode property
758 test, for example, \p{Nd}+, there was the theoretical possibility of
759 setting up an incorrect bitmap of starting bytes, but fortunately it could
760 not have actually happened in practice until change 8 above was made (it
761 added property types that matched character-matching opcodes).
763 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
764 possible starting bytes for non-anchored patterns.
766 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
767 \R, and also a number of cases that involve Unicode properties, both
768 explicit and implicit when PCRE_UCP is set.
770 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
771 input, it could crash or give wrong results if characters with values
772 greater than 0xc0 were present in the subject string. (Detail: it assumed
773 UTF-8 input when processing these items.)
775 16. Added a lot of (int) casts to avoid compiler warnings in systems where
776 size_t is 64-bit (#991).
778 17. Added a check for running out of memory when PCRE is compiled with
779 --disable-stack-for-recursion (#990).
781 18. If the last data line in a file for pcretest does not have a newline on
782 the end, a newline was missing in the output.
784 19. The default pcre_chartables.c file recognizes only ASCII characters (values
785 less than 128) in its various bitmaps. However, there is a facility for
786 generating tables according to the current locale when PCRE is compiled. It
787 turns out that in some environments, 0x85 and 0xa0, which are Unicode space
788 characters, are recognized by isspace() and therefore were getting set in
789 these tables, and indeed these tables seem to approximate to ISO 8859. This
790 caused a problem in UTF-8 mode when pcre_study() was used to create a list
791 of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
792 which of course cannot start UTF-8 characters. I have changed the code so
793 that only real ASCII characters (less than 128) and the correct starting
794 bytes for UTF-8 encodings are set for characters greater than 127 when in
795 UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
796 altogether.)
798 20. Added the /T option to pcretest so as to be able to run tests with non-
799 standard character tables, thus making it possible to include the tests
800 used for 19 above in the standard set of tests.
802 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
803 reference to a subpattern the other side of a comment that contains an
804 opening parenthesis caused either an internal compiling error, or a
805 reference to the wrong subpattern.
808 Version 8.02 19-Mar-2010
809 ------------------------
811 1. The Unicode data tables have been updated to Unicode 5.2.0.
813 2. Added the option --libs-cpp to pcre-config, but only when C++ support is
814 configured.
816 3. Updated the licensing terms in the pcregexp.pas file, as agreed with the
817 original author of that file, following a query about its status.
819 4. On systems that do not have stdint.h (e.g. Solaris), check for and include
820 inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
822 5. A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
823 quantifier applied to a forward-referencing subroutine call, could compile
824 incorrect code or give the error "internal error: previously-checked
825 referenced subpattern not found".
827 6. Both MS Visual Studio and Symbian OS have problems with initializing
828 variables to point to external functions. For these systems, therefore,
829 pcre_malloc etc. are now initialized to local functions that call the
830 relevant global functions.
832 7. There were two entries missing in the vectors called coptable and poptable
833 in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
834 I've fixed the data, and added a kludgy way of testing at compile time that
835 the lengths are correct (equal to the number of opcodes).
837 8. Following on from 7, I added a similar kludge to check the length of the
838 eint vector in pcreposix.c.
840 9. Error texts for pcre_compile() are held as one long string to avoid too
841 much relocation at load time. To find a text, the string is searched,
842 counting zeros. There was no check for running off the end of the string,
843 which could happen if a new error number was added without updating the
844 string.
846 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
848 11. \K was not working if it appeared in an atomic group or in a group that
849 was called as a "subroutine", or in an assertion. Perl 5.11 documents that
850 \K is "not well defined" if used in an assertion. PCRE now accepts it if
851 the assertion is positive, but not if it is negative.
853 12. Change 11 fortuitously reduced the size of the stack frame used in the
854 "match()" function of pcre_exec.c by one pointer. Forthcoming
855 implementation of support for (*MARK) will need an extra pointer on the
856 stack; I have reserved it now, so that the stack frame size does not
857 decrease.
859 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
860 item in branch that calls a recursion is a subroutine call - as in the
861 second branch in the above example - was incorrectly given the compile-
862 time error "recursive call could loop indefinitely" because pcre_compile()
863 was not correctly checking the subroutine for matching a non-empty string.
865 14. The checks for overrunning compiling workspace could trigger after an
866 overrun had occurred. This is a "should never occur" error, but it can be
867 triggered by pathological patterns such as hundreds of nested parentheses.
868 The checks now trigger 100 bytes before the end of the workspace.
870 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
873 Version 8.01 19-Jan-2010
874 ------------------------
876 1. If a pattern contained a conditional subpattern with only one branch (in
877 particular, this includes all (*DEFINE) patterns), a call to pcre_study()
878 computed the wrong minimum data length (which is of course zero for such
879 subpatterns). This could cause incorrect "no match" results.
881 2. For patterns such as (?i)a(?-i)b|c where an option setting at the start of
882 the pattern is reset in the first branch, pcre_compile() failed with
883 "internal error: code overflow at offset...". This happened only when
884 the reset was to the original external option setting. (An optimization
885 abstracts leading options settings into an external setting, which was the
886 cause of this.)
888 3. A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
889 of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
890 assertion pattern did not match (meaning that the assertion was true), it
891 was incorrectly treated as false if the SKIP had been reached during the
892 matching. This also applied to assertions used as conditions.
894 4. If an item that is not supported by pcre_dfa_exec() was encountered in an
895 assertion subpattern, including such a pattern used as a condition,
896 unpredictable results occurred, instead of the error return
899 5. The C++ GlobalReplace function was not working like Perl for the special
900 situation when an empty string is matched. It now does the fancy magic
901 stuff that is necessary.
903 6. In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
904 removed. (These were left over from very, very early versions of PCRE.)
906 7. Some cosmetic changes to the code to make life easier when compiling it
907 as part of something else:
909 (a) Change DEBUG to PCRE_DEBUG.
911 (b) In pcre_compile(), rename the member of the "branch_chain" structure
912 called "current" as "current_branch", to prevent a collision with the
913 Linux macro when compiled as a kernel module.
915 (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
916 prevent a collision with the Linux macro when compiled as a kernel
917 module.
919 8. In pcre_compile() there are some checks for integer overflows that used to
920 cast potentially large values to (double). This has been changed to that
921 when building, a check for int64_t is made, and if it is found, it is used
922 instead, thus avoiding the use of floating point arithmetic. (There is no
923 other use of FP in PCRE.) If int64_t is not found, the fallback is to
924 double.
926 9. Added two casts to avoid signed/unsigned warnings from VS Studio Express
927 2005 (difference between two addresses compared to an unsigned value).
929 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
930 custom one, because of the following reported problem in Windows:
932 - libbz2 uses the Pascal calling convention (WINAPI) for the functions
933 under Win32.
934 - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
935 therefore missing the function definition.
936 - The compiler thus generates a "C" signature for the test function.
937 - The linker fails to find the "C" function.
938 - PCRE fails to configure if asked to do so against libbz2.
940 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
941 messages were output:
943 Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
944 rerunning libtoolize, to keep the correct libtool macros in-tree.
945 Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
947 I have done both of these things.
949 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
950 most of the time, it *can* run out if it is given a pattern that contains a
951 runaway infinite recursion. I updated the discussion in the pcrestack man
952 page.
954 13. Now that we have gone to the x.xx style of version numbers, the minor
955 version may start with zero. Using 08 or 09 is a bad idea because users
956 might check the value of PCRE_MINOR in their code, and 08 or 09 may be
957 interpreted as invalid octal numbers. I've updated the previous comment in
958 configure.ac, and also added a check that gives an error if 08 or 09 are
959 used.
961 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
962 causing partial matching to fail when the end of the subject matched \W
963 in a UTF-8 pattern where \W was quantified with a minimum of 3.
965 15. There were some discrepancies between the declarations in pcre_internal.h
966 of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
967 their definitions. The declarations used "const uschar *" and the
968 definitions used USPTR. Even though USPTR is normally defined as "const
969 unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
970 reported that: "This difference in casting confuses some C++ compilers, for
971 example, SunCC recognizes above declarations as different functions and
972 generates broken code for hbpcre." I have changed the declarations to use
973 USPTR.
975 16. GNU libtool is named differently on some systems. The autogen.sh script now
976 tries several variants such as glibtoolize (MacOSX) and libtoolize1x
977 (FreeBSD).
979 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
980 (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
981 comment: "Figure out how to create a longlong from a string: strtoll and
982 equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
983 instance, but it only takes 2 args instead of 3!"
985 18. A subtle bug concerned with back references has been fixed by a change of
986 specification, with a corresponding code fix. A pattern such as
987 ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
988 refers, was giving matches when it shouldn't. For example, xa=xaaa would
989 match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
990 same bug. Such groups have to be quantified to be useful, or contained
991 inside another quantified group. (If there's no repetition, the reference
992 can never match.) The problem arises because, having left the group and
993 moved on to the rest of the pattern, a later failure that backtracks into
994 the group uses the captured value from the final iteration of the group
995 rather than the correct earlier one. I have fixed this in PCRE by forcing
996 any group that contains a reference to itself to be an atomic group; that
997 is, there cannot be any backtracking into it once it has completed. This is
998 similar to recursive and subroutine calls.
1001 Version 8.00 19-Oct-09
1002 ----------------------
1004 1. The table for translating pcre_compile() error codes into POSIX error codes
1005 was out-of-date, and there was no check on the pcre_compile() error code
1006 being within the table. This could lead to an OK return being given in
1007 error.
1009 2. Changed the call to open a subject file in pcregrep from fopen(pathname,
1010 "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
1011 in a Windows environment.
1013 3. The pcregrep --count option prints the count for each file even when it is
1014 zero, as does GNU grep. However, pcregrep was also printing all files when
1015 --files-with-matches was added. Now, when both options are given, it prints
1016 counts only for those files that have at least one match. (GNU grep just
1017 prints the file name in this circumstance, but including the count seems
1018 more useful - otherwise, why use --count?) Also ensured that the
1019 combination -clh just lists non-zero counts, with no names.
1021 4. The long form of the pcregrep -F option was incorrectly implemented as
1022 --fixed_strings instead of --fixed-strings. This is an incompatible change,
1023 but it seems right to fix it, and I didn't think it was worth preserving
1024 the old behaviour.
1026 5. The command line items --regex=pattern and --regexp=pattern were not
1027 recognized by pcregrep, which required --regex pattern or --regexp pattern
1028 (with a space rather than an '='). The man page documented the '=' forms,
1029 which are compatible with GNU grep; these now work.
1031 6. No libpcreposix.pc file was created for pkg-config; there was just
1032 libpcre.pc and libpcrecpp.pc. The omission has been rectified.
1034 7. Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
1035 when UCP support is not needed, by modifying the Python script that
1036 generates it from Unicode data files. This should not matter if the module
1037 is correctly used as a library, but I received one complaint about 50K of
1038 unwanted data. My guess is that the person linked everything into his
1039 program rather than using a library. Anyway, it does no harm.
1041 8. A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
1042 was a minimum greater than 1 for a wide character in a possessive
1043 repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
1044 which had an unlimited repeat of a nested, fixed maximum repeat of a wide
1045 character. Chaos in the form of incorrect output or a compiling loop could
1046 result.
1048 9. The restrictions on what a pattern can contain when partial matching is
1049 requested for pcre_exec() have been removed. All patterns can now be
1050 partially matched by this function. In addition, if there are at least two
1051 slots in the offset vector, the offset of the earliest inspected character
1052 for the match and the offset of the end of the subject are set in them when
1053 PCRE_ERROR_PARTIAL is returned.
1055 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
1056 synonymous with PCRE_PARTIAL, for backwards compatibility, and
1057 PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
1058 and may be more useful for multi-segment matching.
1060 11. Partial matching with pcre_exec() is now more intuitive. A partial match
1061 used to be given if ever the end of the subject was reached; now it is
1062 given only if matching could not proceed because another character was
1063 needed. This makes a difference in some odd cases such as Z(*FAIL) with the
1064 string "Z", which now yields "no match" instead of "partial match". In the
1065 case of pcre_dfa_exec(), "no match" is given if every matching path for the
1066 final character ended with (*FAIL).
1068 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
1069 if the pattern had a "must contain" character that was already found in the
1070 earlier partial match, unless partial matching was again requested. For
1071 example, with the pattern /dog.(body)?/, the "must contain" character is
1072 "g". If the first part-match was for the string "dog", restarting with
1073 "sbody" failed. This bug has been fixed.
1075 13. The string returned by pcre_dfa_exec() after a partial match has been
1076 changed so that it starts at the first inspected character rather than the
1077 first character of the match. This makes a difference only if the pattern
1078 starts with a lookbehind assertion or \b or \B (\K is not supported by
1079 pcre_dfa_exec()). It's an incompatible change, but it makes the two
1080 matching functions compatible, and I think it's the right thing to do.
1082 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
1083 so that the demonstration program is easily available in environments where
1084 PCRE has not been installed from source.
1086 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
1087 libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
1088 library.
1090 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
1091 It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
1092 is not the first non-POSIX option to be added. Clearly some people find
1093 these options useful.
1095 17. If a caller to the POSIX matching function regexec() passes a non-zero
1096 value for nmatch with a NULL value for pmatch, the value of
1097 nmatch is forced to zero.
1099 18. RunGrepTest did not have a test for the availability of the -u option of
1100 the diff command, as RunTest does. It now checks in the same way as
1101 RunTest, and also checks for the -b option.
1103 19. If an odd number of negated classes containing just a single character
1104 interposed, within parentheses, between a forward reference to a named
1105 subpattern and the definition of the subpattern, compilation crashed with
1106 an internal error, complaining that it could not find the referenced
1107 subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
1108 [The bug was that it was starting one character too far in when skipping
1109 over the character class, thus treating the ] as data rather than
1110 terminating the class. This meant it could skip too much.]
1112 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
1113 /g option in pcretest when the pattern contains \K, which makes it possible
1114 to have an empty string match not at the start, even when the pattern is
1115 anchored. Updated pcretest and pcredemo to use this option.
1117 21. If the maximum number of capturing subpatterns in a recursion was greater
1118 than the maximum at the outer level, the higher number was returned, but
1119 with unset values at the outer level. The correct (outer level) value is
1120 now given.
1122 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
1123 PCRE did not set those parentheses (unlike Perl). I have now found a way to
1124 make it do so. The string so far is captured, making this feature
1125 compatible with Perl.
1127 23. The tests have been re-organized, adding tests 11 and 12, to make it
1128 possible to check the Perl 5.10 features against Perl 5.10.
1130 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
1131 pattern matches a fixed length string. PCRE did not allow this; now it
1132 does. Neither allows recursion.
1134 25. I finally figured out how to implement a request to provide the minimum
1135 length of subject string that was needed in order to match a given pattern.
1136 (It was back references and recursion that I had previously got hung up
1137 on.) This code has now been added to pcre_study(); it finds a lower bound
1138 to the length of subject needed. It is not necessarily the greatest lower
1139 bound, but using it to avoid searching strings that are too short does give
1140 some useful speed-ups. The value is available to calling programs via
1141 pcre_fullinfo().
1143 26. While implementing 25, I discovered to my embarrassment that pcretest had
1144 not been passing the result of pcre_study() to pcre_dfa_exec(), so the
1145 study optimizations had never been tested with that matching function.
1146 Oops. What is worse, even when it was passed study data, there was a bug in
1147 pcre_dfa_exec() that meant it never actually used it. Double oops. There
1148 were also very few tests of studied patterns with pcre_dfa_exec().
1150 27. If (?| is used to create subpatterns with duplicate numbers, they are now
1151 allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
1152 on the other side of the coin, they are no longer allowed to have different
1153 names, because these cannot be distinguished in PCRE, and this has caused
1154 confusion. (This is a difference from Perl.)
1156 28. When duplicate subpattern names are present (necessarily with different
1157 numbers, as required by 27 above), and a test is made by name in a
1158 conditional pattern, either for a subpattern having been matched, or for
1159 recursion in such a pattern, all the associated numbered subpatterns are
1160 tested, and the overall condition is true if the condition is true for any
1161 one of them. This is the way Perl works, and is also more like the way
1162 testing by number works.
1165 Version 7.9 11-Apr-09
1166 ---------------------
1168 1. When building with support for bzlib/zlib (pcregrep) and/or readline
1169 (pcretest), all targets were linked against these libraries. This included
1170 libpcre, libpcreposix, and libpcrecpp, even though they do not use these
1171 libraries. This caused unwanted dependencies to be created. This problem
1172 has been fixed, and now only pcregrep is linked with bzlib/zlib and only
1173 pcretest is linked with readline.
1175 2. The "typedef int BOOL" in pcre_internal.h that was included inside the
1176 "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
1177 moved outside it again, because FALSE and TRUE are already defined in AIX,
1178 but BOOL is not.
1180 3. The pcre_config() function was treating the PCRE_MATCH_LIMIT and
1181 PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
1183 4. The pcregrep documentation said spaces were inserted as well as colons (or
1184 hyphens) following file names and line numbers when outputting matching
1185 lines. This is not true; no spaces are inserted. I have also clarified the
1186 wording for the --colour (or --color) option.
1188 5. In pcregrep, when --colour was used with -o, the list of matching strings
1189 was not coloured; this is different to GNU grep, so I have changed it to be
1190 the same.
1192 6. When --colo(u)r was used in pcregrep, only the first matching substring in
1193 each matching line was coloured. Now it goes on to look for further matches
1194 of any of the test patterns, which is the same behaviour as GNU grep.
1196 7. A pattern that could match an empty string could cause pcregrep to loop; it
1197 doesn't make sense to accept an empty string match in pcregrep, so I have
1198 locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
1199 seems to be how GNU grep behaves.
1201 8. The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
1202 start or after a newline", because the conditional assertion was not being
1203 correctly handled. The rule now is that both the assertion and what follows
1204 in the first alternative must satisfy the test.
1206 9. If auto-callout was enabled in a pattern with a conditional group whose
1207 condition was an assertion, PCRE could crash during matching, both with
1208 pcre_exec() and pcre_dfa_exec().
1210 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
1211 used for matching.
1213 11. Unicode property support in character classes was not working for
1214 characters (bytes) greater than 127 when not in UTF-8 mode.
1216 12. Added the -M command line option to pcretest.
1218 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
1220 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
1222 16. Added comments and documentation about mis-use of no_arg in the C++
1223 wrapper.
1225 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
1226 from Martin Jerabek that uses macro names for all relevant character and
1227 string constants.
1229 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
1230 SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
1231 SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
1232 these, but not everybody uses configure.
1234 19. A conditional group that had only one branch was not being correctly
1235 recognized as an item that could match an empty string. This meant that an
1236 enclosing group might also not be so recognized, causing infinite looping
1237 (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
1238 with the subject "ab", where knowledge that the repeated group can match
1239 nothing is needed in order to break the loop.
1241 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
1242 exec(), but without supplying a callout function, matching went wrong.
1244 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
1245 leak if the size of the offset vector was greater than 30. When the vector
1246 is smaller, the saved offsets during recursion go onto a local stack
1247 vector, but for larger vectors malloc() is used. It was failing to free
1248 when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
1249 error, in fact).
1251 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
1252 heapframe that is used only when UTF-8 support is enabled. This caused no
1253 problem, but was untidy.
1255 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
1256 CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
1257 included within another project.
1259 24. Steven Van Ingelgem's patches to add more options to the CMake support,
1260 slightly modified by me:
1262 (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
1263 not building pcregrep.
1265 (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
1266 if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
1268 25. Forward references, both numeric and by name, in patterns that made use of
1269 duplicate group numbers, could behave incorrectly or give incorrect errors,
1270 because when scanning forward to find the reference group, PCRE was not
1271 taking into account the duplicate group numbers. A pattern such as
1272 ^X(?3)(a)(?|(b)|(q))(Y) is an example.
1274 26. Changed a few more instances of "const unsigned char *" to USPTR, making
1275 the feature of a custom pointer more persuasive (as requested by a user).
1277 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
1278 pcretest.c, inside #ifndefs, because it seems they are sometimes already
1279 pre-defined.
1281 28. Added support for (*UTF8) at the start of a pattern.
1283 29. Arrange for flags added by the "release type" setting in CMake to be shown
1284 in the configuration summary.
1287 Version 7.8 05-Sep-08
1288 ---------------------
1290 1. Replaced UCP searching code with optimized version as implemented for Ad
1291 Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
1292 stage table and inline lookup instead of a function, giving speed ups of 2
1293 to 5 times on some simple patterns that I tested. Permission was given to
1294 distribute the MultiStage2.py script that generates the tables (it's not in
1295 the tarball, but is in the Subversion repository).
1297 2. Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
1298 scripts.
1300 3. Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
1301 a group with a zero qualifier. The result of the study could be incorrect,
1302 or the function might crash, depending on the pattern.
1304 4. Caseless matching was not working for non-ASCII characters in back
1305 references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
1306 It now works when Unicode Property Support is available.
1308 5. In pcretest, an escape such as \x{de} in the data was always generating
1309 a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
1310 non-UTF-8 mode. If the value is greater than 255, it gives a warning about
1311 truncation.
1313 6. Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
1315 7. Added two (int) casts to pcregrep when printing the difference of two
1316 pointers, in case they are 64-bit values.
1318 8. Added comments about Mac OS X stack usage to the pcrestack man page and to
1319 test 2 if it fails.
1321 9. Added PCRE_CALL_CONVENTION just before the names of all exported functions,
1322 and a #define of that name to empty if it is not externally set. This is to
1323 allow users of MSVC to set it if necessary.
1325 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
1326 the convenience functions in the pcre_get.c source file.
1328 11. An option change at the start of a pattern that had top-level alternatives
1329 could cause overwriting and/or a crash. This command provoked a crash in
1330 some environments:
1332 printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
1334 This potential security problem was recorded as CVE-2008-2371.
1336 12. For a pattern where the match had to start at the beginning or immediately
1337 after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
1338 pcre_dfa_exec() could read past the end of the passed subject if there was
1339 no match. To help with detecting such bugs (e.g. with valgrind), I modified
1340 pcretest so that it places the subject at the end of its malloc-ed buffer.
1342 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
1343 exec() might read past the end of the data buffer in UTF-8 mode.
1345 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
1346 the data contained the byte 0x85 as part of a UTF-8 character within its
1347 first line. This applied both to normal and DFA matching.
1349 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
1350 /^[^d]*?$/8 failed to match "abc".
1352 16. Added a missing copyright notice to pcrecpp_internal.h.
1354 17. Make it more clear in the documentation that values returned from
1355 pcre_exec() in ovector are byte offsets, not character counts.
1357 18. Tidied a few places to stop certain compilers from issuing warnings.
1359 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
1360 supplied by Stefan Weber. I made a further small update for 7.8 because
1361 there is a change of source arrangements: the pcre_searchfuncs.c module is
1362 replaced by pcre_ucd.c.
1365 Version 7.7 07-May-08
1366 ---------------------
1368 1. Applied Craig's patch to sort out a long long problem: "If we can't convert
1369 a string to a long long, pretend we don't even have a long long." This is
1370 done by checking for the strtoq, strtoll, and _strtoi64 functions.
1372 2. Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
1373 pre-7.6 versions, which defined a global no_arg variable instead of putting
1374 it in the RE class. (See also #8 below.)
1376 3. Remove a line of dead code, identified by coverity and reported by Nuno
1377 Lopes.
1379 4. Fixed two related pcregrep bugs involving -r with --include or --exclude:
1381 (1) The include/exclude patterns were being applied to the whole pathnames
1382 of files, instead of just to the final components.
1384 (2) If there was more than one level of directory, the subdirectories were
1385 skipped unless they satisfied the include/exclude conditions. This is
1386 inconsistent with GNU grep (and could even be seen as contrary to the
1387 pcregrep specification - which I improved to make it absolutely clear).
1388 The action now is always to scan all levels of directory, and just
1389 apply the include/exclude patterns to regular files.
1391 5. Added the --include_dir and --exclude_dir patterns to pcregrep, and used
1392 --exclude_dir in the tests to avoid scanning .svn directories.
1394 6. Applied Craig's patch to the QuoteMeta function so that it escapes the
1395 NUL character as backslash + 0 rather than backslash + NUL, because PCRE
1396 doesn't support NULs in patterns.
1398 7. Added some missing "const"s to declarations of static tables in
1399 pcre_compile.c and pcre_dfa_exec.c.
1401 8. Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
1402 caused by fix #2 above. (Subsequently also a second patch to fix the
1403 first patch. And a third patch - this was a messy problem.)
1405 9. Applied Craig's patch to remove the use of push_back().
1407 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
1408 matching function regexec().
1410 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
1411 which, however, unlike Perl's \g{...}, are subroutine calls, not back
1412 references. PCRE supports relative numbers with this syntax (I don't think
1413 Oniguruma does).
1415 12. Previously, a group with a zero repeat such as (...){0} was completely
1416 omitted from the compiled regex. However, this means that if the group
1417 was called as a subroutine from elsewhere in the pattern, things went wrong
1418 (an internal error was given). Such groups are now left in the compiled
1419 pattern, with a new opcode that causes them to be skipped at execution
1420 time.
1422 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
1423 to the way PCRE behaves:
1425 (a) A lone ] character is dis-allowed (Perl treats it as data).
1427 (b) A back reference to an unmatched subpattern matches an empty string
1428 (Perl fails the current match path).
1430 (c) A data ] in a character class must be notated as \] because if the
1431 first data character in a class is ], it defines an empty class. (In
1432 Perl it is not possible to have an empty class.) The empty class []
1433 never matches; it forces failure and is equivalent to (*FAIL) or (?!).
1434 The negative empty class [^] matches any one character, independently
1435 of the DOTALL setting.
1437 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
1438 non-existent subpattern following a character class starting with ']' and
1439 containing () gave an internal compiling error instead of "reference to
1440 non-existent subpattern". Fortunately, when the pattern did exist, the
1441 compiled code was correct. (When scanning forwards to check for the
1442 existencd of the subpattern, it was treating the data ']' as terminating
1443 the class, so got the count wrong. When actually compiling, the reference
1444 was subsequently set up correctly.)
1446 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
1447 it was being rejected as not supported by pcre_dfa_exec(), even though
1448 other assertions are supported. I have made pcre_dfa_exec() support
1449 (*FAIL).
1451 16. The implementation of 13c above involved the invention of a new opcode,
1452 OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
1453 cannot be changed at match time, I realized I could make a small
1454 improvement to matching performance by compiling OP_ALLANY instead of
1455 OP_ANY for "." when DOTALL was set, and then removing the runtime tests
1456 on the OP_ANY path.
1458 17. Compiling pcretest on Windows with readline support failed without the
1459 following two fixes: (1) Make the unistd.h include conditional on
1460 HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
1462 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
1463 ncurses library to be included for pcretest when ReadLine support is
1464 requested, but also to allow for it to be overridden. This patch came from
1465 Daniel Bergström.
1467 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
1468 as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
1469 any errors with the current Unicode tables. Thanks to Peter Kankowski for
1470 spotting this.
1473 Version 7.6 28-Jan-08
1474 ---------------------
1476 1. A character class containing a very large number of characters with
1477 codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
1478 overflow.
1480 2. Patch to cut out the "long long" test in pcrecpp_unittest when
1481 HAVE_LONG_LONG is not defined.
1483 3. Applied Christian Ehrlicher's patch to update the CMake build files to
1484 bring them up to date and include new features. This patch includes:
1486 - Fixed PH's badly added libz and libbz2 support.
1487 - Fixed a problem with static linking.
1488 - Added pcredemo. [But later removed - see 7 below.]
1489 - Fixed dftables problem and added an option.
1490 - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
1492 - Added readline support for pcretest.
1493 - Added an listing of the option settings after cmake has run.
1495 4. A user submitted a patch to Makefile that makes it easy to create
1496 "pcre.dll" under mingw when using Configure/Make. I added stuff to
1497 Makefile.am that cause it to include this special target, without
1498 affecting anything else. Note that the same mingw target plus all
1499 the other distribution libraries and programs are now supported
1500 when configuring with CMake (see 6 below) instead of with
1501 Configure/Make.
1503 5. Applied Craig's patch that moves no_arg into the RE class in the C++ code.
1504 This is an attempt to solve the reported problem "pcrecpp::no_arg is not
1505 exported in the Windows port". It has not yet been confirmed that the patch
1506 solves the problem, but it does no harm.
1508 6. Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
1509 NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
1510 with CMake, and also correct the comment about stack recursion.
1512 7. Remove the automatic building of pcredemo from the ./configure system and
1513 from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
1514 of a program that users should build themselves after PCRE is installed, so
1515 building it automatically is not really right. What is more, it gave
1516 trouble in some build environments.
1518 8. Further tidies to CMakeLists.txt from Sheri and Christian.
1521 Version 7.5 10-Jan-08
1522 ---------------------
1524 1. Applied a patch from Craig: "This patch makes it possible to 'ignore'
1525 values in parens when parsing an RE using the C++ wrapper."
1527 2. Negative specials like \S did not work in character classes in UTF-8 mode.
1528 Characters greater than 255 were excluded from the class instead of being
1529 included.
1531 3. The same bug as (2) above applied to negated POSIX classes such as
1532 [:^space:].
1534 4. PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
1535 defined or documented. It seems to have been a typo for PCRE_STATIC, so
1536 I have changed it.
1538 5. The construct (?&) was not diagnosed as a syntax error (it referenced the
1539 first named subpattern) and a construct such as (?&a) would reference the
1540 first named subpattern whose name started with "a" (in other words, the
1541 length check was missing). Both these problems are fixed. "Subpattern name
1542 expected" is now given for (?&) (a zero-length name), and this patch also
1543 makes it give the same error for \k'' (previously it complained that that
1544 was a reference to a non-existent subpattern).
1546 6. The erroneous patterns (?+-a) and (?-+a) give different error messages;
1547 this is right because (?- can be followed by option settings as well as by
1548 digits. I have, however, made the messages clearer.
1550 7. Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
1551 than the number used in the conditional) now cause a compile-time error.
1552 This is actually not compatible with Perl, which accepts such patterns, but
1553 treats the conditional as always being FALSE (as PCRE used to), but it
1554 seems to me that giving a diagnostic is better.
1556 8. Change "alphameric" to the more common word "alphanumeric" in comments
1557 and messages.
1559 9. Fix two occurrences of "backslash" in comments that should have been
1560 "backspace".
1562 10. Remove two redundant lines of code that can never be obeyed (their function
1563 was moved elsewhere).
1565 11. The program that makes PCRE's Unicode character property table had a bug
1566 which caused it to generate incorrect table entries for sequences of
1567 characters that have the same character type, but are in different scripts.
1568 It amalgamated them into a single range, with the script of the first of
1569 them. In other words, some characters were in the wrong script. There were
1570 thirteen such cases, affecting characters in the following ranges:
1572 U+002b0 - U+002c1
1573 U+0060c - U+0060d
1574 U+0061e - U+00612
1575 U+0064b - U+0065e
1576 U+0074d - U+0076d
1577 U+01800 - U+01805
1578 U+01d00 - U+01d77
1579 U+01d9b - U+01dbf
1580 U+0200b - U+0200f
1581 U+030fc - U+030fe
1582 U+03260 - U+0327f
1583 U+0fb46 - U+0fbb1
1584 U+10450 - U+1049d
1586 12. The -o option (show only the matching part of a line) for pcregrep was not
1587 compatible with GNU grep in that, if there was more than one match in a
1588 line, it showed only the first of them. It now behaves in the same way as
1589 GNU grep.
1591 13. If the -o and -v options were combined for pcregrep, it printed a blank
1592 line for every non-matching line. GNU grep prints nothing, and pcregrep now
1593 does the same. The return code can be used to tell if there were any
1594 non-matching lines.
1596 14. Added --file-offsets and --line-offsets to pcregrep.
1598 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
1599 infinitely looping recursion. The bug was that positive lookaheads were not
1600 being skipped when checking for a possible empty match (negative lookaheads
1601 and both kinds of lookbehind were skipped).
1603 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
1604 inclusion of <windows.h> to before rather than after the definition of
1605 INVALID_FILE_ATTRIBUTES (patch from David Byron).
1607 17. Specifying a possessive quantifier with a specific limit for a Unicode
1608 character property caused pcre_compile() to compile bad code, which led at
1609 runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
1610 are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
1611 caused the error; without that there was no problem.
1613 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
1615 19. Added --enable-pcretest-libreadline.
1617 20. In pcrecpp.cc, the variable 'count' was incremented twice in
1618 RE::GlobalReplace(). As a result, the number of replacements returned was
1619 double what it should be. I removed one of the increments, but Craig sent a
1620 later patch that removed the other one (the right fix) and added unit tests
1621 that check the return values (which was not done before).
1623 21. Several CMake things:
1625 (1) Arranged that, when cmake is used on Unix, the libraries end up with
1626 the names libpcre and libpcreposix, not just pcre and pcreposix.
1628 (2) The above change means that pcretest and pcregrep are now correctly
1629 linked with the newly-built libraries, not previously installed ones.
1633 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
1634 crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
1635 UTF-8 newline character). The key issue is that the pattern starts .*;
1636 this means that the match must be either at the beginning, or after a
1637 newline. The bug was in the code for advancing after a failed match and
1638 checking that the new position followed a newline. It was not taking
1639 account of UTF-8 characters correctly.
1641 23. PCRE was behaving differently from Perl in the way it recognized POSIX
1642 character classes. PCRE was not treating the sequence [:...:] as a
1643 character class unless the ... were all letters. Perl, however, seems to
1644 allow any characters between [: and :], though of course it rejects as
1645 unknown any "names" that contain non-letters, because all the known class
1646 names consist only of letters. Thus, Perl gives an error for [[:1234:]],
1647 for example, whereas PCRE did not - it did not recognize a POSIX character
1648 class. This seemed a bit dangerous, so the code has been changed to be
1649 closer to Perl. The behaviour is not identical to Perl, because PCRE will
1650 diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
1651 treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
1652 Perl does, and where it didn't before.
1654 24. Rewrite so as to remove the single use of %n from pcregrep because in some
1655 Windows environments %n is disabled by default.
1658 Version 7.4 21-Sep-07
1659 ---------------------
1661 1. Change 7.3/28 was implemented for classes by looking at the bitmap. This
1662 means that a class such as [\s] counted as "explicit reference to CR or
1663 LF". That isn't really right - the whole point of the change was to try to
1664 help when there was an actual mention of one of the two characters. So now
1665 the change happens only if \r or \n (or a literal CR or LF) character is
1666 encountered.
1668 2. The 32-bit options word was also used for 6 internal flags, but the numbers
1669 of both had grown to the point where there were only 3 bits left.
1670 Fortunately, there was spare space in the data structure, and so I have
1671 moved the internal flags into a new 16-bit field to free up more option
1672 bits.
1674 3. The appearance of (?J) at the start of a pattern set the DUPNAMES option,
1675 but did not set the internal JCHANGED flag - either of these is enough to
1676 control the way the "get" function works - but the PCRE_INFO_JCHANGED
1677 facility is supposed to tell if (?J) was ever used, so now (?J) at the
1678 start sets both bits.
1680 4. Added options (at build time, compile time, exec time) to change \R from
1681 matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
1683 5. doc/pcresyntax.html was missing from the distribution.
1685 6. Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
1686 compatibility, even though it is no longer used.
1688 7. Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
1689 strtoull to pcrecpp.cc to select the available functions in WIN32 when the
1690 windows.h file is present (where different names are used). [This was
1691 reversed later after testing - see 16 below.]
1693 8. Changed all #include <config.h> to #include "config.h". There were also
1694 some further <pcre.h> cases that I changed to "pcre.h".
1696 9. When pcregrep was used with the --colour option, it missed the line ending
1697 sequence off the lines that it output.
1699 10. It was pointed out to me that arrays of string pointers cause lots of
1700 relocations when a shared library is dynamically loaded. A technique of
1701 using a single long string with a table of offsets can drastically reduce
1702 these. I have refactored PCRE in four places to do this. The result is
1703 dramatic:
1705 Originally: 290
1706 After changing UCP table: 187
1707 After changing error message table: 43
1708 After changing table of "verbs" 36
1709 After changing table of Posix names 22
1711 Thanks to the folks working on Gregex for glib for this insight.
1713 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
1714 unicode-properties was also set.
1716 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
1718 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
1719 checked only for CRLF.
1721 14. Added casts to pcretest.c to avoid compiler warnings.
1723 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
1725 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
1726 and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
1727 entirely. This removes changes made in 7 above.
1729 17. The CMake files have been updated, and there is now more information about
1730 building with CMake in the NON-UNIX-USE document.
1733 Version 7.3 28-Aug-07
1734 ---------------------
1736 1. In the rejigging of the build system that eventually resulted in 7.1, the
1737 line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
1738 brackets there is not right, since it causes compilers to look for an
1739 installed pcre.h, not the version that is in the source that is being
1740 compiled (which of course may be different). I have changed it back to:
1742 #include "pcre.h"
1744 I have a vague recollection that the change was concerned with compiling in
1745 different directories, but in the new build system, that is taken care of
1746 by the VPATH setting the Makefile.
1748 2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
1749 when the subject happened to end in the byte 0x85 (e.g. if the last
1750 character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
1751 characters but of course it shouldn't be taken as a newline when it is part
1752 of another character. The bug was that, for an unlimited repeat of . in
1753 not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
1754 characters when looking for a newline.
1756 3. A small performance improvement in the DOTALL UTF-8 mode .* case.
1758 4. Debugging: adjusted the names of opcodes for different kinds of parentheses
1759 in debug output.
1761 5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
1762 long printing in the pcrecpp unittest when running under MinGW.
1764 6. ESC_K was left out of the EBCDIC table.
1766 7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
1767 parentheses; I made it 1000, which seemed large enough. Unfortunately, the
1768 limit also applies to "virtual nesting" when a pattern is recursive, and in
1769 this case 1000 isn't so big. I have been able to remove this limit at the
1770 expense of backing off one optimization in certain circumstances. Normally,
1771 when pcre_exec() would call its internal match() function recursively and
1772 immediately return the result unconditionally, it uses a "tail recursion"
1773 feature to save stack. However, when a subpattern that can match an empty
1774 string has an unlimited repetition quantifier, it no longer makes this
1775 optimization. That gives it a stack frame in which to save the data for
1776 checking that an empty string has been matched. Previously this was taken
1777 from the 1000-entry workspace that had been reserved. So now there is no
1778 explicit limit, but more stack is used.
1780 8. Applied Daniel's patches to solve problems with the import/export magic
1781 syntax that is required for Windows, and which was going wrong for the
1782 pcreposix and pcrecpp parts of the library. These were overlooked when this
1783 problem was solved for the main library.
1785 9. There were some crude static tests to avoid integer overflow when computing
1786 the size of patterns that contain repeated groups with explicit upper
1787 limits. As the maximum quantifier is 65535, the maximum group length was
1788 set at 30,000 so that the product of these two numbers did not overflow a
1789 32-bit integer. However, it turns out that people want to use groups that
1790 are longer than 30,000 bytes (though not repeat them that many times).
1791 Change 7.0/17 (the refactoring of the way the pattern size is computed) has
1792 made it possible to implement the integer overflow checks in a much more
1793 dynamic way, which I have now done. The artificial limitation on group
1794 length has been removed - we now have only the limit on the total length of
1795 the compiled pattern, which depends on the LINK_SIZE setting.
1797 10. Fixed a bug in the documentation for get/copy named substring when
1798 duplicate names are permitted. If none of the named substrings are set, the
1799 functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
1800 empty string.
1802 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
1803 instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
1804 because the ] is interpreted as the first data character and the
1805 terminating ] is not found. PCRE has been made compatible with Perl in this
1806 regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
1807 cause memory overwriting.
1809 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
1810 string has been matched (to stop an infinite loop). It was not recognizing
1811 a conditional subpattern that could match an empty string if that
1812 subpattern was within another subpattern. For example, it looped when
1813 trying to match (((?(1)X|))*) but it was OK with ((?(1)X|)*) where the
1814 condition was not nested. This bug has been fixed.
1816 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
1817 past the start of the subject in the presence of bytes with the top bit
1818 set, for example "\x8aBCD".
1820 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
1821 (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
1823 14. Optimized (?!) to (*FAIL).
1825 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
1826 This restricts code points to be within the range 0 to 0x10FFFF, excluding
1827 the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
1828 full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
1829 does: it's just the validity check that is more restrictive.
1831 16. Inserted checks for integer overflows during escape sequence (backslash)
1832 processing, and also fixed erroneous offset values for syntax errors during
1833 backslash processing.
1835 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
1836 for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
1838 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
1839 caused an overrun.
1841 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
1842 something other than just ASCII characters) inside a group that had an
1843 unlimited repeat caused a loop at compile time (while checking to see
1844 whether the group could match an empty string).
1846 20. Debugging a pattern containing \p or \P could cause a crash. For example,
1847 [\P{Any}] did so. (Error in the code for printing property names.)
1849 21. An orphan \E inside a character class could cause a crash.
1851 22. A repeated capturing bracket such as (A)? could cause a wild memory
1852 reference during compilation.
1854 23. There are several functions in pcre_compile() that scan along a compiled
1855 expression for various reasons (e.g. to see if it's fixed length for look
1856 behind). There were bugs in these functions when a repeated \p or \P was
1857 present in the pattern. These operators have additional parameters compared
1858 with \d, etc, and these were not being taken into account when moving along
1859 the compiled data. Specifically:
1861 (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
1862 length.
1864 (b) An item such as \pL+ within a repeated group could cause crashes or
1865 loops.
1867 (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
1868 "reference to non-existent subpattern" error.
1870 (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
1872 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
1873 characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
1875 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
1877 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
1878 character were causing crashes (broken optimization).
1880 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
1881 \p or \P) caused a compile-time loop.
1883 28. More problems have arisen in unanchored patterns when CRLF is a valid line
1884 break. For example, the unstudied pattern [\r\n]A does not match the string
1885 "\r\nA" because change 7.0/46 below moves the current point on by two
1886 characters after failing to match at the start. However, the pattern \nA
1887 *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
1888 the same is true. There doesn't seem any very clean way out of this, but
1889 what I have chosen to do makes the common cases work: PCRE now takes note
1890 of whether there can be an explicit match for \r or \n anywhere in the
1891 pattern, and if so, 7.0/46 no longer applies. As part of this change,
1892 there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
1893 pattern has explicit CR or LF references.
1895 29. Added (*CR) etc for changing newline setting at start of pattern.
1898 Version 7.2 19-Jun-07
1899 ---------------------
1901 1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
1902 which is apparently normally available under Windows.
1904 2. Re-jig the pcregrep tests with different newline settings in an attempt
1905 to make them independent of the local environment's newline setting.
1907 3. Add code to configure.ac to remove -g from the CFLAGS default settings.
1909 4. Some of the "internals" tests were previously cut out when the link size
1910 was not 2, because the output contained actual offsets. The recent new
1911 "Z" feature of pcretest means that these can be cut out, making the tests
1912 usable with all link sizes.
1914 5. Implemented Stan Switzer's goto replacement for longjmp() when not using
1915 stack recursion. This gives a massive performance boost under BSD, but just
1916 a small improvement under Linux. However, it saves one field in the frame
1917 in all cases.
1919 6. Added more features from the forthcoming Perl 5.10:
1921 (a) (?-n) (where n is a string of digits) is a relative subroutine or
1922 recursion call. It refers to the nth most recently opened parentheses.
1924 (b) (?+n) is also a relative subroutine call; it refers to the nth next
1925 to be opened parentheses.
1927 (c) Conditions that refer to capturing parentheses can be specified
1928 relatively, for example, (?(-2)... or (?(+3)...
1930 (d) \K resets the start of the current match so that everything before
1931 is not part of it.
1933 (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
1935 (f) \g{name} is another synonym - part of Perl 5.10's unification of
1936 reference syntax.
1938 (g) (?| introduces a group in which the numbering of parentheses in each
1939 alternative starts with the same number.
1941 (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
1943 7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
1946 8. A pattern such as (.*(.)?)* caused pcre_exec() to fail by either not
1947 terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
1948 for detecting groups that can match an empty string.
1950 9. A pattern with a very large number of alternatives (more than several
1951 hundred) was running out of internal workspace during the pre-compile
1952 phase, where pcre_compile() figures out how much memory will be needed. A
1953 bit of new cunning has reduced the workspace needed for groups with
1954 alternatives. The 1000-alternative test pattern now uses 12 bytes of
1955 workspace instead of running out of the 4096 that are available.
1957 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
1959 11. Applied patch from Google to remove an optimization that didn't quite work.
1960 The report of the bug said:
1962 pcrecpp::RE("a*").FullMatch("aaa") matches, while
1963 pcrecpp::RE("a*?").FullMatch("aaa") does not, and
1964 pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
1966 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
1967 it matched the wrong number of bytes.
1970 Version 7.1 24-Apr-07
1971 ---------------------
1973 1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
1974 that is more "standard", making use of automake and other Autotools. There
1975 is some re-arrangement of the files and adjustment of comments consequent
1976 on this.
1978 2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
1979 for recursive directory scanning broke on some systems because the files
1980 are not scanned in any specific order and on different systems the order
1981 was different. A call to "sort" has been inserted into RunGrepTest for the
1982 approprate test as a short-term fix. In the longer term there may be an
1983 alternative.
1985 3. I had an email from Eric Raymond about problems translating some of PCRE's
1986 man pages to HTML (despite the fact that I distribute HTML pages, some
1987 people do their own conversions for various reasons). The problems
1988 concerned the use of low-level troff macros .br and .in. I have therefore
1989 removed all such uses from the man pages (some were redundant, some could
1990 be replaced by .nf/.fi pairs). The 132html script that I use to generate
1991 HTML has been updated to handle .nf/.fi and to complain if it encounters
1992 .br or .in.
1994 4. Updated comments in configure.ac that get placed in config.h.in and also
1995 arranged for config.h to be included in the distribution, with the name
1996 config.h.generic, for the benefit of those who have to compile without
1997 Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
1999 5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
2000 Weber: (1) pcre_internal.h was missing some function renames; (2) updated
2001 makevp.bat for the current PCRE, using the additional files
2002 makevp_c.txt, makevp_l.txt, and pcregexp.pas.
2004 6. A Windows user reported a minor discrepancy with test 2, which turned out
2005 to be caused by a trailing space on an input line that had got lost in his
2006 copy. The trailing space was an accident, so I've just removed it.
2008 7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
2009 that is needed.
2011 8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
2012 as "const" (a) because they are and (b) because it helps the PHP
2013 maintainers who have recently made a script to detect big data structures
2014 in the php code that should be moved to the .rodata section. I remembered
2015 to update Builducptable as well, so it won't revert if ucptable.h is ever
2016 re-created.
2018 9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
2019 pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
2020 order to be able to cut out the UTF-8 tables in the latter when UTF-8
2021 support is not required. This saves 1.5-2K of code, which is important in
2022 some applications.
2024 Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
2025 so as not to refer to the tables, even though these functions will never be
2026 called when UTF-8 support is disabled. Otherwise there are problems with a
2027 shared library.
2029 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
2031 (a) It was defining its arguments as char * instead of void *.
2033 (b) It was assuming that all moves were upwards in memory; this was true
2034 a long time ago when I wrote it, but is no longer the case.
2036 The emulated memove() is provided for those environments that have neither
2037 memmove() nor bcopy(). I didn't think anyone used it these days, but that
2038 is clearly not the case, as these two bugs were recently reported.
2040 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
2041 and Detrail to create the HTML documentation, the .txt form of the man
2042 pages, and it removes trailing spaces from listed files. It also creates
2043 pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
2044 case, it wraps all the #defines with #ifndefs. This script should be run
2045 before "make dist".
2047 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
2048 with Unicode property support.
2050 (a) For a maximizing quantifier, if the two different cases of the
2051 character were of different lengths in their UTF-8 codings (there are
2052 some cases like this - I found 11), and the matching function had to
2053 back up over a mixture of the two cases, it incorrectly assumed they
2054 were both the same length.
2056 (b) When PCRE was configured to use the heap rather than the stack for
2057 recursion during matching, it was not correctly preserving the data for
2058 the other case of a UTF-8 character when checking ahead for a match
2059 while processing a minimizing repeat. If the check also involved
2060 matching a wide character, but failed, corruption could cause an
2061 erroneous result when trying to check for a repeat of the original
2062 character.
2064 13. Some tidying changes to the testing mechanism:
2066 (a) The RunTest script now detects the internal link size and whether there
2067 is UTF-8 and UCP support by running ./pcretest -C instead of relying on
2068 values substituted by "configure". (The RunGrepTest script already did
2069 this for UTF-8.) The configure.ac script no longer substitutes the
2070 relevant variables.
2072 (b) The debugging options /B and /D in pcretest show the compiled bytecode
2073 with length and offset values. This means that the output is different
2074 for different internal link sizes. Test 2 is skipped for link sizes
2075 other than 2 because of this, bypassing the problem. Unfortunately,
2076 there was also a test in test 3 (the locale tests) that used /B and
2077 failed for link sizes other than 2. Rather than cut the whole test out,
2078 I have added a new /Z option to pcretest that replaces the length and
2079 offset values with spaces. This is now used to make test 3 independent
2080 of link size. (Test 2 will be tidied up later.)
2082 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
2083 segmentation fault instead of returning the appropriate error message.
2085 15. In multiline mode when the newline sequence was set to "any", the pattern
2086 ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
2087 This doesn't seem right; it now treats the CRLF combination as the line
2088 ending, and so does not match in that case. It's only a pattern such as ^$
2089 that would hit this one: something like ^ABC$ would have failed after \r
2090 and then tried again after \r\n.
2092 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
2093 in an attempt to make files that differ only in their line terminators
2094 compare equal. This works on Linux.
2096 17. Under certain error circumstances pcregrep might try to free random memory
2097 as it exited. This is now fixed, thanks to valgrind.
2099 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
2100 "abc\r\n\r\n", it found an unwanted second match after the second \r. This
2101 was because its rules for how to advance for /g after matching an empty
2102 string at the end of a line did not allow for this case. They now check for
2103 it specially.
2105 20. pcretest is supposed to handle patterns and data of any length, by
2106 extending its buffers when necessary. It was getting this wrong when the
2107 buffer for a data line had to be extended.
2109 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
2110 CRLF as a newline sequence.
2112 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
2113 out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
2114 I have nevertheless tidied it up.
2116 23. Added some casts to kill warnings from HP-UX ia64 compiler.
2118 24. Added a man page for pcre-config.
2121 Version 7.0 19-Dec-06
2122 ---------------------
2124 1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
2125 moving to gcc 4.1.1.
2127 2. The -S option for pcretest uses setrlimit(); I had omitted to #include
2128 sys/time.h, which is documented as needed for this function. It doesn't
2129 seem to matter on Linux, but it showed up on some releases of OS X.
2131 3. It seems that there are systems where bytes whose values are greater than
2132 127 match isprint() in the "C" locale. The "C" locale should be the
2133 default when a C program starts up. In most systems, only ASCII printing
2134 characters match isprint(). This difference caused the output from pcretest
2135 to vary, making some of the tests fail. I have changed pcretest so that:
2137 (a) When it is outputting text in the compiled version of a pattern, bytes
2138 other than 32-126 are always shown as hex escapes.
2140 (b) When it is outputting text that is a matched part of a subject string,
2141 it does the same, unless a different locale has been set for the match
2142 (using the /L modifier). In this case, it uses isprint() to decide.
2144 4. Fixed a major bug that caused incorrect computation of the amount of memory
2145 required for a compiled pattern when options that changed within the
2146 pattern affected the logic of the preliminary scan that determines the
2147 length. The relevant options are -x, and -i in UTF-8 mode. The result was
2148 that the computed length was too small. The symptoms of this bug were
2149 either the PCRE error "internal error: code overflow" from pcre_compile(),
2150 or a glibc crash with a message such as "pcretest: free(): invalid next
2151 size (fast)". Examples of patterns that provoked this bug (shown in
2152 pcretest format) are:
2154 /(?-x: )/x
2155 /(?x)(?-x: \s*#\s*)/
2156 /((?i)[\x{c0}])/8
2157 /(?i:[\x{c0}])/8
2159 HOWEVER: Change 17 below makes this fix obsolete as the memory computation
2160 is now done differently.
2162 5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
2163 wrapper classes; (b) implement a new function in the C++ scanner that is
2164 more efficient than the old way of doing things because it avoids levels of
2165 recursion in the regex matching; (c) add a paragraph to the documentation
2166 for the FullMatch() function.
2168 6. The escape sequence \n was being treated as whatever was defined as
2169 "newline". Not only was this contrary to the documentation, which states
2170 that \n is character 10 (hex 0A), but it also went horribly wrong when
2171 "newline" was defined as CRLF. This has been fixed.
2173 7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
2174 was being set to -1 for the "end of line" case (supposedly a value that no
2175 character can have). Though this value is never used (the check for end of
2176 line is "zero bytes in current character"), it caused compiler complaints.
2177 I've changed it to 0xffffffff.
2179 8. In pcre_version.c, the version string was being built by a sequence of
2180 C macros that, in the event of PCRE_PRERELEASE being defined as an empty
2181 string (as it is for production releases) called a macro with an empty
2182 argument. The C standard says the result of this is undefined. The gcc
2183 compiler treats it as an empty string (which was what was wanted) but it is
2184 reported that Visual C gives an error. The source has been hacked around to
2185 avoid this problem.
2187 9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
2188 builds of pcretest, and changed the call to _setmode() to use _O_BINARY
2189 instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
2190 of them did).
2192 10. Originally, pcretest opened its input and output without "b"; then I was
2193 told that "b" was needed in some environments, so it was added for release
2194 5.0 to both the input and output. (It makes no difference on Unix-like
2195 systems.) Later I was told that it is wrong for the input on Windows. I've
2196 now abstracted the modes into two macros, to make it easier to fiddle with
2197 them, and removed "b" from the input mode under Windows.
2199 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
2201 12. Added -help and --help to pcretest as an official way of being reminded
2202 of the options.
2204 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
2205 and pcrecpp.cc because they annoy compilers at high warning levels.
2207 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
2209 15. Fixed an occurrence of == in configure.ac that should have been = (shell
2210 scripts are not C programs :-) and which was not noticed because it works
2211 on Linux.
2213 16. pcretest is supposed to handle any length of pattern and data line (as one
2214 line or as a continued sequence of lines) by extending its input buffer if
2215 necessary. This feature was broken for very long pattern lines, leading to
2216 a string of junk being passed to pcre_compile() if the pattern was longer
2217 than about 50K.
2219 17. I have done a major re-factoring of the way pcre_compile() computes the
2220 amount of memory needed for a compiled pattern. Previously, there was code
2221 that made a preliminary scan of the pattern in order to do this. That was
2222 OK when PCRE was new, but as the facilities have expanded, it has become
2223 harder and harder to keep it in step with the real compile phase, and there
2224 have been a number of bugs (see for example, 4 above). I have now found a
2225 cunning way of running the real compile function in a "fake" mode that
2226 enables it to compute how much memory it would need, while actually only
2227 ever using a few hundred bytes of working memory and without too many
2228 tests of the mode. This should make future maintenance and development
2229 easier. A side effect of this work is that the limit of 200 on the nesting
2230 depth of parentheses has been removed (though this was never a serious
2231 limitation, I suspect). However, there is a downside: pcre_compile() now
2232 runs more slowly than before (30% or more, depending on the pattern). I
2233 hope this isn't a big issue. There is no effect on runtime performance.
2235 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
2236 newline (only possible for the last line of a file) and it was a
2237 pattern that set a locale (followed by /Lsomething), pcretest crashed.
2239 19. Added additional timing features to pcretest. (1) The -tm option now times
2240 matching only, not compiling. (2) Both -t and -tm can be followed, as a
2241 separate command line item, by a number that specifies the number of
2242 repeats to use when timing. The default is 50000; this gives better
2243 precision, but takes uncomfortably long for very large patterns.
2245 20. Extended pcre_study() to be more clever in cases where a branch of a
2246 subpattern has no definite first character. For example, (a*|b*)[cd] would
2247 previously give no result from pcre_study(). Now it recognizes that the
2248 first character must be a, b, c, or d.
2250 21. There was an incorrect error "recursive call could loop indefinitely" if
2251 a subpattern (or the entire pattern) that was being tested for matching an
2252 empty string contained only one non-empty item after a nested subpattern.
2253 For example, the pattern (?>\x{100}*)\d(?R) provoked this error
2254 incorrectly, because the \d was being skipped in the check.
2256 22. The pcretest program now has a new pattern option /B and a command line
2257 option -b, which is equivalent to adding /B to every pattern. This causes
2258 it to show the compiled bytecode, without the additional information that
2259 -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
2260 is the same as /B/I).
2262 23. A new optimization is now able automatically to treat some sequences such
2263 as a*b as a*+b. More specifically, if something simple (such as a character
2264 or a simple class like \d) has an unlimited quantifier, and is followed by
2265 something that cannot possibly match the quantified thing, the quantifier
2266 is automatically "possessified".
2268 24. A recursive reference to a subpattern whose number was greater than 39
2269 went wrong under certain circumstances in UTF-8 mode. This bug could also
2270 have affected the operation of pcre_study().
2272 25. Realized that a little bit of performance could be had by replacing
2273 (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
2275 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
2277 27. Possessive quantifiers such as a++ were previously implemented by turning
2278 them into atomic groups such as ($>a+). Now they have their own opcodes,
2279 which improves performance. This includes the automatically created ones
2280 from 23 above.
2282 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
2283 lookahead was broken if it was not anchored. PCRE was mistakenly expecting
2284 the first matched character to be a colon. This applied both to named and
2285 numbered groups.
2287 29. The ucpinternal.h header file was missing its idempotency #ifdef.
2289 30. I was sent a "project" file called libpcre.a.dev which I understand makes
2290 building PCRE on Windows easier, so I have included it in the distribution.
2292 31. There is now a check in pcretest against a ridiculously large number being
2293 returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
2294 loop, the loop is abandoned.
2296 32. Forward references to subpatterns in conditions such as (?(2)...) where
2297 subpattern 2 is defined later cause pcre_compile() to search forwards in
2298 the pattern for the relevant set of parentheses. This search went wrong
2299 when there were unescaped parentheses in a character class, parentheses
2300 escaped with \Q...\E, or parentheses in a #-comment in /x mode.
2302 33. "Subroutine" calls and backreferences were previously restricted to
2303 referencing subpatterns earlier in the regex. This restriction has now
2304 been removed.
2306 34. Added a number of extra features that are going to be in Perl 5.10. On the
2307 whole, these are just syntactic alternatives for features that PCRE had
2308 previously implemented using the Python syntax or my own invention. The
2309 other formats are all retained for compatibility.
2311 (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
2312 as (?P<name>...). The new forms, as well as being in Perl 5.10, are
2313 also .NET compatible.
2315 (b) A recursion or subroutine call to a named group can now be defined as
2316 (?&name) as well as (?P>name).
2318 (c) A backreference to a named group can now be defined as \k<name> or
2319 \k'name' as well as (?P=name). The new forms, as well as being in Perl
2320 5.10, are also .NET compatible.
2322 (d) A conditional reference to a named group can now use the syntax
2323 (?(<name>) or (?('name') as well as (?(name).
2325 (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
2326 groups (named and numbered) that are never evaluated inline, but can be
2327 called as "subroutines" from elsewhere. In effect, the DEFINE condition
2328 is always false. There may be only one alternative in such a group.
2330 (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
2331 as the simple (?(R). The condition is true only if the most recent
2332 recursion is that of the given number or name. It does not search out
2333 through the entire recursion stack.
2335 (g) The escape \gN or \g{N} has been added, where N is a positive or
2336 negative number, specifying an absolute or relative reference.
2338 35. Tidied to get rid of some further signed/unsigned compiler warnings and
2339 some "unreachable code" warnings.
2341 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
2342 things, this adds five new scripts.
2344 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
2345 There were also incompatibilities regarding the handling of \Q..\E inside
2346 character classes, for example with patterns like [\Qa\E-\Qz\E] where the
2347 hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
2349 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
2350 matches an empty string, and forcibly breaks the loop. There were bugs in
2351 this code in non-simple cases. For a pattern such as ^(a()*)* matched
2352 against aaaa the result was just "a" rather than "aaaa", for example. Two
2353 separate and independent bugs (that affected different cases) have been
2354 fixed.
2356 39. Refactored the code to abolish the use of different opcodes for small
2357 capturing bracket numbers. This is a tidy that I avoided doing when I
2358 removed the limit on the number of capturing brackets for 3.5 back in 2001.
2359 The new approach is not only tidier, it makes it possible to reduce the
2360 memory needed to fix the previous bug (38).
2362 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
2363 sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
2364 processing dot, circumflex, or dollar metacharacters, or #-comments in /x
2365 mode.
2367 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
2368 report.
2370 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
2371 copy construction and assignment in the C++ wrapper.
2373 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
2374 couple of bugs that could have given wrong results in the "--newline=crlf"
2375 case.
2377 44. Added a number of casts and did some reorganization of signed/unsigned int
2378 variables following suggestions from Dair Grant. Also renamed the variable
2379 "this" as "item" because it is a C++ keyword.
2381 45. Arranged for dftables to add
2383 #include "pcre_internal.h"
2385 to pcre_chartables.c because without it, gcc 4.x may remove the array
2386 definition from the final binary if PCRE is built into a static library and
2387 dead code stripping is activated.
2389 46. For an unanchored pattern, if a match attempt fails at the start of a
2390 newline sequence, and the newline setting is CRLF or ANY, and the next two
2391 characters are CRLF, advance by two characters instead of one.
2394 Version 6.7 04-Jul-06
2395 ---------------------
2397 1. In order to handle tests when input lines are enormously long, pcretest has
2398 been re-factored so that it automatically extends its buffers when
2399 necessary. The code is crude, but this _is_ just a test program. The
2400 default size has been increased from 32K to 50K.
2402 2. The code in pcre_study() was using the value of the re argument before
2403 testing it for NULL. (Of course, in any sensible call of the function, it
2404 won't be NULL.)
2406 3. The memmove() emulation function in pcre_internal.h, which is used on
2407 systems that lack both memmove() and bcopy() - that is, hardly ever -
2408 was missing a "static" storage class specifier.
2410 4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
2411 containing an extended class (one that cannot be represented by a bitmap
2412 because it contains high-valued characters or Unicode property items, e.g.
2413 [\pZ]). Almost always one would set UTF-8 mode when processing such a
2414 pattern, but PCRE should not loop if you do not (it no longer does).
2415 [Detail: two cases were found: (a) a repeated subpattern containing an
2416 extended class; (b) a recursive reference to a subpattern that followed a
2417 previous extended class. It wasn't skipping over the extended class
2418 correctly when UTF-8 mode was not set.]
2420 5. A negated single-character class was not being recognized as fixed-length
2421 in lookbehind assertions such as (?<=[^f]), leading to an incorrect
2422 compile error "lookbehind assertion is not fixed length".
2424 6. The RunPerlTest auxiliary script was showing an unexpected difference
2425 between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
2426 write a Perl script that can interpret lines of an input file either as
2427 byte characters or as UTF-8, which is what "perltest" was being required to
2428 do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
2429 can't do is switch easily at run time between having the "use utf8;" pragma
2430 or not. In the end, I fudged it by using the RunPerlTest script to insert
2431 "use utf8;" explicitly for the UTF-8 tests.
2433 7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
2434 the end of the subject string, contrary to the documentation and to what
2435 Perl does. This was true of both matching functions. Now it matches only at
2436 the start of the subject and immediately after *internal* newlines.
2438 8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
2439 a pointer to an int instead of a pointer to an unsigned long int. This
2440 caused problems on 64-bit systems.
2442 9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
2443 instance of the 'standard' template library not being so standard".
2445 10. There was no check on the number of named subpatterns nor the maximum
2446 length of a subpattern name. The product of these values is used to compute
2447 the size of the memory block for a compiled pattern. By supplying a very
2448 long subpattern name and a large number of named subpatterns, the size
2449 computation could be caused to overflow. This is now prevented by limiting
2450 the length of names to 32 characters, and the number of named subpatterns
2451 to 10,000.
2453 11. Subpatterns that are repeated with specific counts have to be replicated in
2454 the compiled pattern. The size of memory for this was computed from the
2455 length of the subpattern and the repeat count. The latter is limited to
2456 65535, but there was no limit on the former, meaning that integer overflow
2457 could in principle occur. The compiled length of a repeated subpattern is
2458 now limited to 30,000 bytes in order to prevent this.
2460 12. Added the optional facility to have named substrings with the same name.
2462 13. Added the ability to use a named substring as a condition, using the
2463 Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
2464 are numbers (not recommended). Forward references are permitted.
2466 14. Added forward references in named backreferences (if you see what I mean).
2468 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
2469 pattern could run off the end of the subject. For example, the pattern
2470 "(?s)(.{1,5})"8 did this with the subject "ab".
2472 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
2473 PCRE_CASELESS was set when matching characters that were quantified with ?
2474 or *.
2476 17. A character class other than a single negated character that had a minimum
2477 but no maximum quantifier - for example [ab]{6,} - was not handled
2478 correctly by pce_dfa_exec(). It would match only one character.
2480 18. A valid (though odd) pattern that looked like a POSIX character
2481 class but used an invalid character after [ (for example [[,abc,]]) caused
2482 pcre_compile() to give the error "Failed: internal error: code overflow" or
2483 in some cases to crash with a glibc free() error. This could even happen if
2484 the pattern terminated after [[ but there just happened to be a sequence of
2485 letters, a binary zero, and a closing ] in the memory that followed.
2487 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
2488 over the years. Originally (before any Unicode support), just the bottom 8
2489 bits were taken. Thus, for example, \500 really meant \100. Nowadays the
2490 output from "man perlunicode" includes this:
2492 The regular expression compiler produces polymorphic opcodes. That
2493 is, the pattern adapts to the data and automatically switches to
2494 the Unicode character scheme when presented with Unicode data--or
2495 instead uses a traditional byte scheme when presented with byte
2496 data.
2498 Sadly, a wide octal escape does not cause a switch, and in a string with
2499 no other multibyte characters, these octal escapes are treated as before.
2500 Thus, in Perl, the pattern /\500/ actually matches \100 but the pattern
2501 /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
2502 Unicode string.
2504 I have not perpetrated such confusion in PCRE. Up till now, it took just
2505 the bottom 8 bits, as in old Perl. I have now made octal escapes with
2506 values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
2507 translate to the appropriate multibyte character.
2509 29. Applied some refactoring to reduce the number of warnings from Microsoft
2510 and Borland compilers. This has included removing the fudge introduced
2511 seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
2512 a warning about an unused variable.
2514 21. PCRE has not included VT (character 0x0b) in the set of whitespace
2515 characters since release 4.0, because Perl (from release 5.004) does not.
2516 [Or at least, is documented not to: some releases seem to be in conflict
2517 with the documentation.] However, when a pattern was studied with
2518 pcre_study() and all its branches started with \s, PCRE still included VT
2519 as a possible starting character. Of course, this did no harm; it just
2520 caused an unnecessary match attempt.
2522 22. Removed a now-redundant internal flag bit that recorded the fact that case
2523 dependency changed within the pattern. This was once needed for "required
2524 byte" processing, but is no longer used. This recovers a now-scarce options
2525 bit. Also moved the least significant internal flag bit to the most-
2526 significant bit of the word, which was not previously used (hangover from
2527 the days when it was an int rather than a uint) to free up another bit for
2528 the future.
2530 23. Added support for CRLF line endings as well as CR and LF. As well as the
2531 default being selectable at build time, it can now be changed at runtime
2532 via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
2533 specify that it is scanning data with non-default line endings.
2535 24. Changed the definition of CXXLINK to make it agree with the definition of
2536 LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
2538 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
2539 recursions. This makes a big different to stack usage for some patterns.
2541 26. If a subpattern containing a named recursion or subroutine reference such
2542 as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
2543 the space required for the compiled pattern went wrong and gave too small a
2544 value. Depending on the environment, this could lead to "Failed: internal
2545 error: code overflow at offset 49" or "glibc detected double free or
2546 corruption" errors.
2548 27. Applied patches from Google (a) to support the new newline modes and (b) to
2549 advance over multibyte UTF-8 characters in GlobalReplace.
2551 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
2552 difference for some implementation of PCRE in some Windows version.
2554 29. Added some extra testing facilities to pcretest:
2556 \q<number> in a data line sets the "match limit" value
2557 \Q<number> in a data line sets the "match recursion limt" value
2558 -S <number> sets the stack size, where <number> is in megabytes
2560 The -S option isn't available for Windows.
2563 Version 6.6 06-Feb-06
2564 ---------------------
2566 1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
2567 in pcreposix.h. I have copied the definition from pcre.h.
2569 2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
2570 because pcre.h is no longer a built file.
2572 3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
2573 not normally included in the compiled code.
2576 Version 6.5 01-Feb-06
2577 ---------------------
2579 1. When using the partial match feature with pcre_dfa_exec(), it was not
2580 anchoring the second and subsequent partial matches at the new starting
2581 point. This could lead to incorrect results. For example, with the pattern
2582 /1234/, partially matching against "123" and then "a4" gave a match.
2584 2. Changes to pcregrep:
2586 (a) All non-match returns from pcre_exec() were being treated as failures
2587 to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
2588 error message is output. Some extra information is given for the
2590 probably the only errors that are likely to be caused by users (by
2591 specifying a regex that has nested indefinite repeats, for instance).
2592 If there are more than 20 of these errors, pcregrep is abandoned.
2594 (b) A binary zero was treated as data while matching, but terminated the
2595 output line if it was written out. This has been fixed: binary zeroes
2596 are now no different to any other data bytes.
2598 (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
2599 used to set a locale for matching. The --locale=xxxx long option has
2600 been added (no short equivalent) to specify a locale explicitly on the
2601 pcregrep command, overriding the environment variables.
2603 (d) When -B was used with -n, some line numbers in the output were one less
2604 than they should have been.
2606 (e) Added the -o (--only-matching) option.
2608 (f) If -A or -C was used with -c (count only), some lines of context were
2609 accidentally printed for the final match.
2611 (g) Added the -H (--with-filename) option.
2613 (h) The combination of options -rh failed to suppress file names for files
2614 that were found from directory arguments.
2616 (i) Added the -D (--devices) and -d (--directories) options.
2618 (j) Added the -F (--fixed-strings) option.
2620 (k) Allow "-" to be used as a file name for -f as well as for a data file.
2622 (l) Added the --colo(u)r option.
2624 (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
2625 is not present by default.
2627 3. A nasty bug was discovered in the handling of recursive patterns, that is,
2628 items such as (?R) or (?1), when the recursion could match a number of
2629 alternatives. If it matched one of the alternatives, but subsequently,
2630 outside the recursion, there was a failure, the code tried to back up into
2631 the recursion. However, because of the way PCRE is implemented, this is not
2632 possible, and the result was an incorrect result from the match.
2634 In order to prevent this happening, the specification of recursion has
2635 been changed so that all such subpatterns are automatically treated as
2636 atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
2638 4. I had overlooked the fact that, in some locales, there are characters for
2639 which isalpha() is true but neither isupper() nor islower() are true. In
2640 the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
2641 and ordfeminine) are like this. This affected the treatment of \w and \W
2642 when they appeared in character classes, but not when they appeared outside
2643 a character class. The bit map for "word" characters is now created
2644 separately from the results of isalnum() instead of just taking it from the
2645 upper, lower, and digit maps. (Plus the underscore character, of course.)
2647 5. The above bug also affected the handling of POSIX character classes such as
2648 [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
2649 permanent tables. Instead, the bit maps for such a class were previously
2650 created as the appropriate unions of the upper, lower, and digit bitmaps.
2651 Now they are created by subtraction from the [[:word:]] class, which has
2652 its own bitmap.
2654 6. The [[:blank:]] character class matches horizontal, but not vertical space.
2655 It is created by subtracting the vertical space characters (\x09, \x0a,
2656 \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
2657 subtraction was done in the overall bitmap for a character class, meaning
2658 that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
2659 be recognized. This bug has been fixed.
2661 7. Patches from the folks at Google:
2663 (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
2664 real life, but is still worth protecting against".
2666 (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
2667 regular expressions".
2669 (c) pcre_scanner.cc: avoid use of std::count() because not all systems
2670 have it.
2672 (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
2673 "configure" and the latter not, in order to fix a problem somebody had
2674 with compiling the Arg class on HP-UX.
2676 (e) Improve the error-handling of the C++ wrapper a little bit.
2678 (f) New tests for checking recursion limiting.
2680 8. The pcre_memmove() function, which is used only if the environment does not
2681 have a standard memmove() function (and is therefore rarely compiled),
2682 contained two bugs: (a) use of int instead of size_t, and (b) it was not
2683 returning a result (though PCRE never actually uses the result).
2685 9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
2686 large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
2687 returned instead of calling malloc() with an overflowing number that would
2688 most likely cause subsequent chaos.
2690 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
2692 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
2693 with this option is matched, the nmatch and pmatch options of regexec() are
2694 ignored.
2696 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
2697 provided in case anyone wants to the the POSIX interface with UTF-8
2698 strings.
2700 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
2701 C++ linking (needed for some HP-UX environments).
2703 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
2704 (unused parameter) and in the pcre_printint() function (omitted "default"
2705 switch label when the default is to do nothing).
2707 15. Added some code to make it possible, when PCRE is compiled as a C++
2708 library, to replace subject pointers for pcre_exec() with a smart pointer
2709 class, thus making it possible to process discontinuous strings.
2711 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
2712 much the same function. They were added by different people who were trying
2713 to make PCRE easy to compile on non-Unix systems. It has been suggested
2714 that PCRE_EXPORT be abolished now that there is more automatic apparatus
2715 for compiling on Windows systems. I have therefore replaced it with
2716 PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
2717 defaults to "extern" for C or "extern C" for C++, which works fine on
2718 Unix-like systems. It is now possible to override the value of PCRE_DATA_
2719 SCOPE with something explicit in config.h. In addition:
2721 (a) pcreposix.h still had just "extern" instead of either of these macros;
2722 I have replaced it with PCRE_DATA_SCOPE.
2724 (b) Functions such as _pcre_xclass(), which are internal to the library,
2725 but external in the C sense, all had PCRE_EXPORT in their definitions.
2726 This is apparently wrong for the Windows case, so I have removed it.
2727 (It makes no difference on Unix-like systems.)
2729 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
2730 of recursive calls to match(). This is different to MATCH_LIMIT because
2731 that limits the total number of calls to match(), not all of which increase
2732 the depth of recursion. Limiting the recursion depth limits the amount of
2733 stack (or heap if NO_RECURSE is set) that is used. The default can be set
2734 when PCRE is compiled, and changed at run time. A patch from Google adds
2735 this functionality to the C++ interface.
2737 18. Changes to the handling of Unicode character properties:
2739 (a) Updated the table to Unicode 4.1.0.
2741 (b) Recognize characters that are not in the table as "Cn" (undefined).
2743 (c) I revised the way the table is implemented to a much improved format
2744 which includes recognition of ranges. It now supports the ranges that
2745 are defined in UnicodeData.txt, and it also amalgamates other
2746 characters into ranges. This has reduced the number of entries in the
2747 table from around 16,000 to around 3,000, thus reducing its size
2748 considerably. I realized I did not need to use a tree structure after
2749 all - a binary chop search is just as efficient. Having reduced the
2750 number of entries, I extended their size from 6 bytes to 8 bytes to
2751 allow for more data.
2753 (d) Added support for Unicode script names via properties such as \p{Han}.
2755 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
2756 matching that character.
2758 20. When matching a repeated Unicode property with a minimum greater than zero,
2759 (for example \pL{2,}), PCRE could look past the end of the subject if it
2760 reached it while seeking the minimum number of characters. This could
2761 happen only if some of the characters were more than one byte long, because
2762 there is a check for at least the minimum number of bytes.
2764 21. Refactored the implementation of \p and \P so as to be more general, to
2765 allow for more different types of property in future. This has changed the
2766 compiled form incompatibly. Anybody with saved compiled patterns that use
2767 \p or \P will have to recompile them.
2769 22. Added "Any" and "L&" to the supported property types.
2771 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
2772 but give a compile time error if the value is greater than 0xff.
2774 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
2775 accidentally not being installed or uninstalled.
2777 25. The pcre.h file was built from pcre.h.in, but the only changes that were
2778 made were to insert the current release number. This seemed silly, because
2779 it made things harder for people building PCRE on systems that don't run
2780 "configure". I have turned pcre.h into a distributed file, no longer built
2781 by "configure", with the version identification directly included. There is
2782 no longer a pcre.h.in file.
2784 However, this change necessitated a change to the pcre-config script as
2785 well. It is built from pcre-config.in, and one of the substitutions was the
2786 release number. I have updated configure.ac so that ./configure now finds
2787 the release number by grepping pcre.h.
2789 26. Added the ability to run the tests under valgrind.
2792 Version 6.4 05-Sep-05
2793 ---------------------
2795 1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
2796 "--" to be printed when multiple files were scanned, even when none of the
2797 -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
2798 consider it to be a bug, and have restored the previous behaviour.
2800 2. A couple of code tidies to get rid of compiler warnings.
2802 3. The pcretest program used to cheat by referring to symbols in the library
2803 whose names begin with _pcre_. These are internal symbols that are not
2804 really supposed to be visible externally, and in some environments it is
2805 possible to suppress them. The cheating is now confined to including
2806 certain files from the library's source, which is a bit cleaner.
2808 4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
2809 file's purpose clearer.
2811 5. Reorganized pcre_ucp_findchar().
2814 Version 6.3 15-Aug-05
2815 ---------------------
2817 1. The file libpcre.pc.in did not have general read permission in the tarball.
2819 2. There were some problems when building without C++ support:
2821 (a) If C++ support was not built, "make install" and "make test" still
2822 tried to test it.
2824 (b) There were problems when the value of CXX was explicitly set. Some
2825 changes have been made to try to fix these, and ...
2827 (c) --disable-cpp can now be used to explicitly disable C++ support.
2829 (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
2830 backslash in a target when C++ was disabled. This confuses some
2831 versions of "make", apparently. Using an intermediate variable solves
2832 this. (Same for CPP_LOBJ.)
2834 3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
2835 (non-Windows) now includes $(CFLAGS) because these flags are sometimes
2836 necessary on certain architectures.
2838 4. Added a setting of -export-symbols-regex to the link command to remove
2839 those symbols that are exported in the C sense, but actually are local
2840 within the library, and not documented. Their names all begin with
2841 "_pcre_". This is not a perfect job, because (a) we have to except some
2842 symbols that pcretest ("illegally") uses, and (b) the facility isn't always
2843 available (and never for static libraries). I have made a note to try to
2844 find a way round (a) in the future.
2847 Version 6.2 01-Aug-05
2848 ---------------------
2850 1. There was no test for integer overflow of quantifier values. A construction
2851 such as {1111111111111111} would give undefined results. What is worse, if
2852 a minimum quantifier for a parenthesized subpattern overflowed and became
2853 negative, the calculation of the memory size went wrong. This could have
2854 led to memory overwriting.
2856 2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
2858 3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
2859 operating environments where this matters.
2861 4. Applied Giuseppe Maxia's patch to add additional features for controlling
2862 PCRE options from within the C++ wrapper.
2864 5. Named capturing subpatterns were not being correctly counted when a pattern
2865 was compiled. This caused two problems: (a) If there were more than 100
2866 such subpatterns, the calculation of the memory needed for the whole
2867 compiled pattern went wrong, leading to an overflow error. (b) Numerical
2868 back references of the form \12, where the number was greater than 9, were
2869 not recognized as back references, even though there were sufficient
2870 previous subpatterns.
2872 6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
2873 versions of gcc, e.g. 2.95.4.
2876 Version 6.1 21-Jun-05
2877 ---------------------
2879 1. There was one reference to the variable "posix" in pcretest.c that was not
2880 surrounded by "#if !defined NOPOSIX".
2882 2. Make it possible to compile pcretest without DFA support, UTF8 support, or
2883 the cross-check on the old pcre_info() function, for the benefit of the
2884 cut-down version of PCRE that is currently imported into Exim.
2886 3. A (silly) pattern starting with (?i)(?-i) caused an internal space
2887 allocation error. I've done the easy fix, which wastes 2 bytes for sensible
2888 patterns that start (?i) but I don't think that matters. The use of (?i) is
2889 just an example; this all applies to the other options as well.
2891 4. Since libtool seems to echo the compile commands it is issuing, the output
2892 from "make" can be reduced a bit by putting "@" in front of each libtool
2893 compile command.
2895 5. Patch from the folks at Google for configure.in to be a bit more thorough
2896 in checking for a suitable C++ installation before trying to compile the
2897 C++ stuff. This should fix a reported problem when a compiler was present,
2898 but no suitable headers.
2900 6. The man pages all had just "PCRE" as their title. I have changed them to
2901 be the relevant file name. I have also arranged that these names are
2902 retained in the file doc/pcre.txt, which is a concatenation in text format
2903 of all the man pages except the little individual ones for each function.
2905 7. The NON-UNIX-USE file had not been updated for the different set of source
2906 files that come with release 6. I also added a few comments about the C++
2907 wrapper.
2910 Version 6.0 07-Jun-05
2911 ---------------------
2913 1. Some minor internal re-organization to help with my DFA experiments.
2915 2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
2916 didn't matter for the library itself when fully configured, but did matter
2917 when compiling without UCP support, or within Exim, where the ucp files are
2918 not imported.
2920 3. Refactoring of the library code to split up the various functions into
2921 different source modules. The addition of the new DFA matching code (see
2922 below) to a single monolithic source would have made it really too
2923 unwieldy, quite apart from causing all the code to be include in a
2924 statically linked application, when only some functions are used. This is
2925 relevant even without the DFA addition now that patterns can be compiled in
2926 one application and matched in another.
2928 The downside of splitting up is that there have to be some external
2929 functions and data tables that are used internally in different modules of
2930 the library but which are not part of the API. These have all had their
2931 names changed to start with "_pcre_" so that they are unlikely to clash
2932 with other external names.
2934 4. Added an alternate matching function, pcre_dfa_exec(), which matches using
2935 a different (DFA) algorithm. Although it is slower than the original
2936 function, it does have some advantages for certain types of matching
2937 problem.
2939 5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
2940 including restarting after a partial match.
2942 6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
2943 defined when compiling for Windows was sent to me. I have put it into the
2944 code, though I have no means of testing or verifying it.
2946 7. Added the pcre_refcount() auxiliary function.
2948 8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
2949 match before or at the first newline in the subject string. In pcretest,
2950 the /f option on a pattern can be used to set this.
2952 9. A repeated \w when used in UTF-8 mode with characters greater than 256
2953 would behave wrongly. This has been present in PCRE since release 4.0.
2955 10. A number of changes to the pcregrep command:
2957 (a) Refactored how -x works; insert ^(...)$ instead of setting
2958 PCRE_ANCHORED and checking the length, in preparation for adding
2959 something similar for -w.
2961 (b) Added the -w (match as a word) option.
2963 (c) Refactored the way lines are read and buffered so as to have more
2964 than one at a time available.
2966 (d) Implemented a pcregrep test script.
2968 (e) Added the -M (multiline match) option. This allows patterns to match
2969 over several lines of the subject. The buffering ensures that at least
2970 8K, or the rest of the document (whichever is the shorter) is available
2971 for matching (and similarly the previous 8K for lookbehind assertions).
2973 (f) Changed the --help output so that it now says
2975 -w, --word-regex(p)
2977 instead of two lines, one with "regex" and the other with "regexp"
2978 because that confused at least one person since the short forms are the
2979 same. (This required a bit of code, as the output is generated
2980 automatically from a table. It wasn't just a text change.)
2982 (g) -- can be used to terminate pcregrep options if the next thing isn't an
2983 option but starts with a hyphen. Could be a pattern or a path name
2984 starting with a hyphen, for instance.
2986 (h) "-" can be given as a file name to represent stdin.
2988 (i) When file names are being printed, "(standard input)" is used for
2989 the standard input, for compatibility with GNU grep. Previously
2990 "<stdin>" was used.
2992 (j) The option --label=xxx can be used to supply a name to be used for
2993 stdin when file names are being printed. There is no short form.
2995 (k) Re-factored the options decoding logic because we are going to add
2996 two more options that take data. Such options can now be given in four
2997 different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
2999 (l) Added the -A, -B, and -C options for requesting that lines of context
3000 around matches be printed.
3002 (m) Added the -L option to print the names of files that do not contain
3003 any matching lines, that is, the complement of -l.
3005 (n) The return code is 2 if any file cannot be opened, but pcregrep does
3006 continue to scan other files.
3008 (o) The -s option was incorrectly implemented. For compatibility with other
3009 greps, it now suppresses the error message for a non-existent or non-
3010 accessible file (but not the return code). There is a new option called
3011 -q that suppresses the output of matching lines, which was what -s was
3012 previously doing.
3014 (p) Added --include and --exclude options to specify files for inclusion
3015 and exclusion when recursing.
3017 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
3018 Hopefully, it now does.
3020 12. Missing cast in pcre_study().
3022 13. Added an "uninstall" target to the makefile.
3024 14. Replaced "extern" in the function prototypes in Makefile.in with
3025 "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
3026 world, but is set differently for Windows.
3028 15. Added a second compiling function called pcre_compile2(). The only
3029 difference is that it has an extra argument, which is a pointer to an
3030 integer error code. When there is a compile-time failure, this is set
3031 non-zero, in addition to the error test pointer being set to point to an
3032 error message. The new argument may be NULL if no error number is required
3033 (but then you may as well call pcre_compile(), which is now just a
3034 wrapper). This facility is provided because some applications need a
3035 numeric error indication, but it has also enabled me to tidy up the way
3036 compile-time errors are handled in the POSIX wrapper.
3038 16. Added VPATH=.libs to the makefile; this should help when building with one
3039 prefix path and installing with another. (Or so I'm told by someone who
3040 knows more about this stuff than I do.)
3042 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
3043 passes PCRE_DOTALL to the pcre_compile() function, making the "." character
3044 match everything, including newlines. This is not POSIX-compatible, but
3045 somebody wanted the feature. From pcretest it can be activated by using
3046 both the P and the s flags.
3048 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
3050 19. libpcre.pc was being incorrectly installed as executable.
3052 20. A couple of places in pcretest check for end-of-line by looking for '\n';
3053 it now also looks for '\r' so that it will work unmodified on Windows.
3055 21. Added Google's contributed C++ wrapper to the distribution.
3057 22. Added some untidy missing memory free() calls in pcretest, to keep
3058 Electric Fence happy when testing.
3062 Version 5.0 13-Sep-04
3063 ---------------------
3065 1. Internal change: literal characters are no longer packed up into items
3066 containing multiple characters in a single byte-string. Each character
3067 is now matched using a separate opcode. However, there may be more than one
3068 byte in the character in UTF-8 mode.
3070 2. The pcre_callout_block structure has two new fields: pattern_position and
3071 next_item_length. These contain the offset in the pattern to the next match
3072 item, and its length, respectively.
3074 3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
3075 insertion of callouts before each pattern item. Added the /C option to
3076 pcretest to make use of this.
3078 4. On the advice of a Windows user, the lines
3080 #if defined(_WIN32) || defined(WIN32)
3081 _setmode( _fileno( stdout ), 0x8000 );
3082 #endif /* defined(_WIN32) || defined(WIN32) */
3084 have been added to the source of pcretest. This apparently does useful
3085 magic in relation to line terminators.
3087 5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
3088 for the benefit of those environments where the "b" makes a difference.
3090 6. The icc compiler has the same options as gcc, but "configure" doesn't seem
3091 to know about it. I have put a hack into configure.in that adds in code
3092 to set GCC=yes if CC=icc. This seems to end up at a point in the
3093 generated configure script that is early enough to affect the setting of
3094 compiler options, which is what is needed, but I have no means of testing
3095 whether it really works. (The user who reported this had patched the
3096 generated configure script, which of course I cannot do.)
3098 LATER: After change 22 below (new libtool files), the configure script
3099 seems to know about icc (and also ecc). Therefore, I have commented out
3100 this hack in configure.in.
3102 7. Added support for pkg-config (2 patches were sent in).
3104 8. Negated POSIX character classes that used a combination of internal tables
3105 were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
3106 [[:^ascii]]. Typically, they would match almost any characters. The other
3107 POSIX classes were not broken in this way.
3109 9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
3110 to find the match, as PCRE was deluded into thinking that the match had to
3111 start at the start point or following a newline. The same bug applied to
3112 patterns with negative forward assertions or any backward assertions
3113 preceding ".*" at the start, unless the pattern required a fixed first
3114 character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
3116 10. In UTF-8 mode, when moving forwards in the subject after a failed match
3117 starting at the last subject character, bytes beyond the end of the subject
3118 string were read.
3120 11. Renamed the variable "class" as "classbits" to make life easier for C++
3121 users. (Previously there was a macro definition, but it apparently wasn't
3122 enough.)
3124 12. Added the new field "tables" to the extra data so that tables can be passed
3125 in at exec time, or the internal tables can be re-selected. This allows
3126 a compiled regex to be saved and re-used at a later time by a different
3127 program that might have everything at different addresses.
3129 13. Modified the pcre-config script so that, when run on Solaris, it shows a
3130 -R library as well as a -L library.
3132 14. The debugging options of pcretest (-d on the command line or D on a
3133 pattern) showed incorrect output for anything following an extended class
3134 that contained multibyte characters and which was followed by a quantifier.
3136 15. Added optional support for general category Unicode character properties
3137 via the \p, \P, and \X escapes. Unicode property support implies UTF-8
3138 support. It adds about 90K to the size of the library. The meanings of the
3139 inbuilt class escapes such as \d and \s have NOT been changed.
3141 16. Updated pcredemo.c to include calls to free() to release the memory for the
3142 compiled pattern.
3144 17. The generated file chartables.c was being created in the source directory
3145 instead of in the building directory. This caused the build to fail if the
3146 source directory was different from the building directory, and was
3147 read-only.
3149 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
3150 file. No doubt somebody will tell me if they don't make sense... Also added
3151 Dan Mooney's comments about building on OpenVMS.
3153 19. Added support for partial matching via the PCRE_PARTIAL option for
3154 pcre_exec() and the \P data escape in pcretest.
3156 20. Extended pcretest with 3 new pattern features:
3158 (i) A pattern option of the form ">rest-of-line" causes pcretest to
3159 write the compiled pattern to the file whose name is "rest-of-line".
3160 This is a straight binary dump of the data, with the saved pointer to
3161 the character tables forced to be NULL. The study data, if any, is
3162 written too. After writing, pcretest reads a new pattern.
3164 (ii) If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
3165 compiled pattern from the given file. There must not be any
3166 occurrences of "<" in the file name (pretty unlikely); if there are,
3167 pcretest will instead treat the initial "<" as a pattern delimiter.
3168 After reading in the pattern, pcretest goes on to read data lines as
3169 usual.
3171 (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
3172 and 16-bit fields in a compiled pattern, to simulate a pattern that
3173 was compiled on a host of opposite endianness.
3175 21. The pcre-exec() function can now cope with patterns that were compiled on
3176 hosts of opposite endianness, with this restriction:
3178 As for any compiled expression that is saved and used later, the tables
3179 pointer field cannot be preserved; the extra_data field in the arguments
3180 to pcre_exec() should be used to pass in a tables address if a value
3181 other than the default internal tables were used at compile time.
3183 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
3184 now diagnosed as an error. Previously, most of the time, a negative number
3185 would have been treated as zero, but if in addition "ovector" was passed as
3186 NULL, a crash could occur.
3188 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
3189 new versions from the libtool 1.5 distribution (the last one is a copy of
3190 a file called libtool.m4). This seems to have fixed the need to patch
3191 "configure" to support Darwin 1.3 (which I used to do). However, I still
3192 had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
3193 workstation).
3195 24. Changed the PCRE licence to be the more standard "BSD" licence.
3198 Version 4.5 01-Dec-03
3199 ---------------------
3201 1. There has been some re-arrangement of the code for the match() function so
3202 that it can be compiled in a version that does not call itself recursively.
3203 Instead, it keeps those local variables that need separate instances for
3204 each "recursion" in a frame on the heap, and gets/frees frames whenever it
3205 needs to "recurse". Keeping track of where control must go is done by means
3206 of setjmp/longjmp. The whole thing is implemented by a set of macros that
3207 hide most of the details from the main code, and operates only if
3208 NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
3209 "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
3210 operating.
3212 To make it easier for callers to provide specially tailored get/free
3213 functions for this usage, two new functions, pcre_stack_malloc, and
3214 pcre_stack_free, are used. They are always called in strict stacking order,
3215 and the size of block requested is always the same.
3217 The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
3218 PCRE has been compiled to use the stack or the heap for recursion. The
3219 -C option of pcretest uses this to show which version is compiled.
3221 A new data escape \S, is added to pcretest; it causes the amounts of store
3222 obtained and freed by both kinds of malloc/free at match time to be added
3223 to the output.
3225 2. Changed the locale test to use "fr_FR" instead of "fr" because that's
3226 what's available on my current Linux desktop machine.
3228 3. When matching a UTF-8 string, the test for a valid string at the start has
3229 been extended. If start_offset is not zero, PCRE now checks that it points
3230 to a byte that is the start of a UTF-8 character. If not, it returns
3231 PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
3232 this is necessary because there may be backward assertions in the pattern.
3233 When matching the same subject several times, it may save resources to use
3234 PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
3236 4. The code for checking the validity of UTF-8 strings has been tightened so
3237 that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
3238 containing "overlong sequences".
3240 5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
3241 I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
3242 should have been "&", but it just so happened that all the cases this let
3243 through by mistake were picked up later in the function.
3245 6. I had used a variable called "isblank" - this is a C99 function, causing
3246 some compilers to warn. To avoid this, I renamed it (as "blankclass").
3248 7. Cosmetic: (a) only output another newline at the end of pcretest if it is
3249 prompting; (b) run "./pcretest /dev/null" at the start of the test script
3250 so the version is shown; (c) stop "make test" echoing "./RunTest".
3252 8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
3254 9. The prototype for memmove() for systems that don't have it was using
3255 size_t, but the inclusion of the header that defines size_t was later. I've
3256 moved the #includes for the C headers earlier to avoid this.
3258 10. Added some adjustments to the code to make it easier to compiler on certain
3259 special systems:
3261 (a) Some "const" qualifiers were missing.
3262 (b) Added the macro EXPORT before all exported functions; by default this
3263 is defined to be empty.
3264 (c) Changed the dftables auxiliary program (that builds chartables.c) so
3265 that it reads its output file name as an argument instead of writing
3266 to the standard output and assuming this can be redirected.
3268 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
3269 class containing characters with values greater than 255, PCRE compilation
3270 went into a loop.
3272 12. A recursive reference to a subpattern that was within another subpattern
3273 that had a minimum quantifier of zero caused PCRE to crash. For example,
3274 (x(y(?2))z)? provoked this bug with a subject that got as far as the
3275 recursion. If the recursively-called subpattern itself had a zero repeat,
3276 that was OK.
3278 13. In pcretest, the buffer for reading a data line was set at 30K, but the
3279 buffer into which it was copied (for escape processing) was still set at
3280 1024, so long lines caused crashes.
3282 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
3283 "internal error: code overflow...". This applied to any character class
3284 that was followed by a possessive quantifier.
3286 15. Modified the Makefile to add libpcre.la as a prerequisite for
3287 libpcreposix.la because I was told this is needed for a parallel build to
3288 work.
3290 16. If a pattern that contained .* following optional items at the start was
3291 studied, the wrong optimizing data was generated, leading to matching
3292 errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
3293 matching string must start with a or b or c. The correct conclusion for
3294 this pattern is that a match can start with any character.
3297 Version 4.4 13-Aug-03
3298 ---------------------
3300 1. In UTF-8 mode, a character class containing characters with values between
3301 127 and 255 was not handled correctly if the compiled pattern was studied.
3302 In fixing this, I have also improved the studying algorithm for such
3303 classes (slightly).
3305 2. Three internal functions had redundant arguments passed to them. Removal
3306 might give a very teeny performance improvement.
3308 3. Documentation bug: the value of the capture_top field in a callout is *one
3309 more than* the number of the hightest numbered captured substring.
3311 4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
3312 in incorrectly linking with a previously installed version. They now link
3313 explicitly with libpcre.la.
3315 5. configure.in no longer needs to recognize Cygwin specially.
3317 6. A problem in pcre.in for Windows platforms is fixed.
3319 7. If a pattern was successfully studied, and the -d (or /D) flag was given to
3320 pcretest, it used to include the size of the study block as part of its
3321 output. Unfortunately, the structure contains a field that has a different
3322 size on different hardware architectures. This meant that the tests that
3323 showed this size failed. As the block is currently always of a fixed size,
3324 this information isn't actually particularly useful in pcretest output, so
3325 I have just removed it.
3327 8. Three pre-processor statements accidentally did not start in column 1.
3328 Sadly, there are *still* compilers around that complain, even though
3329 standard C has not required this for well over a decade. Sigh.
3331 9. In pcretest, the code for checking callouts passed small integers in the
3332 callout_data field, which is a void * field. However, some picky compilers
3333 complained about the casts involved for this on 64-bit systems. Now
3334 pcretest passes the address of the small integer instead, which should get
3335 rid of the warnings.
3337 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
3338 both compile and run time, and gives an error if an invalid UTF-8 sequence
3339 is found. There is a option for disabling this check in cases where the
3340 string is known to be correct and/or the maximum performance is wanted.
3342 11. In response to a bug report, I changed one line in Makefile.in from
3344 -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
3345 to
3346 -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
3348 to look similar to other lines, but I have no way of telling whether this
3349 is the right thing to do, as I do not use Windows. No doubt I'll get told
3350 if it's wrong...
3353 Version 4.3 21-May-03
3354 ---------------------
3356 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
3357 Makefile.
3359 2. Some refactoring to improve the quality of the code:
3361 (i) The utf8_table... variables are now declared "const".
3363 (ii) The code for \cx, which used the "case flipping" table to upper case
3364 lower case letters, now just substracts 32. This is ASCII-specific,
3365 but the whole concept of \cx is ASCII-specific, so it seems
3366 reasonable.
3368 (iii) PCRE was using its character types table to recognize decimal and
3369 hexadecimal digits in the pattern. This is silly, because it handles
3370 only 0-9, a-f, and A-F, but the character types table is locale-
3371 specific, which means strange things might happen. A private
3372 table is now used for this - though it costs 256 bytes, a table is
3373 much faster than multiple explicit tests. Of course, the standard
3374 character types table is still used for matching digits in subject
3375 strings against \d.
3377 (iv) Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
3378 ending in _t are). So I've renamed it as ESC_tee.
3380 3. The first argument for regexec() in the POSIX wrapper should have been
3381 defined as "const".
3383 4. Changed pcretest to use malloc() for its buffers so that they can be
3384 Electric Fenced for debugging.
3386 5. There were several places in the code where, in UTF-8 mode, PCRE would try
3387 to read one or more bytes before the start of the subject string. Often this
3388 had no effect on PCRE's behaviour, but in some circumstances it could
3389 provoke a segmentation fault.
3391 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
3392 to try to read one or more bytes before the start of the subject string.
3394 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
3395 UTF-8 support could misbehave in various ways if the subject string
3396 contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
3397 area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
3398 back over UTF-8 characters.)
3401 Version 4.2 14-Apr-03
3402 ---------------------
3404 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
3406 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
3407 [ON_WINDOWS]: new variable, "#" on non-Windows platforms
3408 [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
3409 [WIN_PREFIX]: new variable, "cyg" for Cygwin
3410 * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
3412 Note: automatic setting of the BUILD variables is not yet working
3413 set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
3414 compile-time but not at link-time
3415 [LINK]: use for linking executables only
3416 make different versions for Windows and non-Windows
3417 [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
3418 libraries
3419 [LINK_FOR_BUILD]: new variable
3420 [OBJEXT]: use throughout
3421 [EXEEXT]: use throughout
3422 <winshared>: new target
3423 <wininstall>: new target
3424 <dftables.o>: use native compiler
3425 <dftables>: use native linker
3426 <install>: handle Windows platform correctly
3427 <clean>: ditto
3428 <check>: ditto
3429 copy DLL to top builddir before testing
3431 As part of these changes, -no-undefined was removed again. This was reported
3432 to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
3433 in any case.
3435 3. Some tidies to get rid of compiler warnings:
3437 . In the match_data structure, match_limit was an unsigned long int, whereas
3438 match_call_count was an int. I've made them both unsigned long ints.
3440 . In pcretest the fact that a const uschar * doesn't automatically cast to
3441 a void * provoked a warning.
3443 . Turning on some more compiler warnings threw up some "shadow" variables
3444 and a few more missing casts.
3446 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
3447 option, a class that contained a single character with a value between 128
3448 and 255 (e.g. /[\xFF]/) caused PCRE to crash.
3450 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
3451 option, a class that contained several characters, but with at least one
3452 whose value was between 128 and 255 caused PCRE to crash.
3455 Version 4.1 12-Mar-03
3456 ---------------------
3458 1. Compiling with gcc -pedantic found a couple of places where casts were
3459 needed, and a string in dftables.c that was longer than standard compilers are
3460 required to support.
3462 2. Compiling with Sun's compiler found a few more places where the code could
3463 be tidied up in order to avoid warnings.
3465 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
3466 first of these names is deprecated in the latest Autoconf in favour of the name
3467 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
3468 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
3469 analogy I have changed it to CFLAGS_FOR_BUILD.
3471 4. Added -no-undefined to the linking command in the Makefile, because this is
3472 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
3473 linking step for the pcreposix library.
3475 5. PCRE was failing to diagnose the case of two named groups with the same
3476 name.
3478 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
3479 literal character that is needed in the subject for a match, and scans along to
3480 ensure that it is present before embarking on the full matching process. This
3481 saves time in cases of nested unlimited repeats that are never going to match.
3482 Problem: the scan can take a lot of time if the subject is very long (e.g.
3483 megabytes), thus penalizing straightforward matches. It is now done only if the
3484 amount of subject to be scanned is less than 1000 bytes.
3486 7. A lesser problem with the same optimization is that it was recording the
3487 first character of an anchored pattern as "needed", thus provoking a search
3488 right along the subject, even when the first match of the pattern was going to
3489 fail. The "needed" character is now not set for anchored patterns, unless it
3490 follows something in the pattern that is of non-fixed length. Thus, it still
3491 fulfils its original purpose of finding quick non-matches in cases of nested
3492 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
3495 Version 4.0 17-Feb-03
3496 ---------------------
3498 1. If a comment in an extended regex that started immediately after a meta-item
3499 extended to the end of string, PCRE compiled incorrect data. This could lead to
3500 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
3502 2. Moved to autoconf 2.53 and libtool 1.4.2.
3504 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
3505 the special perltest8 script is no longer needed - all the tests can be run
3506 from a single perltest script.
3508 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
3509 by \s. It has now been removed in PCRE. This means it isn't recognized as
3510 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
3511 class [:space:] *does* include VT, thereby creating a mess.
3513 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
3514 space and tab.
3516 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
3517 its new features into the main test script, reducing the number of scripts.
3519 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
3520 were backward compatible, and made the (?i) apply to the whole pattern, as if
3521 /i were given. Now it behaves more logically, and applies the option setting
3522 only to what follows. PCRE has been changed to follow suit. However, if it
3523 finds options settings right at the start of the pattern, it extracts them into
3524 the global options, as before. Thus, they show up in the info data.
3526 8. Added support for the \Q...\E escape sequence. Characters in between are
3527 treated as literals. This is slightly different from Perl in that $ and @ are
3528 also handled as literals inside the quotes. In Perl, they will cause variable
3529 interpolation. Note the following examples:
3531 Pattern PCRE matches Perl matches
3533 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
3534 \Qabc\$xyz\E abc\$xyz abc\$xyz
3535 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
3537 For compatibility with Perl, \Q...\E sequences are recognized inside character
3538 classes as well as outside them.
3540 9. Re-organized 3 code statements in pcretest to avoid "overflow in
3541 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
3542 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
3543 signed/unsigned warnings.
3545 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
3546 option for pcretest, so I've replaced it by a simple function that does just
3547 that job.
3549 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
3550 "pcregrep -".
3552 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
3553 Java package. This provides some syntactic sugar for simple cases of what my
3554 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
3555 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
3556 item, you can use this simplified notation. Note that only makes sense with
3557 greedy quantifiers. Consequently, the use of the possessive quantifier forces
3558 greediness, whatever the setting of the PCRE_UNGREEDY option.
3560 13. A change of greediness default within a pattern was not taking effect at
3561 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
3562 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
3563 was abstracted outside.
3565 14. PCRE now supports the \G assertion. It is true when the current matching
3566 position is at the start point of the match. This differs from \A when the
3567 starting offset is non-zero. Used with the /g option of pcretest (or similar
3568 code), it works in the same way as it does for Perl's /g option. If all
3569 alternatives of a regex begin with \G, the expression is anchored to the start
3570 match position, and the "anchored" flag is set in the compiled expression.
3572 15. Some bugs concerning the handling of certain option changes within patterns
3573 have been fixed. These applied to options other than (?ims). For example,
3574 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
3575 been the other way round. Some of this was related to change 7 above.
3577 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
3578 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
3579 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
3580 POSIX classes only within a class (e.g. /[[:alpha:]]/).
3582 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
3583 mode. Unlike ".", it always matches newline, whatever the setting of
3584 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
3585 assertions. Perl allows it, but it doesn't (in general) work because it can't
3586 calculate the length of the lookbehind. At least, that's the case for Perl
3587 5.8.0 - I've been told they are going to document that it doesn't work in
3588 future.
3590 18. Added an error diagnosis for escapes that PCRE does not support: these are
3591 \L, \l, \N, \P, \p, \U, \u, and \X.
3593 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
3594 reading past the end of the pattern in cases such as /[abcd/.
3596 20. PCRE was getting more memory than necessary for patterns with classes that
3597 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
3599 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
3600 compiling PCRE for use with Virtual Pascal.
3602 22. Small fix to the Makefile to make it work properly if the build is done
3603 outside the source tree.
3605 23. Added a new extension: a condition to go with recursion. If a conditional
3606 subpattern starts with (?(R) the "true" branch is used if recursion has
3607 happened, whereas the "false" branch is used only at the top level.
3609 24. When there was a very long string of literal characters (over 255 bytes
3610 without UTF support, over 250 bytes with UTF support), the computation of how
3611 much memory was required could be incorrect, leading to segfaults or other
3612 strange effects.
3614 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
3615 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
3616 there was a subsequent back reference to those brackets. This meant that, for
3617 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
3618 possible to check for precisely this case. All we can do is abandon the
3619 optimization if .* occurs inside capturing brackets when there are any back
3620 references whatsoever. (See below for a better fix that came later.)
3622 26. The handling of the optimization for finding the first character of a
3623 non-anchored pattern, and for finding a character that is required later in the
3624 match were failing in some cases. This didn't break the matching; it just
3625 failed to optimize when it could. The way this is done has been re-implemented.
3627 27. Fixed typo in error message for invalid (?R item (it said "(?p").
3629 28. Added a new feature that provides some of the functionality that Perl
3630 provides with (?{...}). The facility is termed a "callout". The way it is done
3631 in PCRE is for the caller to provide an optional function, by setting
3632 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
3633 global variable. By default it is unset, which disables all calling out. To get
3634 the function called, the regex must include (?C) at appropriate points. This
3635 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
3636 This provides a means of identifying different callout points. When PCRE
3637 reaches such a point in the regex, if pcre_callout has been set, the external
3638 function is called. It is provided with data in a structure called
3639 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
3640 matching continues; if it returns a non-zero value, the match at the current
3641 point fails. However, backtracking will occur if possible. [This was changed
3642 later and other features added - see item 49 below.]
3644 29. pcretest is upgraded to test the callout functionality. It provides a
3645 callout function that displays information. By default, it shows the start of
3646 the match and the current position in the text. There are some new data escapes
3647 to vary what happens:
3649 \C+ in addition, show current contents of captured substrings
3650 \C- do not supply a callout function
3651 \C!n return 1 when callout number n is reached
3652 \C!n!m return 1 when callout number n is reached for the mth time
3654 30. If pcregrep was called with the -l option and just a single file name, it
3655 output "<stdin>" if a match was found, instead of the file name.
3657 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
3658 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
3659 pcre_exec(). This saves a malloc/free per call. The default value of
3660 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
3661 when configuring.
3663 32. The default maximum size of a compiled pattern is 64K. There have been a
3664 few cases of people hitting this limit. The code now uses macros to handle the
3665 storing of links as offsets within the compiled pattern. It defaults to 2-byte
3666 links, but this can be changed to 3 or 4 bytes by --with-link-size when
3667 configuring. Tests 2 and 5 work only with 2-byte links because they output
3668 debugging information about compiled patterns.
3670 33. Internal code re-arrangements:
3672 (a) Moved the debugging function for printing out a compiled regex into
3673 its own source file (printint.c) and used #include to pull it into
3674 pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
3675 separate copies.
3677 (b) Defined the list of op-code names for debugging as a macro in
3678 internal.h so that it is next to the definition of the opcodes.
3680 (c) Defined a table of op-code lengths for simpler skipping along compiled
3681 code. This is again a macro in internal.h so that it is next to the
3682 definition of the opcodes.
3684 34. Added support for recursive calls to individual subpatterns, along the
3685 lines of Robin Houston's patch (but implemented somewhat differently).
3687 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
3688 allow it to read and process whole directories in Win32. This code was
3689 contributed by Lionel Fourquaux; it has not been tested by me.
3691 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
3692 used to name a group. Names consist of alphanumerics and underscores, and must
3693 be unique. Back references use the syntax (?P=name) and recursive calls use
3694 (?P>name) which is a PCRE extension to the Python extension. Groups still have
3695 numbers. The function pcre_fullinfo() can be used after compilation to extract
3696 a name/number map. There are three relevant calls:
3698 PCRE_INFO_NAMEENTRYSIZE yields the size of each entry in the map
3699 PCRE_INFO_NAMECOUNT yields the number of entries
3700 PCRE_INFO_NAMETABLE yields a pointer to the map.
3702 The map is a vector of fixed-size entries. The size of each entry depends on
3703 the length of the longest name used. The first two bytes of each entry are the
3704 group number, most significant byte first. There follows the corresponding
3705 name, zero terminated. The names are in alphabetical order.
3707 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
3708 case instead of 255. Making it the same both with and without UTF-8 support
3709 means that the same test output works with both.
3711 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
3712 calling malloc() with a zero argument.
3714 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
3715 optimization. I've improved things by keeping a bitmap of backreferences with
3716 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
3717 fact referenced, the optimization can be applied. It is unlikely that a
3718 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
3719 the match to follow \n) will appear inside brackets with a number greater than
3720 31, but if it does, any back reference > 31 suppresses the optimization.
3722 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
3723 of disabling numbered capturing parentheses. Any opening parenthesis that is
3724 not followed by ? behaves as if it were followed by ?: but named parentheses
3725 can still be used for capturing (and they will acquire numbers in the usual
3726 way).
3728 41. Redesigned the return codes from the match() function into yes/no/error so
3729 that errors can be passed back from deep inside the nested calls. A malloc
3730 failure while inside a recursive subpattern call now causes the
3731 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
3733 42. It is now possible to set a limit on the number of times the match()
3734 function is called in a call to pcre_exec(). This facility makes it possible to
3735 limit the amount of recursion and backtracking, though not in a directly
3736 obvious way, because the match() function is used in a number of different
3737 circumstances. The count starts from zero for each position in the subject
3738 string (for non-anchored patterns). The default limit is, for compatibility, a
3739 large number, namely 10 000 000. You can change this in two ways:
3741 (a) When configuring PCRE before making, you can use --with-match-limit=n
3742 to set a default value for the compiled library.
3744 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
3745 a different value is set. See 45 below.
3747 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
3749 43. Added a new function pcre_config(int, void *) to enable run-time extraction
3750 of things that can be changed at compile time. The first argument specifies
3751 what is wanted and the second points to where the information is to be placed.
3752 The current list of available information is:
3756 The output is an integer that is set to one if UTF-8 support is available;
3757 otherwise it is set to zero.
3761 The output is an integer that it set to the value of the code that is used for
3762 newline. It is either LF (10) or CR (13).
3766 The output is an integer that contains the number of bytes used for internal
3767 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
3771 The output is an integer that contains the threshold above which the POSIX
3772 interface uses malloc() for output vectors. See item 31 above.
3776 The output is an unsigned integer that contains the default limit of the number
3777 of match() calls in a pcre_exec() execution. See 42 above.
3779 44. pcretest has been upgraded by the addition of the -C option. This causes it
3780 to extract all the available output from the new pcre_config() function, and to
3781 output it. The program then exits immediately.
3783 45. A need has arisen to pass over additional data with calls to pcre_exec() in
3784 order to support additional features. One way would have been to define
3785 pcre_exec2() (for example) with extra arguments, but this would not have been
3786 extensible, and would also have required all calls to the original function to
3787 be mapped to the new one. Instead, I have chosen to extend the mechanism that
3788 is used for passing in "extra" data from pcre_study().
3790 The pcre_extra structure is now exposed and defined in pcre.h. It currently
3791 contains the following fields:
3793 flags a bitmap indicating which of the following fields are set
3794 study_data opaque data from pcre_study()
3795 match_limit a way of specifying a limit on match() calls for a specific
3796 call to pcre_exec()
3797 callout_data data for callouts (see 49 below)
3799 The flag bits are also defined in pcre.h, and are
3805 The pcre_study() function now returns one of these new pcre_extra blocks, with
3806 the actual study data pointed to by the study_data field, and the
3807 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
3808 before. That is, this change is entirely upwards-compatible and requires no
3809 change to existing code.
3811 If you want to pass in additional data to pcre_exec(), you can either place it
3812 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
3813 block.
3815 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
3816 data string contains the escape sequence \M, pcretest calls pcre_exec() several
3817 times with different match limits, until it finds the minimum value needed for
3818 pcre_exec() to complete. The value is then output. This can be instructive; for
3819 most simple matches the number is quite small, but for pathological cases it
3820 gets very large very quickly.
3822 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
3823 returns the size of the data block pointed to by the study_data field in a
3824 pcre_extra block, that is, the value that was passed as the argument to
3825 pcre_malloc() when PCRE was getting memory in which to place the information
3826 created by pcre_study(). The fourth argument should point to a size_t variable.
3827 pcretest has been extended so that this information is shown after a successful
3828 pcre_study() call when information about the compiled regex is being displayed.
3830 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
3831 because what follows is always an absolute path. (Later: it turns out that this
3832 is more than cosmetic for MinGW, because it doesn't like empty path
3833 components.)
3835 49. Some changes have been made to the callout feature (see 28 above):
3837 (i) A callout function now has three choices for what it returns:
3839 0 => success, carry on matching
3840 > 0 => failure at this point, but backtrack if possible
3841 < 0 => serious error, return this value from pcre_exec()
3843 Negative values should normally be chosen from the set of PCRE_ERROR_xxx
3844 values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
3845 "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
3846 use by callout functions. It will never be used by PCRE itself.
3848 (ii) The pcre_extra structure (see 45 above) has a void * field called
3849 callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
3850 pcre_callout_block structure has a field of the same name. The contents of
3851 the field passed in the pcre_extra structure are passed to the callout
3852 function in the corresponding field in the callout block. This makes it
3853 easier to use the same callout-containing regex from multiple threads. For
3854 testing, the pcretest program has a new data escape
3856 \C*n pass the number n (may be negative) as callout_data
3858 If the callout function in pcretest receives a non-zero value as
3859 callout_data, it returns that value.
3861 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
3862 there were some redundant $(CFLAGS) in commands that are now specified as
3863 $(LINK), which already includes $(CFLAGS).
3865 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
3866 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
3867 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
3868 one-byte characters throughout. Note that case-insensitive matching applies
3869 only to characters whose values are less than 256. PCRE doesn't support the
3870 notion of cases for higher-valued characters.
3872 (i) A character class whose characters are all within 0-255 is handled as
3873 a bit map, and the map is inverted for negative classes. Previously, a
3874 character > 255 always failed to match such a class; however it should
3875 match if the class was a negative one (e.g. [^ab]). This has been fixed.
3877 (ii) A negated character class with a single character < 255 is coded as
3878 "not this character" (OP_NOT). This wasn't working properly when the test
3879 character was multibyte, either singly or repeated.
3881 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
3882 mode, for example: \x{100}{2,3}.
3884 (iv) The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
3885 singly or repeated) now correctly test multibyte characters. However,
3886 PCRE doesn't recognize any characters with values greater than 255 as
3887 digits, spaces, or word characters. Such characters always match \D, \S,
3888 and \W, and never match \d, \s, or \w.
3890 (v) Classes may now contain characters and character ranges with values
3891 greater than 255. For example: [ab\x{100}-\x{400}].
3893 (vi) pcregrep now has a --utf-8 option (synonym -u) which makes it call
3894 PCRE in UTF-8 mode.
3896 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
3897 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
3898 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
3899 value.)
3901 53. The single man page has become too large. I have therefore split it up into
3902 a number of separate man pages. These also give rise to individual HTML pages;
3903 these are now put in a separate directory, and there is an index.html page that
3904 lists them all. Some hyperlinking between the pages has been installed.
3906 54. Added convenience functions for handling named capturing parentheses.
3908 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
3909 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
3910 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
3911 are faulted.
3913 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
3914 calling configure. These values are used when compiling the dftables.c program
3915 which is run to generate the source of the default character tables. They
3916 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
3917 you will need to set these values.
3919 57. Updated the building process for Windows DLL, as provided by Fred Cox.
3922 Version 3.9 02-Jan-02
3923 ---------------------
3925 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
3927 2. If --disable-static was given, the building process failed when trying to
3928 build pcretest and pcregrep. (For some reason it was using libtool to compile
3929 them, which is not right, as they aren't part of the library.)
3932 Version 3.8 18-Dec-01
3933 ---------------------
3935 1. The experimental UTF-8 code was completely screwed up. It was packing the
3936 bytes in the wrong order. How dumb can you get?
3939 Version 3.7 29-Oct-01
3940 ---------------------
3942 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
3943 This caused pcretest, when used on the test data, to segfault. Unfortunately,
3944 this didn't happen under Solaris 8, where I normally test things.
3946 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
3947 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
3948 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
3949 this fix an hour or so after the initial 3.7 release.)
3952 Version 3.6 23-Oct-01
3953 ---------------------
3955 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
3956 offsets passed as NULL with zero offset count.
3958 2. The config.guess and config.sub files had not been updated when I moved to
3959 the latest autoconf.
3962 Version 3.5 15-Aug-01
3963 ---------------------
3965 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
3966 had been forgotten.
3968 2. By using declared but undefined structures, we can avoid using "void"
3969 definitions in pcre.h while keeping the internal definitions of the structures
3970 private.
3972 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
3973 user point of view, this means that both static and shared libraries are built
3974 by default, but this can be individually controlled. More of the work of
3975 handling this static/shared cases is now inside libtool instead of PCRE's make
3976 file.
3978 4. The pcretest utility is now installed along with pcregrep because it is
3979 useful for users (to test regexs) and by doing this, it automatically gets
3980 relinked by libtool. The documentation has been turned into a man page, so
3981 there are now .1, .txt, and .html versions in /doc.
3983 5. Upgrades to pcregrep:
3984 (i) Added long-form option names like gnu grep.
3985 (ii) Added --help to list all options with an explanatory phrase.
3986 (iii) Added -r, --recursive to recurse into sub-directories.
3987 (iv) Added -f, --file to read patterns from a file.
3989 6. pcre_exec() was referring to its "code" argument before testing that
3990 argument for NULL (and giving an error if it was NULL).
3992 7. Upgraded Makefile.in to allow for compiling in a different directory from
3993 the source directory.
3995 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
3996 options bits, the pointer it was passed was to an int instead of to an unsigned
3997 long int. This mattered only on 64-bit systems.
3999 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
4000 generated) instead of pcre.in, which it its source. Also made the same change
4001 in several of the .c files.
4003 10. A new release of gcc defines printf() as a macro, which broke pcretest
4004 because it had an ifdef in the middle of a string argument for printf(). Fixed
4005 by using separate calls to printf().
4007 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
4008 script, to force use of CR or LF instead of \n in the source. On non-Unix
4009 systems, the value can be set in config.h.
4011 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
4012 absolute limit. Changed the text of the error message to make this clear, and
4013 likewise updated the man page.
4015 13. The limit of 99 on the number of capturing subpatterns has been removed.
4016 The new limit is 65535, which I hope will not be a "real" limit.
4019 Version 3.4 22-Aug-00
4020 ---------------------
4022 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
4024 2. Diagnose condition (?(0) as an error instead of crashing on matching.
4027 Version 3.3 01-Aug-00
4028 ---------------------
4030 1. If an octal character was given, but the value was greater than \377, it
4031 was not getting masked to the least significant bits, as documented. This could
4032 lead to crashes in some systems.
4034 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
4035 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
4037 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
4038 These just pass their arguments on to (pcre_free)(), but they are provided
4039 because some uses of PCRE bind it to non-C systems that can call its functions,
4040 but cannot call free() or pcre_free() directly.
4042 4. Add "make test" as a synonym for "make check". Corrected some comments in
4043 the Makefile.
4045 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
4046 Makefile.
4048 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
4049 command called pgrep for grepping around the active processes.
4051 7. Added the beginnings of support for UTF-8 character strings.
4053 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
4054 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
4055 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
4056 out for the ar command.)
4059 Version 3.2 12-May-00
4060 ---------------------
4062 This is purely a bug fixing release.
4064 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
4065 of ZA. This was just one example of several cases that could provoke this bug,
4066 which was introduced by change 9 of version 2.00. The code for breaking
4067 infinite loops after an iteration that matches an empty string was't working
4068 correctly.
4070 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
4071 when matched against abbab (for example). After matching an empty string, it
4072 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
4073 caused it to match further down the string than it should.
4075 3. The code contained an inclusion of sys/types.h. It isn't clear why this
4076 was there because it doesn't seem to be needed, and it causes trouble on some
4077 systems, as it is not a Standard C header. It has been removed.
4079 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
4080 were reported on the Macintosh. The changes were from
4082 while ((c = *(++ptr)) != 0 && c != '\n');
4083 to
4084 while ((c = *(++ptr)) != 0 && c != '\n') ;
4086 Totally extraordinary, but if that's what it takes...
4088 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
4089 available. Added HAVE_BCOPY and an autoconf test for it; if neither
4090 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
4091 assumes the way PCRE uses memmove() (always moving upwards).
4093 6. PCRE is being used in one environment where strchr() is not available. There
4094 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
4095 faster code anyway.
4098 Version 3.1 09-Feb-00
4099 ---------------------
4101 The only change in this release is the fixing of some bugs in Makefile.in for
4102 the "install" target:
4104 (1) It was failing to install pcreposix.h.
4106 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
4109 Version 3.0 01-Feb-00
4110 ---------------------
4112 1. Add support for the /+ modifier to perltest (to output $` like it does in
4113 pcretest).
4115 2. Add support for the /g modifier to perltest.
4117 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
4118 matches null strings.
4120 4. Fix perltest so that it doesn't do unwanted things when fed an empty
4121 pattern. Perl treats empty patterns specially - it reuses the most recent
4122 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
4123 effect.
4125 5. The POSIX interface was broken in that it was just handing over the POSIX
4126 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
4127 required a bigger vector, with some working space on the end. This means that
4128 the POSIX wrapper now has to get and free some memory, and copy the results.
4130 6. Added some simple autoconf support, placing the test data and the
4131 documentation in separate directories, re-organizing some of the
4132 information files, and making it build pcre-config (a GNU standard). Also added
4133 libtool support for building PCRE as a shared library, which is now the
4134 default.
4136 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
4137 09 are not valid octal constants. Single digits will be used for minor values
4138 less than 10.
4140 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
4141 existing programs that set these in the POSIX interface can use PCRE without
4142 modification.
4144 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
4145 return all that pcre_info() returns, plus additional data. The pcre_info()
4146 function is retained for compatibility, but is considered to be obsolete.
4148 10. Added experimental recursion feature (?R) to handle one common case that
4149 Perl 5.6 will be able to do with (?p{...}).
4151 11. Added support for POSIX character classes like [:alpha:], which Perl is
4152 adopting.
4155 Version 2.08 31-Aug-99
4156 ----------------------
4158 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
4159 trying to match at the startoffset position, but instead was moving forward to
4160 the next newline as if a previous match had failed.
4162 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
4163 and could get into a loop if a null string was matched other than at the start
4164 of the subject.
4166 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
4167 be distinguished at compile time, and for completeness also added PCRE_DATE.
4169 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
4170 in GnuWin32 environments.
4173 Version 2.07 29-Jul-99
4174 ----------------------
4176 1. The documentation is now supplied in plain text form and HTML as well as in
4177 the form of man page sources.
4179 2. C++ compilers don't like assigning (void *) values to other pointer types.
4180 In particular this affects malloc(). Although there is no problem in Standard
4181 C, I've put in casts to keep C++ compilers happy.
4183 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
4184 should be (const char *).
4186 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
4187 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
4188 However, I haven't made this a standard facility. The documentation doesn't
4189 mention it, and the Makefile doesn't support it.
4191 5. The Makefile now contains an "install" target, with editable destinations at
4192 the top of the file. The pcretest program is not installed.
4194 6. pgrep -V now gives the PCRE version number and date.
4196 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
4197 causing the entire string to be ignored, instead of just the last character.
4199 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
4200 non-matching string, it can take a very, very long time, even for strings of
4201 quite modest length, because of the nested recursion. PCRE now does better in
4202 some of these cases. It does this by remembering the last required literal
4203 character in the pattern, and pre-searching the subject to ensure it is present
4204 before running the real match. In other words, it applies a heuristic to detect
4205 some types of certain failure quickly, and in the above example, if presented
4206 with a string that has no trailing " it gives "no match" very quickly.
4208 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
4209 other alternatives are tried instead.
4212 Version 2.06 09-Jun-99
4213 ----------------------
4215 1. Change pcretest's output for amount of store used to show just the code
4216 space, because the remainder (the data block) varies in size between 32-bit and
4217 64-bit systems.
4219 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
4220 start matching at. This allows lookbehinds to work when searching for multiple
4221 occurrences in a string.
4223 3. Added additional options to pcretest for testing multiple occurrences:
4225 /+ outputs the rest of the string that follows a match
4226 /g loops for multiple occurrences, using the new startoffset argument
4227 /G loops for multiple occurrences by passing an incremented pointer
4229 4. PCRE wasn't doing the "first character" optimization for patterns starting
4230 with \b or \B, though it was doing it for other lookbehind assertions. That is,
4231 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
4232 the letter 'x'. On long subject strings, this gives a significant speed-up.
4235 Version 2.05 21-Apr-99
4236 ----------------------
4238 1. Changed the type of magic_number from int to long int so that it works
4239 properly on 16-bit systems.
4241 2. Fixed a bug which caused patterns starting with .* not to work correctly
4242 when the subject string contained newline characters. PCRE was assuming
4243 anchoring for such patterns in all cases, which is not correct because .* will
4244 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
4245 DOTALL is set at top level; otherwise it knows that patterns starting with .*
4246 must be retried after every newline in the subject.
4249 Version 2.04 18-Feb-99
4250 ----------------------
4252 1. For parenthesized subpatterns with repeats whose minimum was zero, the
4253 computation of the store needed to hold the pattern was incorrect (too large).
4254 If such patterns were nested a few deep, this could multiply and become a real
4255 problem.
4257 2. Added /M option to pcretest to show the memory requirement of a specific
4258 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
4260 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
4261 compiled in such a way that the backtracking after subsequent failure was
4262 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
4263 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
4266 Version 2.03 02-Feb-99
4267 ----------------------
4269 1. Fixed typo and small mistake in man page.
4271 2. Added 4th condition (GPL supersedes if conflict) and created separate
4272 LICENCE file containing the conditions.
4274 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
4275 Perl, that is the internal \ allows the delimiter to be included in the
4276 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
4277 the final delimiter, add \ to the end of the pattern (to test the error).
4279 4. Added the convenience functions for extracting substrings after a successful
4280 match. Updated pcretest to make it able to test these functions.
4283 Version 2.02 14-Jan-99
4284 ----------------------
4286 1. Initialized the working variables associated with each extraction so that
4287 their saving and restoring doesn't refer to uninitialized store.
4289 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
4290 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
4291 fix the problem.
4293 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
4294 calls, and wasn't printing the correct value for compiling calls. Increased the
4295 default value of LOOPREPEAT, and the number of significant figures in the
4296 times.
4298 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
4300 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
4301 a building problem on Windows NT with a FAT file system.
4304 Version 2.01 21-Oct-98
4305 ----------------------
4307 1. Changed the API for pcre_compile() to allow for the provision of a pointer
4308 to character tables built by pcre_maketables() in the current locale. If NULL
4309 is passed, the default tables are used.
4312 Version 2.00 24-Sep-98
4313 ----------------------
4315 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
4316 it any more.
4318 2. Allow quantification of (?>) groups, and make it work correctly.
4320 3. The first character computation wasn't working for (?>) groups.
4322 4. Correct the implementation of \Z (it is permitted to match on the \n at the
4323 end of the subject) and add 5.005's \z, which really does match only at the
4324 very end of the subject.
4326 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
4328 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
4329 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
4330 localized options. All options to pcre_study() were also removed.
4332 7. Add other new features from 5.005:
4334 $(?<= positive lookbehind
4335 $(?<! negative lookbehind
4336 (?imsx-imsx) added the unsetting capability
4337 such a setting is global if at outer level; local otherwise
4338 (?imsx-imsx:) non-capturing groups with option setting
4339 (?(cond)re|re) conditional pattern matching
4341 A backreference to itself in a repeated group matches the previous
4342 captured string.
4344 8. General tidying up of studying (both automatic and via "study")
4345 consequential on the addition of new assertions.
4347 9. As in 5.005, unlimited repeated groups that could match an empty substring
4348 are no longer faulted at compile time. Instead, the loop is forcibly broken at
4349 runtime if any iteration does actually match an empty substring.
4351 10. Include the RunTest script in the distribution.
4353 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
4354 discrepancies, some of which were old and were also with respect to 5.004. They
4355 have now been fixed.
4358 Version 1.09 28-Apr-98
4359 ----------------------
4361 1. A negated single character class followed by a quantifier with a minimum
4362 value of one (e.g. [^x]{1,6} ) was not compiled correctly. This could lead to
4363 program crashes, or just wrong answers. This did not apply to negated classes
4364 containing more than one character, or to minima other than one.
4367 Version 1.08 27-Mar-98
4368 ----------------------
4370 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
4372 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
4373 latter must appear before anything that relies on it in the pattern.
4376 Version 1.07 16-Feb-98
4377 ----------------------
4379 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
4380 repeat of a potentially empty string).
4383 Version 1.06 23-Jan-98
4384 ----------------------
4386 1. Added Markus Oberhumer's little patches for C++.
4388 2. Literal strings longer than 255 characters were broken.
4391 Version 1.05 23-Dec-97
4392 ----------------------
4394 1. Negated character classes containing more than one character were failing if
4395 PCRE_CASELESS was set at run time.
4398 Version 1.04 19-Dec-97
4399 ----------------------
4401 1. Corrected the man page, where some "const" qualifiers had been omitted.
4403 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
4404 input syntax.
4406 3. Fixed memory leak which occurred when a regex with back references was
4407 matched with an offsets vector that wasn't big enough. The temporary memory
4408 that is used in this case wasn't being freed if the match failed.
4410 4. Tidied pcretest to ensure it frees memory that it gets.
4412 5. Temporary memory was being obtained in the case where the passed offsets
4413 vector was exactly big enough.
4415 6. Corrected definition of offsetof() from change 5 below.
4417 7. I had screwed up change 6 below and broken the rules for the use of
4418 setjmp(). Now fixed.
4421 Version 1.03 18-Dec-97
4422 ----------------------
4424 1. A erroneous regex with a missing opening parenthesis was correctly
4425 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
4426 on some systems.
4428 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
4429 it was reported that one broken compiler failed on the former because "code" is
4430 also an independent variable.
4432 3. The erroneous regex a[]b caused an array overrun reference.
4434 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
4435 fail on data ending with that character. (It was going on too far, and checking
4436 the next character, typically a binary zero.) This was specific to the
4437 optimized code for single-character negative classes.
4439 5. Added a contributed patch from the TIN world which does the following:
4441 + Add an undef for memmove, in case the the system defines a macro for it.
4443 + Add a definition of offsetof(), in case there isn't one. (I don't know
4444 the reason behind this - offsetof() is part of the ANSI standard - but
4445 it does no harm).
4447 + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
4448 most of the places where whitespace preceded '#'. I have given up and
4449 allowed the remaining 2 cases to be at the margin.
4451 + Rename some variables in pcre to eliminate shadowing. This seems very
4452 pedantic, but does no harm, of course.
4454 6. Moved the call to setjmp() into its own function, to get rid of warnings
4455 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
4457 7. Constructs such as \d{8,} were compiling into the equivalent of
4458 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
4459 outcome, but in this particular case used more store than had been allocated,
4460 which caused the bug to be discovered because it threw up an internal error.
4462 8. The debugging code in both pcre and pcretest for outputting the compiled
4463 form of a regex was going wrong in the case of back references followed by
4464 curly-bracketed repeats.
4467 Version 1.02 12-Dec-97
4468 ----------------------
4470 1. Typos in pcre.3 and comments in the source fixed.
4472 2. Applied a contributed patch to get rid of places where it used to remove
4473 'const' from variables, and fixed some signed/unsigned and uninitialized
4474 variable warnings.
4476 3. Added the "runtest" target to Makefile.
4478 4. Set default compiler flag to -O2 rather than just -O.
4481 Version 1.01 19-Nov-97
4482 ----------------------
4484 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
4485 like /([ab]*)*/, that is, for classes with more than one character in them.
4487 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
4488 as /((?>a*))*/ (a PCRE_EXTRA facility).
4491 Version 1.00 18-Nov-97
4492 ----------------------
4494 1. Added compile-time macros to support systems such as SunOS4 which don't have
4495 memmove() or strerror() but have other things that can be used instead.
4497 2. Arranged that "make clean" removes the executables.
4500 Version 0.99 27-Oct-97
4501 ----------------------
4503 1. Fixed bug in code for optimizing classes with only one character. It was
4504 initializing a 32-byte map regardless, which could cause it to run off the end
4505 of the memory it had got.
4507 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
4510 Version 0.98 22-Oct-97
4511 ----------------------
4513 1. Fixed bug in code for handling temporary memory usage when there are more
4514 back references than supplied space in the ovector. This could cause segfaults.
4517 Version 0.97 21-Oct-97
4518 ----------------------
4520 1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
4522 2. Optimized negated single characters not to use a bit map.
4524 3. Brought error texts together as macro definitions; clarified some of them;
4525 fixed one that was wrong - it said "range out of order" when it meant "invalid
4526 escape sequence".
4528 4. Changed some char * arguments to const char *.
4530 5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
4532 6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
4533 pcretest.
4536 Version 0.96 16-Oct-97
4537 ----------------------
4539 1. Added a simple "pgrep" utility to the distribution.
4541 2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
4542 unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
4543 where "ddd" means "one or more decimal digits".
4545 3. Fixed serious bug. If a pattern had a back reference, but the call to
4546 pcre_exec() didn't supply a large enough ovector to record the related
4547 identifying subpattern, the match always failed. PCRE now remembers the number
4548 of the largest back reference, and gets some temporary memory in which to save
4549 the offsets during matching if necessary, in order to ensure that
4550 backreferences always work.
4552 4. Increased the compatibility with Perl in a number of ways:
4554 (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
4555 to request this handling. The option can be set at compile or exec time.
4557 (b) $ matches before a terminating newline by default; an option
4558 PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
4559 mode). The option can be set at compile or exec time.
4561 (c) The handling of \ followed by a digit other than 0 is now supposed to be
4562 the same as Perl's. If the decimal number it represents is less than 10
4563 or there aren't that many previous left capturing parentheses, an octal
4564 escape is read. Inside a character class, it's always an octal escape,
4565 even if it is a single digit.
4567 (d) An escaped but undefined alphabetic character is taken as a literal,
4568 unless PCRE_EXTRA is set. Currently this just reserves the remaining
4569 escapes.
4571 (e) {0} is now permitted. (The previous item is removed from the compiled
4572 pattern).
4574 5. Changed all the names of code files so that the basic parts are no longer
4575 than 10 characters, and abolished the teeny "globals.c" file.
4577 6. Changed the handling of character classes; they are now done with a 32-byte
4578 bit map always.
4580 7. Added the -d and /D options to pcretest to make it possible to look at the
4581 internals of compilation without having to recompile pcre.
4584 Version 0.95 23-Sep-97
4585 ----------------------
4587 1. Fixed bug in pre-pass concerning escaped "normal" characters such as \x5c or
4588 \x20 at the start of a run of normal characters. These were being treated as
4589 real characters, instead of the source characters being re-checked.
4592 Version 0.94 18-Sep-97
4593 ----------------------
4595 1. The functions are now thread-safe, with the caveat that the global variables
4596 containing pointers to malloc() and free() or alternative functions are the
4597 same for all threads.
4599 2. Get pcre_study() to generate a bitmap of initial characters for non-
4600 anchored patterns when this is possible, and use it if passed to pcre_exec().
4603 Version 0.93 15-Sep-97
4604 ----------------------
4606 1. /(b)|(:+)/ was computing an incorrect first character.
4608 2. Add pcre_study() to the API and the passing of pcre_extra to pcre_exec(),
4609 but not actually doing anything yet.
4611 3. Treat "-" characters in classes that cannot be part of ranges as literals,
4612 as Perl does (e.g. [-az] or [az-]).
4614 4. Set the anchored flag if a branch starts with .* or .*? because that tests
4615 all possible positions.
4617 5. Split up into different modules to avoid including unneeded functions in a
4618 compiled binary. However, compile and exec are still in one module. The "study"
4619 function is split off.
4621 6. The character tables are now in a separate module whose source is generated
4622 by an auxiliary program - but can then be edited by hand if required. There are
4623 now no calls to isalnum(), isspace(), isdigit(), isxdigit(), tolower() or
4624 toupper() in the code.
4626 7. Turn the malloc/free funtions variables into pcre_malloc and pcre_free and
4627 make them global. Abolish the function for setting them, as the caller can now
4628 set them directly.
4631 Version 0.92 11-Sep-97
4632 ----------------------
4634 1. A repeat with a fixed maximum and a minimum of 1 for an ordinary character
4635 (e.g. /a{1,3}/) was broken (I mis-optimized it).
4637 2. Caseless matching was not working in character classes if the characters in
4638 the pattern were in upper case.
4640 3. Make ranges like [W-c] work in the same way as Perl for caseless matching.
4642 4. Make PCRE_ANCHORED public and accept as a compile option.
4644 5. Add an options word to pcre_exec() and accept PCRE_ANCHORED and
4645 PCRE_CASELESS at run time. Add escapes \A and \I to pcretest to cause it to
4646 pass them.
4648 6. Give an error if bad option bits passed at compile or run time.
4650 7. Add PCRE_MULTILINE at compile and exec time, and (?m) as well. Add \M to
4651 pcretest to cause it to pass that flag.
4653 8. Add pcre_info(), to get the number of identifying subpatterns, the stored
4654 options, and the first character, if set.
4656 9. Recognize C+ or C{n,m} where n >= 1 as providing a fixed starting character.
4659 Version 0.91 10-Sep-97
4660 ----------------------
4662 1. PCRE was failing to diagnose unlimited repeats of subpatterns that could
4663 match the empty string as in /(a*)*/. It was looping and ultimately crashing.
4665 2. PCRE was looping on encountering an indefinitely repeated back reference to
4666 a subpattern that had matched an empty string, e.g. /(a|)\1*/. It now does what
4667 Perl does - treats the match as successful.
4669 ****


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

  ViewVC Help
Powered by ViewVC 1.1.5