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

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log


Revision 936 - (show annotations)
Sat Feb 25 17:02:23 2012 UTC (3 years, 6 months ago) by ph10
File size: 228318 byte(s)
Error occurred while calculating annotation data.
Add support for linking pcretest with libedit instead of libreadline.
1 ChangeLog for PCRE
2 ------------------
3
4 Version 8.31
5 -----------------------------
6
7 1. Fixing a wrong JIT test case and some compiler warnings.
8
9 2. Removed a bashism from the RunTest script.
10
11 3. Add a cast to pcre_exec.c to fix the warning "unary minus operator applied
12 to unsigned type, result still unsigned" that was given by an MS compiler
13 on encountering the code "-sizeof(xxx)".
14
15 4. Partial matching support is added to the JIT compiler.
16
17 5. Fixed several bugs concerned with partial matching of items that consist
18 of more than one character:
19
20 (a) /^(..)\1/ did not partially match "aba" because checking references was
21 done on an "all or nothing" basis. This also applied to repeated
22 references.
23
24 (b) \R did not give a hard partial match if \r was found at the end of the
25 subject.
26
27 (c) \X did not give a hard partial match after matching one or more
28 characters at the end of the subject.
29
30 (d) When newline was set to CRLF, a pattern such as /a$/ did not recognize
31 a partial match for the string "\r".
32
33 (e) When newline was set to CRLF, the metacharacter "." did not recognize
34 a partial match for a CR character at the end of the subject string.
35
36 6. If JIT is requested using /S++ or -s++ (instead of just /S+ or -s+) when
37 running pcretest, the text "(JIT)" added to the output whenever JIT is
38 actually used to run the match.
39
40 7. Individual JIT compile options can be set in pcretest by following -s+[+]
41 or /S+[+] with a digit between 1 and 7.
42
43 8. OP_NOT now supports any UTF character not just single-byte ones.
44
45 9. (*MARK) control verb is now supported by the JIT compiler.
46
47 10. The command "./RunTest list" lists the available tests without actually
48 running any of them. (Because I keep forgetting what they all are.)
49
50 11. Add PCRE_INFO_MAXLOOKBEHIND.
51
52 12. Applied a (slightly modified) user-supplied patch that improves performance
53 when the heap is used for recursion (compiled with --disable-stack-for-
54 recursion). Instead of malloc and free for each heap frame each time a
55 logical recursion happens, frames are retained on a chain and re-used where
56 possible. This sometimes gives as much as 30% improvement.
57
58 13. (*COMMIT) is now correctly confined to within a recursive subpattern call.
59
60 14. It is now possible to link pcretest with libedit as an alternative to
61 libreadline.
62
63
64 Version 8.30 04-February-2012
65 -----------------------------
66
67 1. Renamed "isnumber" as "is_a_number" because in some Mac environments this
68 name is defined in ctype.h.
69
70 2. Fixed a bug in fixed-length calculation for lookbehinds that would show up
71 only in quite long subpatterns.
72
73 3. Removed the function pcre_info(), which has been obsolete and deprecated
74 since it was replaced by pcre_fullinfo() in February 2000.
75
76 4. For a non-anchored pattern, if (*SKIP) was given with a name that did not
77 match a (*MARK), and the match failed at the start of the subject, a
78 reference to memory before the start of the subject could occur. This bug
79 was introduced by fix 17 of release 8.21.
80
81 5. A reference to an unset group with zero minimum repetition was giving
82 totally wrong answers (in non-JavaScript-compatibility mode). For example,
83 /(another)?(\1?)test/ matched against "hello world test". This bug was
84 introduced in release 8.13.
85
86 6. Add support for 16-bit character strings (a large amount of work involving
87 many changes and refactorings).
88
89 7. RunGrepTest failed on msys because \r\n was replaced by whitespace when the
90 command "pattern=`printf 'xxx\r\njkl'`" was run. The pattern is now taken
91 from a file.
92
93 8. Ovector size of 2 is also supported by JIT based pcre_exec (the ovector size
94 rounding is not applied in this particular case).
95
96 9. The invalid Unicode surrogate codepoints U+D800 to U+DFFF are now rejected
97 if they appear, or are escaped, in patterns.
98
99 10. Get rid of a number of -Wunused-but-set-variable warnings.
100
101 11. The pattern /(?=(*:x))(q|)/ matches an empty string, and returns the mark
102 "x". The similar pattern /(?=(*:x))((*:y)q|)/ did not return a mark at all.
103 Oddly, Perl behaves the same way. PCRE has been fixed so that this pattern
104 also returns the mark "x". This bug applied to capturing parentheses,
105 non-capturing parentheses, and atomic parentheses. It also applied to some
106 assertions.
107
108 12. Stephen Kelly's patch to CMakeLists.txt allows it to parse the version
109 information out of configure.ac instead of relying on pcre.h.generic, which
110 is not stored in the repository.
111
112 13. Applied Dmitry V. Levin's patch for a more portable method for linking with
113 -lreadline.
114
115 14. ZH added PCRE_CONFIG_JITTARGET; added its output to pcretest -C.
116
117 15. Applied Graycode's patch to put the top-level frame on the stack rather
118 than the heap when not using the stack for recursion. This gives a
119 performance improvement in many cases when recursion is not deep.
120
121 16. Experimental code added to "pcretest -C" to output the stack frame size.
122
123
124 Version 8.21 12-Dec-2011
125 ------------------------
126
127 1. Updating the JIT compiler.
128
129 2. JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
130 are added as well.
131
132 3. Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
133 PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
134 calling _pcre_jit_exec. Some extra comments are added.
135
136 4. (*MARK) settings inside atomic groups that do not contain any capturing
137 parentheses, for example, (?>a(*:m)), were not being passed out. This bug
138 was introduced by change 18 for 8.20.
139
140 5. Supporting of \x, \U and \u in JavaScript compatibility mode based on the
141 ECMA-262 standard.
142
143 6. Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
144 erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
145 This bug was probably introduced by change 9 of 8.13.
146
147 7. While fixing 6 above, I noticed that a number of other items were being
148 incorrectly rejected as "not fixed length". This arose partly because newer
149 opcodes had not been added to the fixed-length checking code. I have (a)
150 corrected the bug and added tests for these items, and (b) arranged for an
151 error to occur if an unknown opcode is encountered while checking for fixed
152 length instead of just assuming "not fixed length". The items that were
153 rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
154 (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
155 repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
156
157 8. A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
158 being incorrectly compiled and would have given unpredicatble results.
159
160 9. A possessively repeated subpattern with minimum repeat count greater than
161 one behaved incorrectly. For example, (A){2,}+ behaved as if it was
162 (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
163 the first (A) could occur when it should not.
164
165 10. Add a cast and remove a redundant test from the code.
166
167 11. JIT should use pcre_malloc/pcre_free for allocation.
168
169 12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
170 best practice nowadays, and helps with cross-compiling. (If the exec_prefix
171 is anything other than /usr, -L is still shown).
172
173 13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
174
175 14. Perl does not support \N without a following name in a [] class; PCRE now
176 also gives an error.
177
178 15. If a forward reference was repeated with an upper limit of around 2000,
179 it caused the error "internal error: overran compiling workspace". The
180 maximum number of forward references (including repeats) was limited by the
181 internal workspace, and dependent on the LINK_SIZE. The code has been
182 rewritten so that the workspace expands (via pcre_malloc) if necessary, and
183 the default depends on LINK_SIZE. There is a new upper limit (for safety)
184 of around 200,000 forward references. While doing this, I also speeded up
185 the filling in of repeated forward references.
186
187 16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
188 incorrectly expecting the subject to contain another "a" after the start.
189
190 17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
191 in the match, the SKIP should be ignored. This was not happening; instead
192 the SKIP was being treated as NOMATCH. For patterns such as
193 /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
194 tested.
195
196 18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
197 now much more compatible with Perl, in particular in cases where the result
198 is a non-match for a non-anchored pattern. For example, if
199 /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
200 "m", where previously it did not return a name. A side effect of this
201 change is that for partial matches, the last encountered mark name is
202 returned, as for non matches. A number of tests that were previously not
203 Perl-compatible have been moved into the Perl-compatible test files. The
204 refactoring has had the pleasing side effect of removing one argument from
205 the match() function, thus reducing its stack requirements.
206
207 19. If the /S+ option was used in pcretest to study a pattern using JIT,
208 subsequent uses of /S (without +) incorrectly behaved like /S+.
209
210 21. Retrieve executable code size support for the JIT compiler and fixing
211 some warnings.
212
213 22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
214 not work when the shorter character appeared right at the end of the
215 subject string.
216
217 23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
218 systems.
219
220 24. Added PCRE_INFO_JITSIZE to pass on the value from (21) above, and also
221 output it when the /M option is used in pcretest.
222
223 25. The CheckMan script was not being included in the distribution. Also, added
224 an explicit "perl" to run Perl scripts from the PrepareRelease script
225 because this is reportedly needed in Windows.
226
227 26. If study data was being save in a file and studying had not found a set of
228 "starts with" bytes for the pattern, the data written to the file (though
229 never used) was taken from uninitialized memory and so caused valgrind to
230 complain.
231
232 27. Updated RunTest.bat as provided by Sheri Pierce.
233
234 28. Fixed a possible uninitialized memory bug in pcre_jit_compile.c.
235
236 29. Computation of memory usage for the table of capturing group names was
237 giving an unnecessarily large value.
238
239
240 Version 8.20 21-Oct-2011
241 ------------------------
242
243 1. Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
244 a POSIX class. After further experiments with Perl, which convinced me that
245 Perl has bugs and confusions, a closing square bracket is no longer allowed
246 in a POSIX name. This bug also affected patterns with classes that started
247 with full stops.
248
249 2. If a pattern such as /(a)b|ac/ is matched against "ac", there is no
250 captured substring, but while checking the failing first alternative,
251 substring 1 is temporarily captured. If the output vector supplied to
252 pcre_exec() was not big enough for this capture, the yield of the function
253 was still zero ("insufficient space for captured substrings"). This cannot
254 be totally fixed without adding another stack variable, which seems a lot
255 of expense for a edge case. However, I have improved the situation in cases
256 such as /(a)(b)x|abc/ matched against "abc", where the return code
257 indicates that fewer than the maximum number of slots in the ovector have
258 been set.
259
260 3. Related to (2) above: when there are more back references in a pattern than
261 slots in the output vector, pcre_exec() uses temporary memory during
262 matching, and copies in the captures as far as possible afterwards. It was
263 using the entire output vector, but this conflicts with the specification
264 that only 2/3 is used for passing back captured substrings. Now it uses
265 only the first 2/3, for compatibility. This is, of course, another edge
266 case.
267
268 4. Zoltan Herczeg's just-in-time compiler support has been integrated into the
269 main code base, and can be used by building with --enable-jit. When this is
270 done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
271 runtime --no-jit option is given.
272
273 5. When the number of matches in a pcre_dfa_exec() run exactly filled the
274 ovector, the return from the function was zero, implying that there were
275 other matches that did not fit. The correct "exactly full" value is now
276 returned.
277
278 6. If a subpattern that was called recursively or as a subroutine contained
279 (*PRUNE) or any other control that caused it to give a non-standard return,
280 invalid errors such as "Error -26 (nested recursion at the same subject
281 position)" or even infinite loops could occur.
282
283 7. If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
284 computing the minimum length on reaching *ACCEPT, and so ended up with the
285 wrong value of 1 rather than 0. Further investigation indicates that
286 computing a minimum subject length in the presence of *ACCEPT is difficult
287 (think back references, subroutine calls), and so I have changed the code
288 so that no minimum is registered for a pattern that contains *ACCEPT.
289
290 8. If (*THEN) was present in the first (true) branch of a conditional group,
291 it was not handled as intended. [But see 16 below.]
292
293 9. Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
294 Sheri Pierce.
295
296 10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
297 the first byte in a match must be "a".
298
299 11. Change 17 for 8.13 increased the recursion depth for patterns like
300 /a(?:.)*?a/ drastically. I've improved things by remembering whether a
301 pattern contains any instances of (*THEN). If it does not, the old
302 optimizations are restored. It would be nice to do this on a per-group
303 basis, but at the moment that is not feasible.
304
305 12. In some environments, the output of pcretest -C is CRLF terminated. This
306 broke RunTest's code that checks for the link size. A single white space
307 character after the value is now allowed for.
308
309 13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
310 For "fr", it uses the Windows-specific input and output files.
311
312 14. If (*THEN) appeared in a group that was called recursively or as a
313 subroutine, it did not work as intended. [But see next item.]
314
315 15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
316 pattern fragments (but not containing any | characters). If A and B are
317 matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
318 was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
319 D. In other words, Perl considers parentheses that do not contain any |
320 characters to be part of a surrounding alternative, whereas PCRE was
321 treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
322 differently. PCRE now behaves in the same way as Perl, except in the case
323 of subroutine/recursion calls such as (?1) which have in any case always
324 been different (but PCRE had them first :-).
325
326 16. Related to 15 above: Perl does not treat the | in a conditional group as
327 creating alternatives. Such a group is treated in the same way as an
328 ordinary group without any | characters when processing (*THEN). PCRE has
329 been changed to match Perl's behaviour.
330
331 17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
332 RunGrepTest script failed.
333
334 18. Change 22 for version 13 caused atomic groups to use more stack. This is
335 inevitable for groups that contain captures, but it can lead to a lot of
336 stack use in large patterns. The old behaviour has been restored for atomic
337 groups that do not contain any capturing parentheses.
338
339 19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
340 suppress the check for a minimum subject length at run time. (If it was
341 given to pcre_exec() or pcre_dfa_exec() it did work.)
342
343 20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
344 fail to work when decoding hex characters in data strings in EBCDIC
345 environments.
346
347 21. It appears that in at least one Mac OS environment, the isxdigit() function
348 is implemented as a macro that evaluates to its argument more than once,
349 contravening the C 90 Standard (I haven't checked a later standard). There
350 was an instance in pcretest which caused it to go wrong when processing
351 \x{...} escapes in subject strings. The has been rewritten to avoid using
352 things like p++ in the argument of isxdigit().
353
354
355 Version 8.13 16-Aug-2011
356 ------------------------
357
358 1. The Unicode data tables have been updated to Unicode 6.0.0.
359
360 2. Two minor typos in pcre_internal.h have been fixed.
361
362 3. Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
363 pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
364 in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
365
366 4. There were a number of related bugs in the code for matching backrefences
367 caselessly in UTF-8 mode when codes for the characters concerned were
368 different numbers of bytes. For example, U+023A and U+2C65 are an upper
369 and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
370 (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
371 code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
372 2-byte code at the end of the subject (it thought there wasn't enough data
373 left).
374
375 5. Comprehensive information about what went wrong is now returned by
376 pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
377 as the output vector has at least 2 elements. The offset of the start of
378 the failing character and a reason code are placed in the vector.
379
380 6. When the UTF-8 string check fails for pcre_compile(), the offset that is
381 now returned is for the first byte of the failing character, instead of the
382 last byte inspected. This is an incompatible change, but I hope it is small
383 enough not to be a problem. It makes the returned offset consistent with
384 pcre_exec() and pcre_dfa_exec().
385
386 7. pcretest now gives a text phrase as well as the error number when
387 pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
388 failure, the offset and reason code are output.
389
390 8. When \R was used with a maximizing quantifier it failed to skip backwards
391 over a \r\n pair if the subsequent match failed. Instead, it just skipped
392 back over a single character (\n). This seems wrong (because it treated the
393 two characters as a single entity when going forwards), conflicts with the
394 documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
395 behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
396 has been changed.
397
398 9. Some internal refactoring has changed the processing so that the handling
399 of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
400 time (the PCRE_DOTALL option was changed this way some time ago: version
401 7.7 change 16). This has made it possible to abolish the OP_OPT op code,
402 which was always a bit of a fudge. It also means that there is one less
403 argument for the match() function, which reduces its stack requirements
404 slightly. This change also fixes an incompatibility with Perl: the pattern
405 (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
406
407 10. More internal refactoring has drastically reduced the number of recursive
408 calls to match() for possessively repeated groups such as (abc)++ when
409 using pcre_exec().
410
411 11. While implementing 10, a number of bugs in the handling of groups were
412 discovered and fixed:
413
414 (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
415 (a|)*(?1) gave a compile-time internal error.
416 ((a|)+)+ did not notice that the outer group could match an empty string.
417 (^a|^)+ was not marked as anchored.
418 (.*a|.*)+ was not marked as matching at start or after a newline.
419
420 12. Yet more internal refactoring has removed another argument from the match()
421 function. Special calls to this function are now indicated by setting a
422 value in a variable in the "match data" data block.
423
424 13. Be more explicit in pcre_study() instead of relying on "default" for
425 opcodes that mean there is no starting character; this means that when new
426 ones are added and accidentally left out of pcre_study(), testing should
427 pick them up.
428
429 14. The -s option of pcretest has been documented for ages as being an old
430 synonym of -m (show memory usage). I have changed it to mean "force study
431 for every regex", that is, assume /S for every regex. This is similar to -i
432 and -d etc. It's slightly incompatible, but I'm hoping nobody is still
433 using it. It makes it easier to run collections of tests with and without
434 study enabled, and thereby test pcre_study() more easily. All the standard
435 tests are now run with and without -s (but some patterns can be marked as
436 "never study" - see 20 below).
437
438 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
439 restoration of the capturing data to the outer values was not happening
440 correctly.
441
442 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
443 empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
444 pattern had matched an empty string, and so incorrectly returned a no
445 match.
446
447 17. There was optimizing code for the last branch of non-capturing parentheses,
448 and also for the obeyed branch of a conditional subexpression, which used
449 tail recursion to cut down on stack usage. Unfortunately, now that there is
450 the possibility of (*THEN) occurring in these branches, tail recursion is
451 no longer possible because the return has to be checked for (*THEN). These
452 two optimizations have therefore been removed. [But see 8.20/11 above.]
453
454 18. If a pattern containing \R was studied, it was assumed that \R always
455 matched two bytes, thus causing the minimum subject length to be
456 incorrectly computed because \R can also match just one byte.
457
458 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
459 was incorrectly computed.
460
461 20. If /S is present twice on a test pattern in pcretest input, it now
462 *disables* studying, thereby overriding the use of -s on the command line
463 (see 14 above). This is necessary for one or two tests to keep the output
464 identical in both cases.
465
466 21. When (*ACCEPT) was used in an assertion that matched an empty string and
467 PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
468
469 22. When an atomic group that contained a capturing parenthesis was
470 successfully matched, but the branch in which it appeared failed, the
471 capturing was not being forgotten if a higher numbered group was later
472 captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
473 group 1 to "a", when in fact it should be unset. This applied to multi-
474 branched capturing and non-capturing groups, repeated or not, and also to
475 positive assertions (capturing in negative assertions does not happen
476 in PCRE) and also to nested atomic groups.
477
478 23. Add the ++ qualifier feature to pcretest, to show the remainder of the
479 subject after a captured substring, to make it easier to tell which of a
480 number of identical substrings has been captured.
481
482 24. The way atomic groups are processed by pcre_exec() has been changed so that
483 if they are repeated, backtracking one repetition now resets captured
484 values correctly. For example, if ((?>(a+)b)+aabab) is matched against
485 "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
486 "aaa". Previously, it would have been "a". As part of this code
487 refactoring, the way recursive calls are handled has also been changed.
488
489 25. If an assertion condition captured any substrings, they were not passed
490 back unless some other capturing happened later. For example, if
491 (?(?=(a))a) was matched against "a", no capturing was returned.
492
493 26. When studying a pattern that contained subroutine calls or assertions,
494 the code for finding the minimum length of a possible match was handling
495 direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
496 group 1 called group 2 while simultaneously a separate group 2 called group
497 1). A stack overflow occurred in this case. I have fixed this by limiting
498 the recursion depth to 10.
499
500 27. Updated RunTest.bat in the distribution to the version supplied by Tom
501 Fortmann. This supports explicit test numbers on the command line, and has
502 argument validation and error reporting.
503
504 28. An instance of \X with an unlimited repeat could fail if at any point the
505 first character it looked at was a mark character.
506
507 29. Some minor code refactoring concerning Unicode properties and scripts
508 should reduce the stack requirement of match() slightly.
509
510 30. Added the '=' option to pcretest to check the setting of unused capturing
511 slots at the end of the pattern, which are documented as being -1, but are
512 not included in the return count.
513
514 31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
515 compiled something random. Now it gives a compile-time error (as does
516 Perl).
517
518 32. A *MARK encountered during the processing of a positive assertion is now
519 recorded and passed back (compatible with Perl).
520
521 33. If --only-matching or --colour was set on a pcregrep call whose pattern
522 had alternative anchored branches, the search for a second match in a line
523 was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
524 matched the line "0102" twice. The same bug affected patterns that started
525 with a backwards assertion. For example /\b01|\b02/ also matched "0102"
526 twice.
527
528 34. Previously, PCRE did not allow quantification of assertions. However, Perl
529 does, and because of capturing effects, quantifying parenthesized
530 assertions may at times be useful. Quantifiers are now allowed for
531 parenthesized assertions.
532
533 35. A minor code tidy in pcre_compile() when checking options for \R usage.
534
535 36. \g was being checked for fancy things in a character class, when it should
536 just be a literal "g".
537
538 37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
539 appearance of a nested POSIX class supersedes an apparent external class.
540 For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
541 unescaped square brackets may also appear as part of class names. For
542 example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
543 more like Perl. (But see 8.20/1 above.)
544
545 38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
546 was because it thought it was \N{name}, which is not supported).
547
548 39. Add minix to OS list not supporting the -S option in pcretest.
549
550 40. PCRE tries to detect cases of infinite recursion at compile time, but it
551 cannot analyze patterns in sufficient detail to catch mutual recursions
552 such as ((?1))((?2)). There is now a runtime test that gives an error if a
553 subgroup is called recursively as a subpattern for a second time at the
554 same position in the subject string. In previous releases this might have
555 been caught by the recursion limit, or it might have run out of stack.
556
557 41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
558 happen only once. PCRE was, however incorrectly giving a compile time error
559 "recursive call could loop indefinitely" because it cannot analyze the
560 pattern in sufficient detail. The compile time test no longer happens when
561 PCRE is compiling a conditional subpattern, but actual runaway loops are
562 now caught at runtime (see 40 above).
563
564 42. It seems that Perl allows any characters other than a closing parenthesis
565 to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
566 has been changed to be the same.
567
568 43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
569 as not to get warnings when autogen.sh is called. Also changed
570 AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
571
572 44. To help people who use pcregrep to scan files containing exceedingly long
573 lines, the following changes have been made:
574
575 (a) The default value of the buffer size parameter has been increased from
576 8K to 20K. (The actual buffer used is three times this size.)
577
578 (b) The default can be changed by ./configure --with-pcregrep-bufsize when
579 PCRE is built.
580
581 (c) A --buffer-size=n option has been added to pcregrep, to allow the size
582 to be set at run time.
583
584 (d) Numerical values in pcregrep options can be followed by K or M, for
585 example --buffer-size=50K.
586
587 (e) If a line being scanned overflows pcregrep's buffer, an error is now
588 given and the return code is set to 2.
589
590 45. Add a pointer to the latest mark to the callout data block.
591
592 46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
593 partial match of an empty string instead of no match. This was specific to
594 the use of ".".
595
596 47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
597 complete match instead of a partial match. This bug was dependent on both
598 the PCRE_UTF8 and PCRE_DOTALL options being set.
599
600 48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
601 starting byte set, because \b was not being ignored.
602
603
604 Version 8.12 15-Jan-2011
605 ------------------------
606
607 1. Fixed some typos in the markup of the man pages, and wrote a script that
608 checks for such things as part of the documentation building process.
609
610 2. On a big-endian 64-bit system, pcregrep did not correctly process the
611 --match-limit and --recursion-limit options (added for 8.11). In
612 particular, this made one of the standard tests fail. (The integer value
613 went into the wrong half of a long int.)
614
615 3. If the --colour option was given to pcregrep with -v (invert match), it
616 did strange things, either producing crazy output, or crashing. It should,
617 of course, ignore a request for colour when reporting lines that do not
618 match.
619
620 4. Another pcregrep bug caused similar problems if --colour was specified with
621 -M (multiline) and the pattern match finished with a line ending.
622
623 5. In pcregrep, when a pattern that ended with a literal newline sequence was
624 matched in multiline mode, the following line was shown as part of the
625 match. This seems wrong, so I have changed it.
626
627 6. Another pcregrep bug in multiline mode, when --colour was specified, caused
628 the check for further matches in the same line (so they could be coloured)
629 to overrun the end of the current line. If another match was found, it was
630 incorrectly shown (and then shown again when found in the next line).
631
632 7. If pcregrep was compiled under Windows, there was a reference to the
633 function pcregrep_exit() before it was defined. I am assuming this was
634 the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
635 reported by a user. I've moved the definition above the reference.
636
637
638 Version 8.11 10-Dec-2010
639 ------------------------
640
641 1. (*THEN) was not working properly if there were untried alternatives prior
642 to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
643 backtracked to try for "b" instead of moving to the next alternative branch
644 at the same level (in this case, to look for "c"). The Perl documentation
645 is clear that when (*THEN) is backtracked onto, it goes to the "next
646 alternative in the innermost enclosing group".
647
648 2. (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
649 such as (A(*COMMIT)B(*THEN)C|D) any failure after matching A should
650 result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
651 (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
652 (*THEN).
653
654 3. If \s appeared in a character class, it removed the VT character from
655 the class, even if it had been included by some previous item, for example
656 in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
657 of \s, but is part of the POSIX "space" class.)
658
659 4. A partial match never returns an empty string (because you can always
660 match an empty string at the end of the subject); however the checking for
661 an empty string was starting at the "start of match" point. This has been
662 changed to the "earliest inspected character" point, because the returned
663 data for a partial match starts at this character. This means that, for
664 example, /(?<=abc)def/ gives a partial match for the subject "abc"
665 (previously it gave "no match").
666
667 5. Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
668 of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
669 previously a full match would be given. However, setting PCRE_PARTIAL_HARD
670 has an implication that the given string is incomplete (because a partial
671 match is preferred over a full match). For this reason, these items now
672 give a partial match in this situation. [Aside: previously, the one case
673 /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
674 match rather than a full match, which was wrong by the old rules, but is
675 now correct.]
676
677 6. There was a bug in the handling of #-introduced comments, recognized when
678 PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
679 If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
680 UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
681 scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
682 but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
683 places in pcre_compile().
684
685 7. Related to (6) above, when pcre_compile() was skipping #-introduced
686 comments when looking ahead for named forward references to subpatterns,
687 the only newline sequence it recognized was NL. It now handles newlines
688 according to the set newline convention.
689
690 8. SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
691 former, but used strtoul(), whereas pcretest avoided strtoul() but did not
692 cater for a lack of strerror(). These oversights have been fixed.
693
694 9. Added --match-limit and --recursion-limit to pcregrep.
695
696 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
697
698 11. When the -o option was used, pcregrep was setting a return code of 1, even
699 when matches were found, and --line-buffered was not being honoured.
700
701 12. Added an optional parentheses number to the -o and --only-matching options
702 of pcregrep.
703
704 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
705 can match an empty string. The code to do it in pcretest and pcredemo
706 needed fixing:
707
708 (a) When the newline convention was "crlf", pcretest got it wrong, skipping
709 only one byte after an empty string match just before CRLF (this case
710 just got forgotten; "any" and "anycrlf" were OK).
711
712 (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
713 mode when an empty string match preceded an ASCII character followed by
714 a non-ASCII character. (The code for advancing by one character rather
715 than one byte was nonsense.)
716
717 (c) The pcredemo.c sample program did not have any code at all to handle
718 the cases when CRLF is a valid newline sequence.
719
720 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
721 as a starting offset was within the subject string. There is now a new
722 error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
723 negative or greater than the length of the string. In order to test this,
724 pcretest is extended to allow the setting of negative starting offsets.
725
726 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
727 starting offset points to the beginning of a UTF-8 character was
728 unnecessarily clumsy. I tidied it up.
729
730 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
731 bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
732
733 17. Nobody had reported that the --include_dir option, which was added in
734 release 7.7 should have been called --include-dir (hyphen, not underscore)
735 for compatibility with GNU grep. I have changed it to --include-dir, but
736 left --include_dir as an undocumented synonym, and the same for
737 --exclude-dir, though that is not available in GNU grep, at least as of
738 release 2.5.4.
739
740 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
741 characters from a string of bytes) have been redefined so as not to use
742 loops, in order to improve performance in some environments. At the same
743 time, I abstracted some of the common code into auxiliary macros to save
744 repetition (this should not affect the compiled code).
745
746 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
747 compile-time error is now given if \c is not followed by an ASCII
748 character, that is, a byte less than 128. (In EBCDIC mode, the code is
749 different, and any byte value is allowed.)
750
751 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
752 START_OPTIMIZE option, which is now allowed at compile time - but just
753 passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
754 to pcregrep and other applications that have no direct access to PCRE
755 options. The new /Y option in pcretest sets this option when calling
756 pcre_compile().
757
758 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
759 back references. Groups containing recursive back references were forced to
760 be atomic by that change, but in the case of named groups, the amount of
761 memory required was incorrectly computed, leading to "Failed: internal
762 error: code overflow". This has been fixed.
763
764 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
765 pcretest.c, to avoid build problems in some Borland environments.
766
767
768 Version 8.10 25-Jun-2010
769 ------------------------
770
771 1. Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
772 THEN.
773
774 2. (*ACCEPT) was not working when inside an atomic group.
775
776 3. Inside a character class, \B is treated as a literal by default, but
777 faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
778 causes the error). The code is unchanged, but I tidied the documentation.
779
780 4. Inside a character class, PCRE always treated \R and \X as literals,
781 whereas Perl faults them if its -w option is set. I have changed PCRE so
782 that it faults them when PCRE_EXTRA is set.
783
784 5. Added support for \N, which always matches any character other than
785 newline. (It is the same as "." when PCRE_DOTALL is not set.)
786
787 6. When compiling pcregrep with newer versions of gcc which may have
788 FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
789 declared with attribute warn_unused_result" were given. Just casting the
790 result to (void) does not stop the warnings; a more elaborate fudge is
791 needed. I've used a macro to implement this.
792
793 7. Minor change to pcretest.c to avoid a compiler warning.
794
795 8. Added four artifical Unicode properties to help with an option to make
796 \s etc use properties (see next item). The new properties are: Xan
797 (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
798
799 9. Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
800 use Unicode properties. (*UCP) at the start of a pattern can be used to set
801 this option. Modified pcretest to add /W to test this facility. Added
802 REG_UCP to make it available via the POSIX interface.
803
804 10. Added --line-buffered to pcregrep.
805
806 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
807 studied, and the match started with a letter with a code point greater than
808 127 whose first byte was different to the first byte of the other case of
809 the letter, the other case of this starting letter was not recognized
810 (#976).
811
812 12. If a pattern that was studied started with a repeated Unicode property
813 test, for example, \p{Nd}+, there was the theoretical possibility of
814 setting up an incorrect bitmap of starting bytes, but fortunately it could
815 not have actually happened in practice until change 8 above was made (it
816 added property types that matched character-matching opcodes).
817
818 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
819 possible starting bytes for non-anchored patterns.
820
821 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
822 \R, and also a number of cases that involve Unicode properties, both
823 explicit and implicit when PCRE_UCP is set.
824
825 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
826 input, it could crash or give wrong results if characters with values
827 greater than 0xc0 were present in the subject string. (Detail: it assumed
828 UTF-8 input when processing these items.)
829
830 16. Added a lot of (int) casts to avoid compiler warnings in systems where
831 size_t is 64-bit (#991).
832
833 17. Added a check for running out of memory when PCRE is compiled with
834 --disable-stack-for-recursion (#990).
835
836 18. If the last data line in a file for pcretest does not have a newline on
837 the end, a newline was missing in the output.
838
839 19. The default pcre_chartables.c file recognizes only ASCII characters (values
840 less than 128) in its various bitmaps. However, there is a facility for
841 generating tables according to the current locale when PCRE is compiled. It
842 turns out that in some environments, 0x85 and 0xa0, which are Unicode space
843 characters, are recognized by isspace() and therefore were getting set in
844 these tables, and indeed these tables seem to approximate to ISO 8859. This
845 caused a problem in UTF-8 mode when pcre_study() was used to create a list
846 of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
847 which of course cannot start UTF-8 characters. I have changed the code so
848 that only real ASCII characters (less than 128) and the correct starting
849 bytes for UTF-8 encodings are set for characters greater than 127 when in
850 UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
851 altogether.)
852
853 20. Added the /T option to pcretest so as to be able to run tests with non-
854 standard character tables, thus making it possible to include the tests
855 used for 19 above in the standard set of tests.
856
857 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
858 reference to a subpattern the other side of a comment that contains an
859 opening parenthesis caused either an internal compiling error, or a
860 reference to the wrong subpattern.
861
862
863 Version 8.02 19-Mar-2010
864 ------------------------
865
866 1. The Unicode data tables have been updated to Unicode 5.2.0.
867
868 2. Added the option --libs-cpp to pcre-config, but only when C++ support is
869 configured.
870
871 3. Updated the licensing terms in the pcregexp.pas file, as agreed with the
872 original author of that file, following a query about its status.
873
874 4. On systems that do not have stdint.h (e.g. Solaris), check for and include
875 inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
876
877 5. A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
878 quantifier applied to a forward-referencing subroutine call, could compile
879 incorrect code or give the error "internal error: previously-checked
880 referenced subpattern not found".
881
882 6. Both MS Visual Studio and Symbian OS have problems with initializing
883 variables to point to external functions. For these systems, therefore,
884 pcre_malloc etc. are now initialized to local functions that call the
885 relevant global functions.
886
887 7. There were two entries missing in the vectors called coptable and poptable
888 in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
889 I've fixed the data, and added a kludgy way of testing at compile time that
890 the lengths are correct (equal to the number of opcodes).
891
892 8. Following on from 7, I added a similar kludge to check the length of the
893 eint vector in pcreposix.c.
894
895 9. Error texts for pcre_compile() are held as one long string to avoid too
896 much relocation at load time. To find a text, the string is searched,
897 counting zeros. There was no check for running off the end of the string,
898 which could happen if a new error number was added without updating the
899 string.
900
901 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
902
903 11. \K was not working if it appeared in an atomic group or in a group that
904 was called as a "subroutine", or in an assertion. Perl 5.11 documents that
905 \K is "not well defined" if used in an assertion. PCRE now accepts it if
906 the assertion is positive, but not if it is negative.
907
908 12. Change 11 fortuitously reduced the size of the stack frame used in the
909 "match()" function of pcre_exec.c by one pointer. Forthcoming
910 implementation of support for (*MARK) will need an extra pointer on the
911 stack; I have reserved it now, so that the stack frame size does not
912 decrease.
913
914 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
915 item in branch that calls a recursion is a subroutine call - as in the
916 second branch in the above example - was incorrectly given the compile-
917 time error "recursive call could loop indefinitely" because pcre_compile()
918 was not correctly checking the subroutine for matching a non-empty string.
919
920 14. The checks for overrunning compiling workspace could trigger after an
921 overrun had occurred. This is a "should never occur" error, but it can be
922 triggered by pathological patterns such as hundreds of nested parentheses.
923 The checks now trigger 100 bytes before the end of the workspace.
924
925 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
926
927
928 Version 8.01 19-Jan-2010
929 ------------------------
930
931 1. If a pattern contained a conditional subpattern with only one branch (in
932 particular, this includes all (*DEFINE) patterns), a call to pcre_study()
933 computed the wrong minimum data length (which is of course zero for such
934 subpatterns). This could cause incorrect "no match" results.
935
936 2. For patterns such as (?i)a(?-i)b|c where an option setting at the start of
937 the pattern is reset in the first branch, pcre_compile() failed with
938 "internal error: code overflow at offset...". This happened only when
939 the reset was to the original external option setting. (An optimization
940 abstracts leading options settings into an external setting, which was the
941 cause of this.)
942
943 3. A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
944 of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
945 assertion pattern did not match (meaning that the assertion was true), it
946 was incorrectly treated as false if the SKIP had been reached during the
947 matching. This also applied to assertions used as conditions.
948
949 4. If an item that is not supported by pcre_dfa_exec() was encountered in an
950 assertion subpattern, including such a pattern used as a condition,
951 unpredictable results occurred, instead of the error return
952 PCRE_ERROR_DFA_UITEM.
953
954 5. The C++ GlobalReplace function was not working like Perl for the special
955 situation when an empty string is matched. It now does the fancy magic
956 stuff that is necessary.
957
958 6. In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
959 removed. (These were left over from very, very early versions of PCRE.)
960
961 7. Some cosmetic changes to the code to make life easier when compiling it
962 as part of something else:
963
964 (a) Change DEBUG to PCRE_DEBUG.
965
966 (b) In pcre_compile(), rename the member of the "branch_chain" structure
967 called "current" as "current_branch", to prevent a collision with the
968 Linux macro when compiled as a kernel module.
969
970 (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
971 prevent a collision with the Linux macro when compiled as a kernel
972 module.
973
974 8. In pcre_compile() there are some checks for integer overflows that used to
975 cast potentially large values to (double). This has been changed to that
976 when building, a check for int64_t is made, and if it is found, it is used
977 instead, thus avoiding the use of floating point arithmetic. (There is no
978 other use of FP in PCRE.) If int64_t is not found, the fallback is to
979 double.
980
981 9. Added two casts to avoid signed/unsigned warnings from VS Studio Express
982 2005 (difference between two addresses compared to an unsigned value).
983
984 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
985 custom one, because of the following reported problem in Windows:
986
987 - libbz2 uses the Pascal calling convention (WINAPI) for the functions
988 under Win32.
989 - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
990 therefore missing the function definition.
991 - The compiler thus generates a "C" signature for the test function.
992 - The linker fails to find the "C" function.
993 - PCRE fails to configure if asked to do so against libbz2.
994
995 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
996 messages were output:
997
998 Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
999 rerunning libtoolize, to keep the correct libtool macros in-tree.
1000 Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
1001
1002 I have done both of these things.
1003
1004 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
1005 most of the time, it *can* run out if it is given a pattern that contains a
1006 runaway infinite recursion. I updated the discussion in the pcrestack man
1007 page.
1008
1009 13. Now that we have gone to the x.xx style of version numbers, the minor
1010 version may start with zero. Using 08 or 09 is a bad idea because users
1011 might check the value of PCRE_MINOR in their code, and 08 or 09 may be
1012 interpreted as invalid octal numbers. I've updated the previous comment in
1013 configure.ac, and also added a check that gives an error if 08 or 09 are
1014 used.
1015
1016 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
1017 causing partial matching to fail when the end of the subject matched \W
1018 in a UTF-8 pattern where \W was quantified with a minimum of 3.
1019
1020 15. There were some discrepancies between the declarations in pcre_internal.h
1021 of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
1022 their definitions. The declarations used "const uschar *" and the
1023 definitions used USPTR. Even though USPTR is normally defined as "const
1024 unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
1025 reported that: "This difference in casting confuses some C++ compilers, for
1026 example, SunCC recognizes above declarations as different functions and
1027 generates broken code for hbpcre." I have changed the declarations to use
1028 USPTR.
1029
1030 16. GNU libtool is named differently on some systems. The autogen.sh script now
1031 tries several variants such as glibtoolize (MacOSX) and libtoolize1x
1032 (FreeBSD).
1033
1034 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
1035 (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
1036 comment: "Figure out how to create a longlong from a string: strtoll and
1037 equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
1038 instance, but it only takes 2 args instead of 3!"
1039
1040 18. A subtle bug concerned with back references has been fixed by a change of
1041 specification, with a corresponding code fix. A pattern such as
1042 ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
1043 refers, was giving matches when it shouldn't. For example, xa=xaaa would
1044 match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
1045 same bug. Such groups have to be quantified to be useful, or contained
1046 inside another quantified group. (If there's no repetition, the reference
1047 can never match.) The problem arises because, having left the group and
1048 moved on to the rest of the pattern, a later failure that backtracks into
1049 the group uses the captured value from the final iteration of the group
1050 rather than the correct earlier one. I have fixed this in PCRE by forcing
1051 any group that contains a reference to itself to be an atomic group; that
1052 is, there cannot be any backtracking into it once it has completed. This is
1053 similar to recursive and subroutine calls.
1054
1055
1056 Version 8.00 19-Oct-09
1057 ----------------------
1058
1059 1. The table for translating pcre_compile() error codes into POSIX error codes
1060 was out-of-date, and there was no check on the pcre_compile() error code
1061 being within the table. This could lead to an OK return being given in
1062 error.
1063
1064 2. Changed the call to open a subject file in pcregrep from fopen(pathname,
1065 "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
1066 in a Windows environment.
1067
1068 3. The pcregrep --count option prints the count for each file even when it is
1069 zero, as does GNU grep. However, pcregrep was also printing all files when
1070 --files-with-matches was added. Now, when both options are given, it prints
1071 counts only for those files that have at least one match. (GNU grep just
1072 prints the file name in this circumstance, but including the count seems
1073 more useful - otherwise, why use --count?) Also ensured that the
1074 combination -clh just lists non-zero counts, with no names.
1075
1076 4. The long form of the pcregrep -F option was incorrectly implemented as
1077 --fixed_strings instead of --fixed-strings. This is an incompatible change,
1078 but it seems right to fix it, and I didn't think it was worth preserving
1079 the old behaviour.
1080
1081 5. The command line items --regex=pattern and --regexp=pattern were not
1082 recognized by pcregrep, which required --regex pattern or --regexp pattern
1083 (with a space rather than an '='). The man page documented the '=' forms,
1084 which are compatible with GNU grep; these now work.
1085
1086 6. No libpcreposix.pc file was created for pkg-config; there was just
1087 libpcre.pc and libpcrecpp.pc. The omission has been rectified.
1088
1089 7. Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
1090 when UCP support is not needed, by modifying the Python script that
1091 generates it from Unicode data files. This should not matter if the module
1092 is correctly used as a library, but I received one complaint about 50K of
1093 unwanted data. My guess is that the person linked everything into his
1094 program rather than using a library. Anyway, it does no harm.
1095
1096 8. A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
1097 was a minimum greater than 1 for a wide character in a possessive
1098 repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
1099 which had an unlimited repeat of a nested, fixed maximum repeat of a wide
1100 character. Chaos in the form of incorrect output or a compiling loop could
1101 result.
1102
1103 9. The restrictions on what a pattern can contain when partial matching is
1104 requested for pcre_exec() have been removed. All patterns can now be
1105 partially matched by this function. In addition, if there are at least two
1106 slots in the offset vector, the offset of the earliest inspected character
1107 for the match and the offset of the end of the subject are set in them when
1108 PCRE_ERROR_PARTIAL is returned.
1109
1110 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
1111 synonymous with PCRE_PARTIAL, for backwards compatibility, and
1112 PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
1113 and may be more useful for multi-segment matching.
1114
1115 11. Partial matching with pcre_exec() is now more intuitive. A partial match
1116 used to be given if ever the end of the subject was reached; now it is
1117 given only if matching could not proceed because another character was
1118 needed. This makes a difference in some odd cases such as Z(*FAIL) with the
1119 string "Z", which now yields "no match" instead of "partial match". In the
1120 case of pcre_dfa_exec(), "no match" is given if every matching path for the
1121 final character ended with (*FAIL).
1122
1123 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
1124 if the pattern had a "must contain" character that was already found in the
1125 earlier partial match, unless partial matching was again requested. For
1126 example, with the pattern /dog.(body)?/, the "must contain" character is
1127 "g". If the first part-match was for the string "dog", restarting with
1128 "sbody" failed. This bug has been fixed.
1129
1130 13. The string returned by pcre_dfa_exec() after a partial match has been
1131 changed so that it starts at the first inspected character rather than the
1132 first character of the match. This makes a difference only if the pattern
1133 starts with a lookbehind assertion or \b or \B (\K is not supported by
1134 pcre_dfa_exec()). It's an incompatible change, but it makes the two
1135 matching functions compatible, and I think it's the right thing to do.
1136
1137 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
1138 so that the demonstration program is easily available in environments where
1139 PCRE has not been installed from source.
1140
1141 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
1142 libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
1143 library.
1144
1145 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
1146 It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
1147 is not the first non-POSIX option to be added. Clearly some people find
1148 these options useful.
1149
1150 17. If a caller to the POSIX matching function regexec() passes a non-zero
1151 value for nmatch with a NULL value for pmatch, the value of
1152 nmatch is forced to zero.
1153
1154 18. RunGrepTest did not have a test for the availability of the -u option of
1155 the diff command, as RunTest does. It now checks in the same way as
1156 RunTest, and also checks for the -b option.
1157
1158 19. If an odd number of negated classes containing just a single character
1159 interposed, within parentheses, between a forward reference to a named
1160 subpattern and the definition of the subpattern, compilation crashed with
1161 an internal error, complaining that it could not find the referenced
1162 subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
1163 [The bug was that it was starting one character too far in when skipping
1164 over the character class, thus treating the ] as data rather than
1165 terminating the class. This meant it could skip too much.]
1166
1167 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
1168 /g option in pcretest when the pattern contains \K, which makes it possible
1169 to have an empty string match not at the start, even when the pattern is
1170 anchored. Updated pcretest and pcredemo to use this option.
1171
1172 21. If the maximum number of capturing subpatterns in a recursion was greater
1173 than the maximum at the outer level, the higher number was returned, but
1174 with unset values at the outer level. The correct (outer level) value is
1175 now given.
1176
1177 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
1178 PCRE did not set those parentheses (unlike Perl). I have now found a way to
1179 make it do so. The string so far is captured, making this feature
1180 compatible with Perl.
1181
1182 23. The tests have been re-organized, adding tests 11 and 12, to make it
1183 possible to check the Perl 5.10 features against Perl 5.10.
1184
1185 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
1186 pattern matches a fixed length string. PCRE did not allow this; now it
1187 does. Neither allows recursion.
1188
1189 25. I finally figured out how to implement a request to provide the minimum
1190 length of subject string that was needed in order to match a given pattern.
1191 (It was back references and recursion that I had previously got hung up
1192 on.) This code has now been added to pcre_study(); it finds a lower bound
1193 to the length of subject needed. It is not necessarily the greatest lower
1194 bound, but using it to avoid searching strings that are too short does give
1195 some useful speed-ups. The value is available to calling programs via
1196 pcre_fullinfo().
1197
1198 26. While implementing 25, I discovered to my embarrassment that pcretest had
1199 not been passing the result of pcre_study() to pcre_dfa_exec(), so the
1200 study optimizations had never been tested with that matching function.
1201 Oops. What is worse, even when it was passed study data, there was a bug in
1202 pcre_dfa_exec() that meant it never actually used it. Double oops. There
1203 were also very few tests of studied patterns with pcre_dfa_exec().
1204
1205 27. If (?| is used to create subpatterns with duplicate numbers, they are now
1206 allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
1207 on the other side of the coin, they are no longer allowed to have different
1208 names, because these cannot be distinguished in PCRE, and this has caused
1209 confusion. (This is a difference from Perl.)
1210
1211 28. When duplicate subpattern names are present (necessarily with different
1212 numbers, as required by 27 above), and a test is made by name in a
1213 conditional pattern, either for a subpattern having been matched, or for
1214 recursion in such a pattern, all the associated numbered subpatterns are
1215 tested, and the overall condition is true if the condition is true for any
1216 one of them. This is the way Perl works, and is also more like the way
1217 testing by number works.
1218
1219
1220 Version 7.9 11-Apr-09
1221 ---------------------
1222
1223 1. When building with support for bzlib/zlib (pcregrep) and/or readline
1224 (pcretest), all targets were linked against these libraries. This included
1225 libpcre, libpcreposix, and libpcrecpp, even though they do not use these
1226 libraries. This caused unwanted dependencies to be created. This problem
1227 has been fixed, and now only pcregrep is linked with bzlib/zlib and only
1228 pcretest is linked with readline.
1229
1230 2. The "typedef int BOOL" in pcre_internal.h that was included inside the
1231 "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
1232 moved outside it again, because FALSE and TRUE are already defined in AIX,
1233 but BOOL is not.
1234
1235 3. The pcre_config() function was treating the PCRE_MATCH_LIMIT and
1236 PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
1237
1238 4. The pcregrep documentation said spaces were inserted as well as colons (or
1239 hyphens) following file names and line numbers when outputting matching
1240 lines. This is not true; no spaces are inserted. I have also clarified the
1241 wording for the --colour (or --color) option.
1242
1243 5. In pcregrep, when --colour was used with -o, the list of matching strings
1244 was not coloured; this is different to GNU grep, so I have changed it to be
1245 the same.
1246
1247 6. When --colo(u)r was used in pcregrep, only the first matching substring in
1248 each matching line was coloured. Now it goes on to look for further matches
1249 of any of the test patterns, which is the same behaviour as GNU grep.
1250
1251 7. A pattern that could match an empty string could cause pcregrep to loop; it
1252 doesn't make sense to accept an empty string match in pcregrep, so I have
1253 locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
1254 seems to be how GNU grep behaves.
1255
1256 8. The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
1257 start or after a newline", because the conditional assertion was not being
1258 correctly handled. The rule now is that both the assertion and what follows
1259 in the first alternative must satisfy the test.
1260
1261 9. If auto-callout was enabled in a pattern with a conditional group whose
1262 condition was an assertion, PCRE could crash during matching, both with
1263 pcre_exec() and pcre_dfa_exec().
1264
1265 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
1266 used for matching.
1267
1268 11. Unicode property support in character classes was not working for
1269 characters (bytes) greater than 127 when not in UTF-8 mode.
1270
1271 12. Added the -M command line option to pcretest.
1272
1273 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
1274
1275 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
1276
1277 16. Added comments and documentation about mis-use of no_arg in the C++
1278 wrapper.
1279
1280 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
1281 from Martin Jerabek that uses macro names for all relevant character and
1282 string constants.
1283
1284 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
1285 SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
1286 SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
1287 these, but not everybody uses configure.
1288
1289 19. A conditional group that had only one branch was not being correctly
1290 recognized as an item that could match an empty string. This meant that an
1291 enclosing group might also not be so recognized, causing infinite looping
1292 (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
1293 with the subject "ab", where knowledge that the repeated group can match
1294 nothing is needed in order to break the loop.
1295
1296 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
1297 exec(), but without supplying a callout function, matching went wrong.
1298
1299 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
1300 leak if the size of the offset vector was greater than 30. When the vector
1301 is smaller, the saved offsets during recursion go onto a local stack
1302 vector, but for larger vectors malloc() is used. It was failing to free
1303 when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
1304 error, in fact).
1305
1306 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
1307 heapframe that is used only when UTF-8 support is enabled. This caused no
1308 problem, but was untidy.
1309
1310 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
1311 CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
1312 included within another project.
1313
1314 24. Steven Van Ingelgem's patches to add more options to the CMake support,
1315 slightly modified by me:
1316
1317 (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
1318 not building pcregrep.
1319
1320 (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
1321 if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
1322
1323 25. Forward references, both numeric and by name, in patterns that made use of
1324 duplicate group numbers, could behave incorrectly or give incorrect errors,
1325 because when scanning forward to find the reference group, PCRE was not
1326 taking into account the duplicate group numbers. A pattern such as
1327 ^X(?3)(a)(?|(b)|(q))(Y) is an example.
1328
1329 26. Changed a few more instances of "const unsigned char *" to USPTR, making
1330 the feature of a custom pointer more persuasive (as requested by a user).
1331
1332 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
1333 pcretest.c, inside #ifndefs, because it seems they are sometimes already
1334 pre-defined.
1335
1336 28. Added support for (*UTF8) at the start of a pattern.
1337
1338 29. Arrange for flags added by the "release type" setting in CMake to be shown
1339 in the configuration summary.
1340
1341
1342 Version 7.8 05-Sep-08
1343 ---------------------
1344
1345 1. Replaced UCP searching code with optimized version as implemented for Ad
1346 Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
1347 stage table and inline lookup instead of a function, giving speed ups of 2
1348 to 5 times on some simple patterns that I tested. Permission was given to
1349 distribute the MultiStage2.py script that generates the tables (it's not in
1350 the tarball, but is in the Subversion repository).
1351
1352 2. Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
1353 scripts.
1354
1355 3. Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
1356 a group with a zero qualifier. The result of the study could be incorrect,
1357 or the function might crash, depending on the pattern.
1358
1359 4. Caseless matching was not working for non-ASCII characters in back
1360 references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
1361 It now works when Unicode Property Support is available.
1362
1363 5. In pcretest, an escape such as \x{de} in the data was always generating
1364 a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
1365 non-UTF-8 mode. If the value is greater than 255, it gives a warning about
1366 truncation.
1367
1368 6. Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
1369
1370 7. Added two (int) casts to pcregrep when printing the difference of two
1371 pointers, in case they are 64-bit values.
1372
1373 8. Added comments about Mac OS X stack usage to the pcrestack man page and to
1374 test 2 if it fails.
1375
1376 9. Added PCRE_CALL_CONVENTION just before the names of all exported functions,
1377 and a #define of that name to empty if it is not externally set. This is to
1378 allow users of MSVC to set it if necessary.
1379
1380 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
1381 the convenience functions in the pcre_get.c source file.
1382
1383 11. An option change at the start of a pattern that had top-level alternatives
1384 could cause overwriting and/or a crash. This command provoked a crash in
1385 some environments:
1386
1387 printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
1388
1389 This potential security problem was recorded as CVE-2008-2371.
1390
1391 12. For a pattern where the match had to start at the beginning or immediately
1392 after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
1393 pcre_dfa_exec() could read past the end of the passed subject if there was
1394 no match. To help with detecting such bugs (e.g. with valgrind), I modified
1395 pcretest so that it places the subject at the end of its malloc-ed buffer.
1396
1397 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
1398 exec() might read past the end of the data buffer in UTF-8 mode.
1399
1400 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
1401 the data contained the byte 0x85 as part of a UTF-8 character within its
1402 first line. This applied both to normal and DFA matching.
1403
1404 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
1405 /^[^d]*?$/8 failed to match "abc".
1406
1407 16. Added a missing copyright notice to pcrecpp_internal.h.
1408
1409 17. Make it more clear in the documentation that values returned from
1410 pcre_exec() in ovector are byte offsets, not character counts.
1411
1412 18. Tidied a few places to stop certain compilers from issuing warnings.
1413
1414 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
1415 supplied by Stefan Weber. I made a further small update for 7.8 because
1416 there is a change of source arrangements: the pcre_searchfuncs.c module is
1417 replaced by pcre_ucd.c.
1418
1419
1420 Version 7.7 07-May-08
1421 ---------------------
1422
1423 1. Applied Craig's patch to sort out a long long problem: "If we can't convert
1424 a string to a long long, pretend we don't even have a long long." This is
1425 done by checking for the strtoq, strtoll, and _strtoi64 functions.
1426
1427 2. Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
1428 pre-7.6 versions, which defined a global no_arg variable instead of putting
1429 it in the RE class. (See also #8 below.)
1430
1431 3. Remove a line of dead code, identified by coverity and reported by Nuno
1432 Lopes.
1433
1434 4. Fixed two related pcregrep bugs involving -r with --include or --exclude:
1435
1436 (1) The include/exclude patterns were being applied to the whole pathnames
1437 of files, instead of just to the final components.
1438
1439 (2) If there was more than one level of directory, the subdirectories were
1440 skipped unless they satisfied the include/exclude conditions. This is
1441 inconsistent with GNU grep (and could even be seen as contrary to the
1442 pcregrep specification - which I improved to make it absolutely clear).
1443 The action now is always to scan all levels of directory, and just
1444 apply the include/exclude patterns to regular files.
1445
1446 5. Added the --include_dir and --exclude_dir patterns to pcregrep, and used
1447 --exclude_dir in the tests to avoid scanning .svn directories.
1448
1449 6. Applied Craig's patch to the QuoteMeta function so that it escapes the
1450 NUL character as backslash + 0 rather than backslash + NUL, because PCRE
1451 doesn't support NULs in patterns.
1452
1453 7. Added some missing "const"s to declarations of static tables in
1454 pcre_compile.c and pcre_dfa_exec.c.
1455
1456 8. Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
1457 caused by fix #2 above. (Subsequently also a second patch to fix the
1458 first patch. And a third patch - this was a messy problem.)
1459
1460 9. Applied Craig's patch to remove the use of push_back().
1461
1462 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
1463 matching function regexec().
1464
1465 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
1466 which, however, unlike Perl's \g{...}, are subroutine calls, not back
1467 references. PCRE supports relative numbers with this syntax (I don't think
1468 Oniguruma does).
1469
1470 12. Previously, a group with a zero repeat such as (...){0} was completely
1471 omitted from the compiled regex. However, this means that if the group
1472 was called as a subroutine from elsewhere in the pattern, things went wrong
1473 (an internal error was given). Such groups are now left in the compiled
1474 pattern, with a new opcode that causes them to be skipped at execution
1475 time.
1476
1477 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
1478 to the way PCRE behaves:
1479
1480 (a) A lone ] character is dis-allowed (Perl treats it as data).
1481
1482 (b) A back reference to an unmatched subpattern matches an empty string
1483 (Perl fails the current match path).
1484
1485 (c) A data ] in a character class must be notated as \] because if the
1486 first data character in a class is ], it defines an empty class. (In
1487 Perl it is not possible to have an empty class.) The empty class []
1488 never matches; it forces failure and is equivalent to (*FAIL) or (?!).
1489 The negative empty class [^] matches any one character, independently
1490 of the DOTALL setting.
1491
1492 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
1493 non-existent subpattern following a character class starting with ']' and
1494 containing () gave an internal compiling error instead of "reference to
1495 non-existent subpattern". Fortunately, when the pattern did exist, the
1496 compiled code was correct. (When scanning forwards to check for the
1497 existencd of the subpattern, it was treating the data ']' as terminating
1498 the class, so got the count wrong. When actually compiling, the reference
1499 was subsequently set up correctly.)
1500
1501 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
1502 it was being rejected as not supported by pcre_dfa_exec(), even though
1503 other assertions are supported. I have made pcre_dfa_exec() support
1504 (*FAIL).
1505
1506 16. The implementation of 13c above involved the invention of a new opcode,
1507 OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
1508 cannot be changed at match time, I realized I could make a small
1509 improvement to matching performance by compiling OP_ALLANY instead of
1510 OP_ANY for "." when DOTALL was set, and then removing the runtime tests
1511 on the OP_ANY path.
1512
1513 17. Compiling pcretest on Windows with readline support failed without the
1514 following two fixes: (1) Make the unistd.h include conditional on
1515 HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
1516
1517 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
1518 ncurses library to be included for pcretest when ReadLine support is
1519 requested, but also to allow for it to be overridden. This patch came from
1520 Daniel Bergström.
1521
1522 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
1523 as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
1524 any errors with the current Unicode tables. Thanks to Peter Kankowski for
1525 spotting this.
1526
1527
1528 Version 7.6 28-Jan-08
1529 ---------------------
1530
1531 1. A character class containing a very large number of characters with
1532 codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
1533 overflow.
1534
1535 2. Patch to cut out the "long long" test in pcrecpp_unittest when
1536 HAVE_LONG_LONG is not defined.
1537
1538 3. Applied Christian Ehrlicher's patch to update the CMake build files to
1539 bring them up to date and include new features. This patch includes:
1540
1541 - Fixed PH's badly added libz and libbz2 support.
1542 - Fixed a problem with static linking.
1543 - Added pcredemo. [But later removed - see 7 below.]
1544 - Fixed dftables problem and added an option.
1545 - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
1546 HAVE_LONG_LONG.
1547 - Added readline support for pcretest.
1548 - Added an listing of the option settings after cmake has run.
1549
1550 4. A user submitted a patch to Makefile that makes it easy to create
1551 "pcre.dll" under mingw when using Configure/Make. I added stuff to
1552 Makefile.am that cause it to include this special target, without
1553 affecting anything else. Note that the same mingw target plus all
1554 the other distribution libraries and programs are now supported
1555 when configuring with CMake (see 6 below) instead of with
1556 Configure/Make.
1557
1558 5. Applied Craig's patch that moves no_arg into the RE class in the C++ code.
1559 This is an attempt to solve the reported problem "pcrecpp::no_arg is not
1560 exported in the Windows port". It has not yet been confirmed that the patch
1561 solves the problem, but it does no harm.
1562
1563 6. Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
1564 NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
1565 with CMake, and also correct the comment about stack recursion.
1566
1567 7. Remove the automatic building of pcredemo from the ./configure system and
1568 from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
1569 of a program that users should build themselves after PCRE is installed, so
1570 building it automatically is not really right. What is more, it gave
1571 trouble in some build environments.
1572
1573 8. Further tidies to CMakeLists.txt from Sheri and Christian.
1574
1575
1576 Version 7.5 10-Jan-08
1577 ---------------------
1578
1579 1. Applied a patch from Craig: "This patch makes it possible to 'ignore'
1580 values in parens when parsing an RE using the C++ wrapper."
1581
1582 2. Negative specials like \S did not work in character classes in UTF-8 mode.
1583 Characters greater than 255 were excluded from the class instead of being
1584 included.
1585
1586 3. The same bug as (2) above applied to negated POSIX classes such as
1587 [:^space:].
1588
1589 4. PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
1590 defined or documented. It seems to have been a typo for PCRE_STATIC, so
1591 I have changed it.
1592
1593 5. The construct (?&) was not diagnosed as a syntax error (it referenced the
1594 first named subpattern) and a construct such as (?&a) would reference the
1595 first named subpattern whose name started with "a" (in other words, the
1596 length check was missing). Both these problems are fixed. "Subpattern name
1597 expected" is now given for (?&) (a zero-length name), and this patch also
1598 makes it give the same error for \k'' (previously it complained that that
1599 was a reference to a non-existent subpattern).
1600
1601 6. The erroneous patterns (?+-a) and (?-+a) give different error messages;
1602 this is right because (?- can be followed by option settings as well as by
1603 digits. I have, however, made the messages clearer.
1604
1605 7. Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
1606 than the number used in the conditional) now cause a compile-time error.
1607 This is actually not compatible with Perl, which accepts such patterns, but
1608 treats the conditional as always being FALSE (as PCRE used to), but it
1609 seems to me that giving a diagnostic is better.
1610
1611 8. Change "alphameric" to the more common word "alphanumeric" in comments
1612 and messages.
1613
1614 9. Fix two occurrences of "backslash" in comments that should have been
1615 "backspace".
1616
1617 10. Remove two redundant lines of code that can never be obeyed (their function
1618 was moved elsewhere).
1619
1620 11. The program that makes PCRE's Unicode character property table had a bug
1621 which caused it to generate incorrect table entries for sequences of
1622 characters that have the same character type, but are in different scripts.
1623 It amalgamated them into a single range, with the script of the first of
1624 them. In other words, some characters were in the wrong script. There were
1625 thirteen such cases, affecting characters in the following ranges:
1626
1627 U+002b0 - U+002c1
1628 U+0060c - U+0060d
1629 U+0061e - U+00612
1630 U+0064b - U+0065e
1631 U+0074d - U+0076d
1632 U+01800 - U+01805
1633 U+01d00 - U+01d77
1634 U+01d9b - U+01dbf
1635 U+0200b - U+0200f
1636 U+030fc - U+030fe
1637 U+03260 - U+0327f
1638 U+0fb46 - U+0fbb1
1639 U+10450 - U+1049d
1640
1641 12. The -o option (show only the matching part of a line) for pcregrep was not
1642 compatible with GNU grep in that, if there was more than one match in a
1643 line, it showed only the first of them. It now behaves in the same way as
1644 GNU grep.
1645
1646 13. If the -o and -v options were combined for pcregrep, it printed a blank
1647 line for every non-matching line. GNU grep prints nothing, and pcregrep now
1648 does the same. The return code can be used to tell if there were any
1649 non-matching lines.
1650
1651 14. Added --file-offsets and --line-offsets to pcregrep.
1652
1653 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
1654 infinitely looping recursion. The bug was that positive lookaheads were not
1655 being skipped when checking for a possible empty match (negative lookaheads
1656 and both kinds of lookbehind were skipped).
1657
1658 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
1659 inclusion of <windows.h> to before rather than after the definition of
1660 INVALID_FILE_ATTRIBUTES (patch from David Byron).
1661
1662 17. Specifying a possessive quantifier with a specific limit for a Unicode
1663 character property caused pcre_compile() to compile bad code, which led at
1664 runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
1665 are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
1666 caused the error; without that there was no problem.
1667
1668 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
1669
1670 19. Added --enable-pcretest-libreadline.
1671
1672 20. In pcrecpp.cc, the variable 'count' was incremented twice in
1673 RE::GlobalReplace(). As a result, the number of replacements returned was
1674 double what it should be. I removed one of the increments, but Craig sent a
1675 later patch that removed the other one (the right fix) and added unit tests
1676 that check the return values (which was not done before).
1677
1678 21. Several CMake things:
1679
1680 (1) Arranged that, when cmake is used on Unix, the libraries end up with
1681 the names libpcre and libpcreposix, not just pcre and pcreposix.
1682
1683 (2) The above change means that pcretest and pcregrep are now correctly
1684 linked with the newly-built libraries, not previously installed ones.
1685
1686 (3) Added PCRE_SUPPORT_LIBREADLINE, PCRE_SUPPORT_LIBZ, PCRE_SUPPORT_LIBBZ2.
1687
1688 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
1689 crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
1690 UTF-8 newline character). The key issue is that the pattern starts .*;
1691 this means that the match must be either at the beginning, or after a
1692 newline. The bug was in the code for advancing after a failed match and
1693 checking that the new position followed a newline. It was not taking
1694 account of UTF-8 characters correctly.
1695
1696 23. PCRE was behaving differently from Perl in the way it recognized POSIX
1697 character classes. PCRE was not treating the sequence [:...:] as a
1698 character class unless the ... were all letters. Perl, however, seems to
1699 allow any characters between [: and :], though of course it rejects as
1700 unknown any "names" that contain non-letters, because all the known class
1701 names consist only of letters. Thus, Perl gives an error for [[:1234:]],
1702 for example, whereas PCRE did not - it did not recognize a POSIX character
1703 class. This seemed a bit dangerous, so the code has been changed to be
1704 closer to Perl. The behaviour is not identical to Perl, because PCRE will
1705 diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
1706 treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
1707 Perl does, and where it didn't before.
1708
1709 24. Rewrite so as to remove the single use of %n from pcregrep because in some
1710 Windows environments %n is disabled by default.
1711
1712
1713 Version 7.4 21-Sep-07
1714 ---------------------
1715
1716 1. Change 7.3/28 was implemented for classes by looking at the bitmap. This
1717 means that a class such as [\s] counted as "explicit reference to CR or
1718 LF". That isn't really right - the whole point of the change was to try to
1719 help when there was an actual mention of one of the two characters. So now
1720 the change happens only if \r or \n (or a literal CR or LF) character is
1721 encountered.
1722
1723 2. The 32-bit options word was also used for 6 internal flags, but the numbers
1724 of both had grown to the point where there were only 3 bits left.
1725 Fortunately, there was spare space in the data structure, and so I have
1726 moved the internal flags into a new 16-bit field to free up more option
1727 bits.
1728
1729 3. The appearance of (?J) at the start of a pattern set the DUPNAMES option,
1730 but did not set the internal JCHANGED flag - either of these is enough to
1731 control the way the "get" function works - but the PCRE_INFO_JCHANGED
1732 facility is supposed to tell if (?J) was ever used, so now (?J) at the
1733 start sets both bits.
1734
1735 4. Added options (at build time, compile time, exec time) to change \R from
1736 matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
1737
1738 5. doc/pcresyntax.html was missing from the distribution.
1739
1740 6. Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
1741 compatibility, even though it is no longer used.
1742
1743 7. Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
1744 strtoull to pcrecpp.cc to select the available functions in WIN32 when the
1745 windows.h file is present (where different names are used). [This was
1746 reversed later after testing - see 16 below.]
1747
1748 8. Changed all #include <config.h> to #include "config.h". There were also
1749 some further <pcre.h> cases that I changed to "pcre.h".
1750
1751 9. When pcregrep was used with the --colour option, it missed the line ending
1752 sequence off the lines that it output.
1753
1754 10. It was pointed out to me that arrays of string pointers cause lots of
1755 relocations when a shared library is dynamically loaded. A technique of
1756 using a single long string with a table of offsets can drastically reduce
1757 these. I have refactored PCRE in four places to do this. The result is
1758 dramatic:
1759
1760 Originally: 290
1761 After changing UCP table: 187
1762 After changing error message table: 43
1763 After changing table of "verbs" 36
1764 After changing table of Posix names 22
1765
1766 Thanks to the folks working on Gregex for glib for this insight.
1767
1768 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
1769 unicode-properties was also set.
1770
1771 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
1772
1773 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
1774 checked only for CRLF.
1775
1776 14. Added casts to pcretest.c to avoid compiler warnings.
1777
1778 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
1779
1780 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
1781 and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
1782 entirely. This removes changes made in 7 above.
1783
1784 17. The CMake files have been updated, and there is now more information about
1785 building with CMake in the NON-UNIX-USE document.
1786
1787
1788 Version 7.3 28-Aug-07
1789 ---------------------
1790
1791 1. In the rejigging of the build system that eventually resulted in 7.1, the
1792 line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
1793 brackets there is not right, since it causes compilers to look for an
1794 installed pcre.h, not the version that is in the source that is being
1795 compiled (which of course may be different). I have changed it back to:
1796
1797 #include "pcre.h"
1798
1799 I have a vague recollection that the change was concerned with compiling in
1800 different directories, but in the new build system, that is taken care of
1801 by the VPATH setting the Makefile.
1802
1803 2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
1804 when the subject happened to end in the byte 0x85 (e.g. if the last
1805 character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
1806 characters but of course it shouldn't be taken as a newline when it is part
1807 of another character. The bug was that, for an unlimited repeat of . in
1808 not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
1809 characters when looking for a newline.
1810
1811 3. A small performance improvement in the DOTALL UTF-8 mode .* case.
1812
1813 4. Debugging: adjusted the names of opcodes for different kinds of parentheses
1814 in debug output.
1815
1816 5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
1817 long printing in the pcrecpp unittest when running under MinGW.
1818
1819 6. ESC_K was left out of the EBCDIC table.
1820
1821 7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
1822 parentheses; I made it 1000, which seemed large enough. Unfortunately, the
1823 limit also applies to "virtual nesting" when a pattern is recursive, and in
1824 this case 1000 isn't so big. I have been able to remove this limit at the
1825 expense of backing off one optimization in certain circumstances. Normally,
1826 when pcre_exec() would call its internal match() function recursively and
1827 immediately return the result unconditionally, it uses a "tail recursion"
1828 feature to save stack. However, when a subpattern that can match an empty
1829 string has an unlimited repetition quantifier, it no longer makes this
1830 optimization. That gives it a stack frame in which to save the data for
1831 checking that an empty string has been matched. Previously this was taken
1832 from the 1000-entry workspace that had been reserved. So now there is no
1833 explicit limit, but more stack is used.
1834
1835 8. Applied Daniel's patches to solve problems with the import/export magic
1836 syntax that is required for Windows, and which was going wrong for the
1837 pcreposix and pcrecpp parts of the library. These were overlooked when this
1838 problem was solved for the main library.
1839
1840 9. There were some crude static tests to avoid integer overflow when computing
1841 the size of patterns that contain repeated groups with explicit upper
1842 limits. As the maximum quantifier is 65535, the maximum group length was
1843 set at 30,000 so that the product of these two numbers did not overflow a
1844 32-bit integer. However, it turns out that people want to use groups that
1845 are longer than 30,000 bytes (though not repeat them that many times).
1846 Change 7.0/17 (the refactoring of the way the pattern size is computed) has
1847 made it possible to implement the integer overflow checks in a much more
1848 dynamic way, which I have now done. The artificial limitation on group
1849 length has been removed - we now have only the limit on the total length of
1850 the compiled pattern, which depends on the LINK_SIZE setting.
1851
1852 10. Fixed a bug in the documentation for get/copy named substring when
1853 duplicate names are permitted. If none of the named substrings are set, the
1854 functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
1855 empty string.
1856
1857 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
1858 instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
1859 because the ] is interpreted as the first data character and the
1860 terminating ] is not found. PCRE has been made compatible with Perl in this
1861 regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
1862 cause memory overwriting.
1863
1864 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
1865 string has been matched (to stop an infinite loop). It was not recognizing
1866 a conditional subpattern that could match an empty string if that
1867 subpattern was within another subpattern. For example, it looped when
1868 trying to match (((?(1)X|))*) but it was OK with ((?(1)X|)*) where the
1869 condition was not nested. This bug has been fixed.
1870
1871 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
1872 past the start of the subject in the presence of bytes with the top bit
1873 set, for example "\x8aBCD".
1874
1875 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
1876 (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
1877
1878 14. Optimized (?!) to (*FAIL).
1879
1880 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
1881 This restricts code points to be within the range 0 to 0x10FFFF, excluding
1882 the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
1883 full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
1884 does: it's just the validity check that is more restrictive.
1885
1886 16. Inserted checks for integer overflows during escape sequence (backslash)
1887 processing, and also fixed erroneous offset values for syntax errors during
1888 backslash processing.
1889
1890 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
1891 for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
1892
1893 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
1894 caused an overrun.
1895
1896 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
1897 something other than just ASCII characters) inside a group that had an
1898 unlimited repeat caused a loop at compile time (while checking to see
1899 whether the group could match an empty string).
1900
1901 20. Debugging a pattern containing \p or \P could cause a crash. For example,
1902 [\P{Any}] did so. (Error in the code for printing property names.)
1903
1904 21. An orphan \E inside a character class could cause a crash.
1905
1906 22. A repeated capturing bracket such as (A)? could cause a wild memory
1907 reference during compilation.
1908
1909 23. There are several functions in pcre_compile() that scan along a compiled
1910 expression for various reasons (e.g. to see if it's fixed length for look
1911 behind). There were bugs in these functions when a repeated \p or \P was
1912 present in the pattern. These operators have additional parameters compared
1913 with \d, etc, and these were not being taken into account when moving along
1914 the compiled data. Specifically:
1915
1916 (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
1917 length.
1918
1919 (b) An item such as \pL+ within a repeated group could cause crashes or
1920 loops.
1921
1922 (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
1923 "reference to non-existent subpattern" error.
1924
1925 (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
1926
1927 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
1928 characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
1929
1930 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
1931
1932 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
1933 character were causing crashes (broken optimization).
1934
1935 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
1936 \p or \P) caused a compile-time loop.
1937
1938 28. More problems have arisen in unanchored patterns when CRLF is a valid line
1939 break. For example, the unstudied pattern [\r\n]A does not match the string
1940 "\r\nA" because change 7.0/46 below moves the current point on by two
1941 characters after failing to match at the start. However, the pattern \nA
1942 *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
1943 the same is true. There doesn't seem any very clean way out of this, but
1944 what I have chosen to do makes the common cases work: PCRE now takes note
1945 of whether there can be an explicit match for \r or \n anywhere in the
1946 pattern, and if so, 7.0/46 no longer applies. As part of this change,
1947 there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
1948 pattern has explicit CR or LF references.
1949
1950 29. Added (*CR) etc for changing newline setting at start of pattern.
1951
1952
1953 Version 7.2 19-Jun-07
1954 ---------------------
1955
1956 1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
1957 which is apparently normally available under Windows.
1958
1959 2. Re-jig the pcregrep tests with different newline settings in an attempt
1960 to make them independent of the local environment's newline setting.
1961
1962 3. Add code to configure.ac to remove -g from the CFLAGS default settings.
1963
1964 4. Some of the "internals" tests were previously cut out when the link size
1965 was not 2, because the output contained actual offsets. The recent new
1966 "Z" feature of pcretest means that these can be cut out, making the tests
1967 usable with all link sizes.
1968
1969 5. Implemented Stan Switzer's goto replacement for longjmp() when not using
1970 stack recursion. This gives a massive performance boost under BSD, but just
1971 a small improvement under Linux. However, it saves one field in the frame
1972 in all cases.
1973
1974 6. Added more features from the forthcoming Perl 5.10:
1975
1976 (a) (?-n) (where n is a string of digits) is a relative subroutine or
1977 recursion call. It refers to the nth most recently opened parentheses.
1978
1979 (b) (?+n) is also a relative subroutine call; it refers to the nth next
1980 to be opened parentheses.
1981
1982 (c) Conditions that refer to capturing parentheses can be specified
1983 relatively, for example, (?(-2)... or (?(+3)...
1984
1985 (d) \K resets the start of the current match so that everything before
1986 is not part of it.
1987
1988 (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
1989
1990 (f) \g{name} is another synonym - part of Perl 5.10's unification of
1991 reference syntax.
1992
1993 (g) (?| introduces a group in which the numbering of parentheses in each
1994 alternative starts with the same number.
1995
1996 (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
1997
1998 7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
1999 PCRE_INFO_JCHANGED.
2000
2001 8. A pattern such as (.*(.)?)* caused pcre_exec() to fail by either not
2002 terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
2003 for detecting groups that can match an empty string.
2004
2005 9. A pattern with a very large number of alternatives (more than several
2006 hundred) was running out of internal workspace during the pre-compile
2007 phase, where pcre_compile() figures out how much memory will be needed. A
2008 bit of new cunning has reduced the workspace needed for groups with
2009 alternatives. The 1000-alternative test pattern now uses 12 bytes of
2010 workspace instead of running out of the 4096 that are available.
2011
2012 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
2013
2014 11. Applied patch from Google to remove an optimization that didn't quite work.
2015 The report of the bug said:
2016
2017 pcrecpp::RE("a*").FullMatch("aaa") matches, while
2018 pcrecpp::RE("a*?").FullMatch("aaa") does not, and
2019 pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
2020
2021 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
2022 it matched the wrong number of bytes.
2023
2024
2025 Version 7.1 24-Apr-07
2026 ---------------------
2027
2028 1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
2029 that is more "standard", making use of automake and other Autotools. There
2030 is some re-arrangement of the files and adjustment of comments consequent
2031 on this.
2032
2033 2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
2034 for recursive directory scanning broke on some systems because the files
2035 are not scanned in any specific order and on different systems the order
2036 was different. A call to "sort" has been inserted into RunGrepTest for the
2037 approprate test as a short-term fix. In the longer term there may be an
2038 alternative.
2039
2040 3. I had an email from Eric Raymond about problems translating some of PCRE's
2041 man pages to HTML (despite the fact that I distribute HTML pages, some
2042 people do their own conversions for various reasons). The problems
2043 concerned the use of low-level troff macros .br and .in. I have therefore
2044 removed all such uses from the man pages (some were redundant, some could
2045 be replaced by .nf/.fi pairs). The 132html script that I use to generate
2046 HTML has been updated to handle .nf/.fi and to complain if it encounters
2047 .br or .in.
2048
2049 4. Updated comments in configure.ac that get placed in config.h.in and also
2050 arranged for config.h to be included in the distribution, with the name
2051 config.h.generic, for the benefit of those who have to compile without
2052 Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
2053
2054 5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
2055 Weber: (1) pcre_internal.h was missing some function renames; (2) updated
2056 makevp.bat for the current PCRE, using the additional files
2057 makevp_c.txt, makevp_l.txt, and pcregexp.pas.
2058
2059 6. A Windows user reported a minor discrepancy with test 2, which turned out
2060 to be caused by a trailing space on an input line that had got lost in his
2061 copy. The trailing space was an accident, so I've just removed it.
2062
2063 7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
2064 that is needed.
2065
2066 8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
2067 as "const" (a) because they are and (b) because it helps the PHP
2068 maintainers who have recently made a script to detect big data structures
2069 in the php code that should be moved to the .rodata section. I remembered
2070 to update Builducptable as well, so it won't revert if ucptable.h is ever
2071 re-created.
2072
2073 9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
2074 pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
2075 order to be able to cut out the UTF-8 tables in the latter when UTF-8
2076 support is not required. This saves 1.5-2K of code, which is important in
2077 some applications.
2078
2079 Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
2080 so as not to refer to the tables, even though these functions will never be
2081 called when UTF-8 support is disabled. Otherwise there are problems with a
2082 shared library.
2083
2084 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
2085
2086 (a) It was defining its arguments as char * instead of void *.
2087
2088 (b) It was assuming that all moves were upwards in memory; this was true
2089 a long time ago when I wrote it, but is no longer the case.
2090
2091 The emulated memove() is provided for those environments that have neither
2092 memmove() nor bcopy(). I didn't think anyone used it these days, but that
2093 is clearly not the case, as these two bugs were recently reported.
2094
2095 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
2096 and Detrail to create the HTML documentation, the .txt form of the man
2097 pages, and it removes trailing spaces from listed files. It also creates
2098 pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
2099 case, it wraps all the #defines with #ifndefs. This script should be run
2100 before "make dist".
2101
2102 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
2103 with Unicode property support.
2104
2105 (a) For a maximizing quantifier, if the two different cases of the
2106 character were of different lengths in their UTF-8 codings (there are
2107 some cases like this - I found 11), and the matching function had to
2108 back up over a mixture of the two cases, it incorrectly assumed they
2109 were both the same length.
2110
2111 (b) When PCRE was configured to use the heap rather than the stack for
2112 recursion during matching, it was not correctly preserving the data for
2113 the other case of a UTF-8 character when checking ahead for a match
2114 while processing a minimizing repeat. If the check also involved
2115 matching a wide character, but failed, corruption could cause an
2116 erroneous result when trying to check for a repeat of the original
2117 character.
2118
2119 13. Some tidying changes to the testing mechanism:
2120
2121 (a) The RunTest script now detects the internal link size and whether there
2122 is UTF-8 and UCP support by running ./pcretest -C instead of relying on
2123 values substituted by "configure". (The RunGrepTest script already did
2124 this for UTF-8.) The configure.ac script no longer substitutes the
2125 relevant variables.
2126
2127 (b) The debugging options /B and /D in pcretest show the compiled bytecode
2128 with length and offset values. This means that the output is different
2129 for different internal link sizes. Test 2 is skipped for link sizes
2130 other than 2 because of this, bypassing the problem. Unfortunately,
2131 there was also a test in test 3 (the locale tests) that used /B and
2132 failed for link sizes other than 2. Rather than cut the whole test out,
2133 I have added a new /Z option to pcretest that replaces the length and
2134 offset values with spaces. This is now used to make test 3 independent
2135 of link size. (Test 2 will be tidied up later.)
2136
2137 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
2138 segmentation fault instead of returning the appropriate error message.
2139
2140 15. In multiline mode when the newline sequence was set to "any", the pattern
2141 ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
2142 This doesn't seem right; it now treats the CRLF combination as the line
2143 ending, and so does not match in that case. It's only a pattern such as ^$
2144 that would hit this one: something like ^ABC$ would have failed after \r
2145 and then tried again after \r\n.
2146
2147 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
2148 in an attempt to make files that differ only in their line terminators
2149 compare equal. This works on Linux.
2150
2151 17. Under certain error circumstances pcregrep might try to free random memory
2152 as it exited. This is now fixed, thanks to valgrind.
2153
2154 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
2155 "abc\r\n\r\n", it found an unwanted second match after the second \r. This
2156 was because its rules for how to advance for /g after matching an empty
2157 string at the end of a line did not allow for this case. They now check for
2158 it specially.
2159
2160 20. pcretest is supposed to handle patterns and data of any length, by
2161 extending its buffers when necessary. It was getting this wrong when the
2162 buffer for a data line had to be extended.
2163
2164 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
2165 CRLF as a newline sequence.
2166
2167 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
2168 out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
2169 I have nevertheless tidied it up.
2170
2171 23. Added some casts to kill warnings from HP-UX ia64 compiler.
2172
2173 24. Added a man page for pcre-config.
2174
2175
2176 Version 7.0 19-Dec-06
2177 ---------------------
2178
2179 1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
2180 moving to gcc 4.1.1.
2181
2182 2. The -S option for pcretest uses setrlimit(); I had omitted to #include
2183 sys/time.h, which is documented as needed for this function. It doesn't
2184 seem to matter on Linux, but it showed up on some releases of OS X.
2185
2186 3. It seems that there are systems where bytes whose values are greater than
2187 127 match isprint() in the "C" locale. The "C" locale should be the
2188 default when a C program starts up. In most systems, only ASCII printing
2189 characters match isprint(). This difference caused the output from pcretest
2190 to vary, making some of the tests fail. I have changed pcretest so that:
2191
2192 (a) When it is outputting text in the compiled version of a pattern, bytes
2193 other than 32-126 are always shown as hex escapes.
2194
2195 (b) When it is outputting text that is a matched part of a subject string,
2196 it does the same, unless a different locale has been set for the match
2197 (using the /L modifier). In this case, it uses isprint() to decide.
2198
2199 4. Fixed a major bug that caused incorrect computation of the amount of memory
2200 required for a compiled pattern when options that changed within the
2201 pattern affected the logic of the preliminary scan that determines the
2202 length. The relevant options are -x, and -i in UTF-8 mode. The result was
2203 that the computed length was too small. The symptoms of this bug were
2204 either the PCRE error "internal error: code overflow" from pcre_compile(),
2205 or a glibc crash with a message such as "pcretest: free(): invalid next
2206 size (fast)". Examples of patterns that provoked this bug (shown in
2207 pcretest format) are:
2208
2209 /(?-x: )/x
2210 /(?x)(?-x: \s*#\s*)/
2211 /((?i)[\x{c0}])/8
2212 /(?i:[\x{c0}])/8
2213
2214 HOWEVER: Change 17 below makes this fix obsolete as the memory computation
2215 is now done differently.
2216
2217 5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
2218 wrapper classes; (b) implement a new function in the C++ scanner that is
2219 more efficient than the old way of doing things because it avoids levels of
2220 recursion in the regex matching; (c) add a paragraph to the documentation
2221 for the FullMatch() function.
2222
2223 6. The escape sequence \n was being treated as whatever was defined as
2224 "newline". Not only was this contrary to the documentation, which states
2225 that \n is character 10 (hex 0A), but it also went horribly wrong when
2226 "newline" was defined as CRLF. This has been fixed.
2227
2228 7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
2229 was being set to -1 for the "end of line" case (supposedly a value that no
2230 character can have). Though this value is never used (the check for end of
2231 line is "zero bytes in current character"), it caused compiler complaints.
2232 I've changed it to 0xffffffff.
2233
2234 8. In pcre_version.c, the version string was being built by a sequence of
2235 C macros that, in the event of PCRE_PRERELEASE being defined as an empty
2236 string (as it is for production releases) called a macro with an empty
2237 argument. The C standard says the result of this is undefined. The gcc
2238 compiler treats it as an empty string (which was what was wanted) but it is
2239 reported that Visual C gives an error. The source has been hacked around to
2240 avoid this problem.
2241
2242 9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
2243 builds of pcretest, and changed the call to _setmode() to use _O_BINARY
2244 instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
2245 of them did).
2246
2247 10. Originally, pcretest opened its input and output without "b"; then I was
2248 told that "b" was needed in some environments, so it was added for release
2249 5.0 to both the input and output. (It makes no difference on Unix-like
2250 systems.) Later I was told that it is wrong for the input on Windows. I've
2251 now abstracted the modes into two macros, to make it easier to fiddle with
2252 them, and removed "b" from the input mode under Windows.
2253
2254 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
2255
2256 12. Added -help and --help to pcretest as an official way of being reminded
2257 of the options.
2258
2259 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
2260 and pcrecpp.cc because they annoy compilers at high warning levels.
2261
2262 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
2263
2264 15. Fixed an occurrence of == in configure.ac that should have been = (shell
2265 scripts are not C programs :-) and which was not noticed because it works
2266 on Linux.
2267
2268 16. pcretest is supposed to handle any length of pattern and data line (as one
2269 line or as a continued sequence of lines) by extending its input buffer if
2270 necessary. This feature was broken for very long pattern lines, leading to
2271 a string of junk being passed to pcre_compile() if the pattern was longer
2272 than about 50K.
2273
2274 17. I have done a major re-factoring of the way pcre_compile() computes the
2275 amount of memory needed for a compiled pattern. Previously, there was code
2276 that made a preliminary scan of the pattern in order to do this. That was
2277 OK when PCRE was new, but as the facilities have expanded, it has become
2278 harder and harder to keep it in step with the real compile phase, and there
2279 have been a number of bugs (see for example, 4 above). I have now found a
2280 cunning way of running the real compile function in a "fake" mode that
2281 enables it to compute how much memory it would need, while actually only
2282 ever using a few hundred bytes of working memory and without too many
2283 tests of the mode. This should make future maintenance and development
2284 easier. A side effect of this work is that the limit of 200 on the nesting
2285 depth of parentheses has been removed (though this was never a serious
2286 limitation, I suspect). However, there is a downside: pcre_compile() now
2287 runs more slowly than before (30% or more, depending on the pattern). I
2288 hope this isn't a big issue. There is no effect on runtime performance.
2289
2290 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
2291 newline (only possible for the last line of a file) and it was a
2292 pattern that set a locale (followed by /Lsomething), pcretest crashed.
2293
2294 19. Added additional timing features to pcretest. (1) The -tm option now times
2295 matching only, not compiling. (2) Both -t and -tm can be followed, as a
2296 separate command line item, by a number that specifies the number of
2297 repeats to use when timing. The default is 50000; this gives better
2298 precision, but takes uncomfortably long for very large patterns.
2299
2300 20. Extended pcre_study() to be more clever in cases where a branch of a
2301 subpattern has no definite first character. For example, (a*|b*)[cd] would
2302 previously give no result from pcre_study(). Now it recognizes that the
2303 first character must be a, b, c, or d.
2304
2305 21. There was an incorrect error "recursive call could loop indefinitely" if
2306 a subpattern (or the entire pattern) that was being tested for matching an
2307 empty string contained only one non-empty item after a nested subpattern.
2308 For example, the pattern (?>\x{100}*)\d(?R) provoked this error
2309 incorrectly, because the \d was being skipped in the check.
2310
2311 22. The pcretest program now has a new pattern option /B and a command line
2312 option -b, which is equivalent to adding /B to every pattern. This causes
2313 it to show the compiled bytecode, without the additional information that
2314 -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
2315 is the same as /B/I).
2316
2317 23. A new optimization is now able automatically to treat some sequences such
2318 as a*b as a*+b. More specifically, if something simple (such as a character
2319 or a simple class like \d) has an unlimited quantifier, and is followed by
2320 something that cannot possibly match the quantified thing, the quantifier
2321 is automatically "possessified".
2322
2323 24. A recursive reference to a subpattern whose number was greater than 39
2324 went wrong under certain circumstances in UTF-8 mode. This bug could also
2325 have affected the operation of pcre_study().
2326
2327 25. Realized that a little bit of performance could be had by replacing
2328 (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
2329
2330 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
2331
2332 27. Possessive quantifiers such as a++ were previously implemented by turning
2333 them into atomic groups such as ($>a+). Now they have their own opcodes,
2334 which improves performance. This includes the automatically created ones
2335 from 23 above.
2336
2337 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
2338 lookahead was broken if it was not anchored. PCRE was mistakenly expecting
2339 the first matched character to be a colon. This applied both to named and
2340 numbered groups.
2341
2342 29. The ucpinternal.h header file was missing its idempotency #ifdef.
2343
2344 30. I was sent a "project" file called libpcre.a.dev which I understand makes
2345 building PCRE on Windows easier, so I have included it in the distribution.
2346
2347 31. There is now a check in pcretest against a ridiculously large number being
2348 returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
2349 loop, the loop is abandoned.
2350
2351 32. Forward references to subpatterns in conditions such as (?(2)...) where
2352 subpattern 2 is defined later cause pcre_compile() to search forwards in
2353 the pattern for the relevant set of parentheses. This search went wrong
2354 when there were unescaped parentheses in a character class, parentheses
2355 escaped with \Q...\E, or parentheses in a #-comment in /x mode.
2356
2357 33. "Subroutine" calls and backreferences were previously restricted to
2358 referencing subpatterns earlier in the regex. This restriction has now
2359 been removed.
2360
2361 34. Added a number of extra features that are going to be in Perl 5.10. On the
2362 whole, these are just syntactic alternatives for features that PCRE had
2363 previously implemented using the Python syntax or my own invention. The
2364 other formats are all retained for compatibility.
2365
2366 (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
2367 as (?P<name>...). The new forms, as well as being in Perl 5.10, are
2368 also .NET compatible.
2369
2370 (b) A recursion or subroutine call to a named group can now be defined as
2371 (?&name) as well as (?P>name).
2372
2373 (c) A backreference to a named group can now be defined as \k<name> or
2374 \k'name' as well as (?P=name). The new forms, as well as being in Perl
2375 5.10, are also .NET compatible.
2376
2377 (d) A conditional reference to a named group can now use the syntax
2378 (?(<name>) or (?('name') as well as (?(name).
2379
2380 (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
2381 groups (named and numbered) that are never evaluated inline, but can be
2382 called as "subroutines" from elsewhere. In effect, the DEFINE condition
2383 is always false. There may be only one alternative in such a group.
2384
2385 (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
2386 as the simple (?(R). The condition is true only if the most recent
2387 recursion is that of the given number or name. It does not search out
2388 through the entire recursion stack.
2389
2390 (g) The escape \gN or \g{N} has been added, where N is a positive or
2391 negative number, specifying an absolute or relative reference.
2392
2393 35. Tidied to get rid of some further signed/unsigned compiler warnings and
2394 some "unreachable code" warnings.
2395
2396 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
2397 things, this adds five new scripts.
2398
2399 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
2400 There were also incompatibilities regarding the handling of \Q..\E inside
2401 character classes, for example with patterns like [\Qa\E-\Qz\E] where the
2402 hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
2403
2404 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
2405 matches an empty string, and forcibly breaks the loop. There were bugs in
2406 this code in non-simple cases. For a pattern such as ^(a()*)* matched
2407 against aaaa the result was just "a" rather than "aaaa", for example. Two
2408 separate and independent bugs (that affected different cases) have been
2409 fixed.
2410
2411 39. Refactored the code to abolish the use of different opcodes for small
2412 capturing bracket numbers. This is a tidy that I avoided doing when I
2413 removed the limit on the number of capturing brackets for 3.5 back in 2001.
2414 The new approach is not only tidier, it makes it possible to reduce the
2415 memory needed to fix the previous bug (38).
2416
2417 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
2418 sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
2419 processing dot, circumflex, or dollar metacharacters, or #-comments in /x
2420 mode.
2421
2422 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
2423 report.
2424
2425 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
2426 copy construction and assignment in the C++ wrapper.
2427
2428 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
2429 couple of bugs that could have given wrong results in the "--newline=crlf"
2430 case.
2431
2432 44. Added a number of casts and did some reorganization of signed/unsigned int
2433 variables following suggestions from Dair Grant. Also renamed the variable
2434 "this" as "item" because it is a C++ keyword.
2435
2436 45. Arranged for dftables to add
2437
2438 #include "pcre_internal.h"
2439
2440 to pcre_chartables.c because without it, gcc 4.x may remove the array
2441 definition from the final binary if PCRE is built into a static library and
2442 dead code stripping is activated.
2443
2444 46. For an unanchored pattern, if a match attempt fails at the start of a
2445 newline sequence, and the newline setting is CRLF or ANY, and the next two
2446 characters are CRLF, advance by two characters instead of one.
2447
2448
2449 Version 6.7 04-Jul-06
2450 ---------------------
2451
2452 1. In order to handle tests when input lines are enormously long, pcretest has
2453 been re-factored so that it automatically extends its buffers when
2454 necessary. The code is crude, but this _is_ just a test program. The
2455 default size has been increased from 32K to 50K.
2456
2457 2. The code in pcre_study() was using the value of the re argument before
2458 testing it for NULL. (Of course, in any sensible call of the function, it
2459 won't be NULL.)
2460
2461 3. The memmove() emulation function in pcre_internal.h, which is used on
2462 systems that lack both memmove() and bcopy() - that is, hardly ever -
2463 was missing a "static" storage class specifier.
2464
2465 4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
2466 containing an extended class (one that cannot be represented by a bitmap
2467 because it contains high-valued characters or Unicode property items, e.g.
2468 [\pZ]). Almost always one would set UTF-8 mode when processing such a
2469 pattern, but PCRE should not loop if you do not (it no longer does).
2470 [Detail: two cases were found: (a) a repeated subpattern containing an
2471 extended class; (b) a recursive reference to a subpattern that followed a
2472 previous extended class. It wasn't skipping over the extended class
2473 correctly when UTF-8 mode was not set.]
2474
2475 5. A negated single-character class was not being recognized as fixed-length
2476 in lookbehind assertions such as (?<=[^f]), leading to an incorrect
2477 compile error "lookbehind assertion is not fixed length".
2478
2479 6. The RunPerlTest auxiliary script was showing an unexpected difference
2480 between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
2481 write a Perl script that can interpret lines of an input file either as
2482 byte characters or as UTF-8, which is what "perltest" was being required to
2483 do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
2484 can't do is switch easily at run time between having the "use utf8;" pragma
2485 or not. In the end, I fudged it by using the RunPerlTest script to insert
2486 "use utf8;" explicitly for the UTF-8 tests.
2487
2488 7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
2489 the end of the subject string, contrary to the documentation and to what
2490 Perl does. This was true of both matching functions. Now it matches only at
2491 the start of the subject and immediately after *internal* newlines.
2492
2493 8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
2494 a pointer to an int instead of a pointer to an unsigned long int. This
2495 caused problems on 64-bit systems.
2496
2497 9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
2498 instance of the 'standard' template library not being so standard".
2499
2500 10. There was no check on the number of named subpatterns nor the maximum
2501 length of a subpattern name. The product of these values is used to compute
2502 the size of the memory block for a compiled pattern. By supplying a very
2503 long subpattern name and a large number of named subpatterns, the size
2504 computation could be caused to overflow. This is now prevented by limiting
2505 the length of names to 32 characters, and the number of named subpatterns
2506 to 10,000.
2507
2508 11. Subpatterns that are repeated with specific counts have to be replicated in
2509 the compiled pattern. The size of memory for this was computed from the
2510 length of the subpattern and the repeat count. The latter is limited to
2511 65535, but there was no limit on the former, meaning that integer overflow
2512 could in principle occur. The compiled length of a repeated subpattern is
2513 now limited to 30,000 bytes in order to prevent this.
2514
2515 12. Added the optional facility to have named substrings with the same name.
2516
2517 13. Added the ability to use a named substring as a condition, using the
2518 Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
2519 are numbers (not recommended). Forward references are permitted.
2520
2521 14. Added forward references in named backreferences (if you see what I mean).
2522
2523 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
2524 pattern could run off the end of the subject. For example, the pattern
2525 "(?s)(.{1,5})"8 did this with the subject "ab".
2526
2527 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
2528 PCRE_CASELESS was set when matching characters that were quantified with ?
2529 or *.
2530
2531 17. A character class other than a single negated character that had a minimum
2532 but no maximum quantifier - for example [ab]{6,} - was not handled
2533 correctly by pce_dfa_exec(). It would match only one character.
2534
2535 18. A valid (though odd) pattern that looked like a POSIX character
2536 class but used an invalid character after [ (for example [[,abc,]]) caused
2537 pcre_compile() to give the error "Failed: internal error: code overflow" or
2538 in some cases to crash with a glibc free() error. This could even happen if
2539 the pattern terminated after [[ but there just happened to be a sequence of
2540 letters, a binary zero, and a closing ] in the memory that followed.
2541
2542 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
2543 over the years. Originally (before any Unicode support), just the bottom 8
2544 bits were taken. Thus, for example, \500 really meant \100. Nowadays the
2545 output from "man perlunicode" includes this:
2546
2547 The regular expression compiler produces polymorphic opcodes. That
2548 is, the pattern adapts to the data and automatically switches to
2549 the Unicode character scheme when presented with Unicode data--or
2550 instead uses a traditional byte scheme when presented with byte
2551 data.
2552
2553 Sadly, a wide octal escape does not cause a switch, and in a string with
2554 no other multibyte characters, these octal escapes are treated as before.
2555 Thus, in Perl, the pattern /\500/ actually matches \100 but the pattern
2556 /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
2557 Unicode string.
2558
2559 I have not perpetrated such confusion in PCRE. Up till now, it took just
2560 the bottom 8 bits, as in old Perl. I have now made octal escapes with
2561 values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
2562 translate to the appropriate multibyte character.
2563
2564 29. Applied some refactoring to reduce the number of warnings from Microsoft
2565 and Borland compilers. This has included removing the fudge introduced
2566 seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
2567 a warning about an unused variable.
2568
2569 21. PCRE has not included VT (character 0x0b) in the set of whitespace
2570 characters since release 4.0, because Perl (from release 5.004) does not.
2571 [Or at least, is documented not to: some releases seem to be in conflict
2572 with the documentation.] However, when a pattern was studied with
2573 pcre_study() and all its branches started with \s, PCRE still included VT
2574 as a possible starting character. Of course, this did no harm; it just
2575 caused an unnecessary match attempt.
2576
2577 22. Removed a now-redundant internal flag bit that recorded the fact that case
2578 dependency changed within the pattern. This was once needed for "required
2579 byte" processing, but is no longer used. This recovers a now-scarce options
2580 bit. Also moved the least significant internal flag bit to the most-
2581 significant bit of the word, which was not previously used (hangover from
2582 the days when it was an int rather than a uint) to free up another bit for
2583 the future.
2584
2585 23. Added support for CRLF line endings as well as CR and LF. As well as the
2586 default being selectable at build time, it can now be changed at runtime
2587 via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
2588 specify that it is scanning data with non-default line endings.
2589
2590 24. Changed the definition of CXXLINK to make it agree with the definition of
2591 LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
2592
2593 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
2594 recursions. This makes a big different to stack usage for some patterns.
2595
2596 26. If a subpattern containing a named recursion or subroutine reference such
2597 as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
2598 the space required for the compiled pattern went wrong and gave too small a
2599 value. Depending on the environment, this could lead to "Failed: internal
2600 error: code overflow at offset 49" or "glibc detected double free or
2601 corruption" errors.
2602
2603 27. Applied patches from Google (a) to support the new newline modes and (b) to
2604 advance over multibyte UTF-8 characters in GlobalReplace.
2605
2606 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
2607 difference for some implementation of PCRE in some Windows version.
2608
2609 29. Added some extra testing facilities to pcretest:
2610
2611 \q<number> in a data line sets the "match limit" value
2612 \Q<number> in a data line sets the "match recursion limt" value
2613 -S <number> sets the stack size, where <number> is in megabytes
2614
2615 The -S option isn't available for Windows.
2616
2617
2618 Version 6.6 06-Feb-06
2619 ---------------------
2620
2621 1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
2622 in pcreposix.h. I have copied the definition from pcre.h.
2623
2624 2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
2625 because pcre.h is no longer a built file.
2626
2627 3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
2628 not normally included in the compiled code.
2629
2630
2631 Version 6.5 01-Feb-06
2632 ---------------------
2633
2634 1. When using the partial match feature with pcre_dfa_exec(), it was not
2635 anchoring the second and subsequent partial matches at the new starting
2636 point. This could lead to incorrect results. For example, with the pattern
2637 /1234/, partially matching against "123" and then "a4" gave a match.
2638
2639 2. Changes to pcregrep:
2640
2641 (a) All non-match returns from pcre_exec() were being treated as failures
2642 to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
2643 error message is output. Some extra information is given for the
2644 PCRE_ERROR_MATCHLIMIT and PCRE_ERROR_RECURSIONLIMIT errors, which are
2645 probably the only errors that are likely to be caused by users (by
2646 specifying a regex that has nested indefinite repeats, for instance).
2647 If there are more than 20 of these errors, pcregrep is abandoned.
2648
2649 (b) A binary zero was treated as data while matching, but terminated the
2650 output line if it was written out. This has been fixed: binary zeroes
2651 are now no different to any other data bytes.
2652
2653 (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
2654 used to set a locale for matching. The --locale=xxxx long option has
2655 been added (no short equivalent) to specify a locale explicitly on the
2656 pcregrep command, overriding the environment variables.
2657
2658 (d) When -B was used with -n, some line numbers in the output were one less
2659 than they should have been.
2660
2661 (e) Added the -o (--only-matching) option.
2662
2663 (f) If -A or -C was used with -c (count only), some lines of context were
2664 accidentally printed for the final match.
2665
2666 (g) Added the -H (--with-filename) option.
2667
2668 (h) The combination of options -rh failed to suppress file names for files
2669 that were found from directory arguments.
2670
2671 (i) Added the -D (--devices) and -d (--directories) options.
2672
2673 (j) Added the -F (--fixed-strings) option.
2674
2675 (k) Allow "-" to be used as a file name for -f as well as for a data file.
2676
2677 (l) Added the --colo(u)r option.
2678
2679 (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
2680 is not present by default.
2681
2682 3. A nasty bug was discovered in the handling of recursive patterns, that is,
2683 items such as (?R) or (?1), when the recursion could match a number of
2684 alternatives. If it matched one of the alternatives, but subsequently,
2685 outside the recursion, there was a failure, the code tried to back up into
2686 the recursion. However, because of the way PCRE is implemented, this is not
2687 possible, and the result was an incorrect result from the match.
2688
2689 In order to prevent this happening, the specification of recursion has
2690 been changed so that all such subpatterns are automatically treated as
2691 atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
2692
2693 4. I had overlooked the fact that, in some locales, there are characters for
2694 which isalpha() is true but neither isupper() nor islower() are true. In
2695 the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
2696 and ordfeminine) are like this. This affected the treatment of \w and \W
2697 when they appeared in character classes, but not when they appeared outside
2698 a character class. The bit map for "word" characters is now created
2699 separately from the results of isalnum() instead of just taking it from the
2700 upper, lower, and digit maps. (Plus the underscore character, of course.)
2701
2702 5. The above bug also affected the handling of POSIX character classes such as
2703 [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
2704 permanent tables. Instead, the bit maps for such a class were previously
2705 created as the appropriate unions of the upper, lower, and digit bitmaps.
2706 Now they are created by subtraction from the [[:word:]] class, which has
2707 its own bitmap.
2708
2709 6. The [[:blank:]] character class matches horizontal, but not vertical space.
2710 It is created by subtracting the vertical space characters (\x09, \x0a,
2711 \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
2712 subtraction was done in the overall bitmap for a character class, meaning
2713 that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
2714 be recognized. This bug has been fixed.
2715
2716 7. Patches from the folks at Google:
2717
2718 (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
2719 real life, but is still worth protecting against".
2720
2721 (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
2722 regular expressions".
2723
2724 (c) pcre_scanner.cc: avoid use of std::count() because not all systems
2725 have it.
2726
2727 (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
2728 "configure" and the latter not, in order to fix a problem somebody had
2729 with compiling the Arg class on HP-UX.
2730
2731 (e) Improve the error-handling of the C++ wrapper a little bit.
2732
2733 (f) New tests for checking recursion limiting.
2734
2735 8. The pcre_memmove() function, which is used only if the environment does not
2736 have a standard memmove() function (and is therefore rarely compiled),
2737 contained two bugs: (a) use of int instead of size_t, and (b) it was not
2738 returning a result (though PCRE never actually uses the result).
2739
2740 9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
2741 large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
2742 returned instead of calling malloc() with an overflowing number that would
2743 most likely cause subsequent chaos.
2744
2745 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
2746
2747 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
2748 with this option is matched, the nmatch and pmatch options of regexec() are
2749 ignored.
2750
2751 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
2752 provided in case anyone wants to the the POSIX interface with UTF-8
2753 strings.
2754
2755 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
2756 C++ linking (needed for some HP-UX environments).
2757
2758 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
2759 (unused parameter) and in the pcre_printint() function (omitted "default"
2760 switch label when the default is to do nothing).
2761
2762 15. Added some code to make it possible, when PCRE is compiled as a C++
2763 library, to replace subject pointers for pcre_exec() with a smart pointer
2764 class, thus making it possible to process discontinuous strings.
2765
2766 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
2767 much the same function. They were added by different people who were trying
2768 to make PCRE easy to compile on non-Unix systems. It has been suggested
2769 that PCRE_EXPORT be abolished now that there is more automatic apparatus
2770 for compiling on Windows systems. I have therefore replaced it with
2771 PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
2772 defaults to "extern" for C or "extern C" for C++, which works fine on
2773 Unix-like systems. It is now possible to override the value of PCRE_DATA_
2774 SCOPE with something explicit in config.h. In addition:
2775
2776 (a) pcreposix.h still had just "extern" instead of either of these macros;
2777 I have replaced it with PCRE_DATA_SCOPE.
2778
2779 (b) Functions such as _pcre_xclass(), which are internal to the library,
2780 but external in the C sense, all had PCRE_EXPORT in their definitions.
2781 This is apparently wrong for the Windows case, so I have removed it.
2782 (It makes no difference on Unix-like systems.)
2783
2784 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
2785 of recursive calls to match(). This is different to MATCH_LIMIT because
2786 that limits the total number of calls to match(), not all of which increase
2787 the depth of recursion. Limiting the recursion depth limits the amount of
2788 stack (or heap if NO_RECURSE is set) that is used. The default can be set
2789 when PCRE is compiled, and changed at run time. A patch from Google adds
2790 this functionality to the C++ interface.
2791
2792 18. Changes to the handling of Unicode character properties:
2793
2794 (a) Updated the table to Unicode 4.1.0.
2795
2796 (b) Recognize characters that are not in the table as "Cn" (undefined).
2797
2798 (c) I revised the way the table is implemented to a much improved format
2799 which includes recognition of ranges. It now supports the ranges that
2800 are defined in UnicodeData.txt, and it also amalgamates other
2801 characters into ranges. This has reduced the number of entries in the
2802 table from around 16,000 to around 3,000, thus reducing its size
2803 considerably. I realized I did not need to use a tree structure after
2804 all - a binary chop search is just as efficient. Having reduced the
2805 number of entries, I extended their size from 6 bytes to 8 bytes to
2806 allow for more data.
2807
2808 (d) Added support for Unicode script names via properties such as \p{Han}.
2809
2810 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
2811 matching that character.
2812
2813 20. When matching a repeated Unicode property with a minimum greater than zero,
2814 (for example \pL{2,}), PCRE could look past the end of the subject if it
2815 reached it while seeking the minimum number of characters. This could
2816 happen only if some of the characters were more than one byte long, because
2817 there is a check for at least the minimum number of bytes.
2818
2819 21. Refactored the implementation of \p and \P so as to be more general, to
2820 allow for more different types of property in future. This has changed the
2821 compiled form incompatibly. Anybody with saved compiled patterns that use
2822 \p or \P will have to recompile them.
2823
2824 22. Added "Any" and "L&" to the supported property types.
2825
2826 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
2827 but give a compile time error if the value is greater than 0xff.
2828
2829 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
2830 accidentally not being installed or uninstalled.
2831
2832 25. The pcre.h file was built from pcre.h.in, but the only changes that were
2833 made were to insert the current release number. This seemed silly, because
2834 it made things harder for people building PCRE on systems that don't run
2835 "configure". I have turned pcre.h into a distributed file, no longer built
2836 by "configure", with the version identification directly included. There is
2837 no longer a pcre.h.in file.
2838
2839 However, this change necessitated a change to the pcre-config script as
2840 well. It is built from pcre-config.in, and one of the substitutions was the
2841 release number. I have updated configure.ac so that ./configure now finds
2842 the release number by grepping pcre.h.
2843
2844 26. Added the ability to run the tests under valgrind.
2845
2846
2847 Version 6.4 05-Sep-05
2848 ---------------------
2849
2850 1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
2851 "--" to be printed when multiple files were scanned, even when none of the
2852 -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
2853 consider it to be a bug, and have restored the previous behaviour.
2854
2855 2. A couple of code tidies to get rid of compiler warnings.
2856
2857 3. The pcretest program used to cheat by referring to symbols in the library
2858 whose names begin with _pcre_. These are internal symbols that are not
2859 really supposed to be visible externally, and in some environments it is
2860 possible to suppress them. The cheating is now confined to including
2861 certain files from the library's source, which is a bit cleaner.
2862
2863 4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
2864 file's purpose clearer.
2865
2866 5. Reorganized pcre_ucp_findchar().
2867
2868
2869 Version 6.3 15-Aug-05
2870 ---------------------
2871
2872 1. The file libpcre.pc.in did not have general read permission in the tarball.
2873
2874 2. There were some problems when building without C++ support:
2875
2876 (a) If C++ support was not built, "make install" and "make test" still
2877 tried to test it.
2878
2879 (b) There were problems when the value of CXX was explicitly set. Some
2880 changes have been made to try to fix these, and ...
2881
2882 (c) --disable-cpp can now be used to explicitly disable C++ support.
2883
2884 (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
2885 backslash in a target when C++ was disabled. This confuses some
2886 versions of "make", apparently. Using an intermediate variable solves
2887 this. (Same for CPP_LOBJ.)
2888
2889 3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
2890 (non-Windows) now includes $(CFLAGS) because these flags are sometimes
2891 necessary on certain architectures.
2892
2893 4. Added a setting of -export-symbols-regex to the link command to remove
2894 those symbols that are exported in the C sense, but actually are local
2895 within the library, and not documented. Their names all begin with
2896 "_pcre_". This is not a perfect job, because (a) we have to except some
2897 symbols that pcretest ("illegally") uses, and (b) the facility isn't always
2898 available (and never for static libraries). I have made a note to try to
2899 find a way round (a) in the future.
2900
2901
2902 Version 6.2 01-Aug-05
2903 ---------------------
2904
2905 1. There was no test for integer overflow of quantifier values. A construction
2906 such as {1111111111111111} would give undefined results. What is worse, if
2907 a minimum quantifier for a parenthesized subpattern overflowed and became
2908 negative, the calculation of the memory size went wrong. This could have
2909 led to memory overwriting.
2910
2911 2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
2912
2913 3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
2914 operating environments where this matters.
2915
2916 4. Applied Giuseppe Maxia's patch to add additional features for controlling
2917 PCRE options from within the C++ wrapper.
2918
2919 5. Named capturing subpatterns were not being correctly counted when a pattern
2920 was compiled. This caused two problems: (a) If there were more than 100
2921 such subpatterns, the calculation of the memory needed for the whole
2922 compiled pattern went wrong, leading to an overflow error. (b) Numerical
2923 back references of the form \12, where the number was greater than 9, were
2924 not recognized as back references, even though there were sufficient
2925 previous subpatterns.
2926
2927 6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
2928 versions of gcc, e.g. 2.95.4.
2929
2930
2931 Version 6.1 21-Jun-05
2932 ---------------------
2933
2934 1. There was one reference to the variable "posix" in pcretest.c that was not
2935 surrounded by "#if !defined NOPOSIX".
2936
2937 2. Make it possible to compile pcretest without DFA support, UTF8 support, or
2938 the cross-check on the old pcre_info() function, for the benefit of the
2939 cut-down version of PCRE that is currently imported into Exim.
2940
2941 3. A (silly) pattern starting with (?i)(?-i) caused an internal space
2942 allocation error. I've done the easy fix, which wastes 2 bytes for sensible
2943 patterns that start (?i) but I don't think that matters. The use of (?i) is
2944 just an example; this all applies to the other options as well.
2945
2946 4. Since libtool seems to echo the compile commands it is issuing, the output
2947 from "make" can be reduced a bit by putting "@" in front of each libtool
2948 compile command.
2949
2950 5. Patch from the folks at Google for configure.in to be a bit more thorough
2951 in checking for a suitable C++ installation before trying to compile the
2952 C++ stuff. This should fix a reported problem when a compiler was present,
2953 but no suitable headers.
2954
2955 6. The man pages all had just "PCRE" as their title. I have changed them to
2956 be the relevant file name. I have also arranged that these names are
2957 retained in the file doc/pcre.txt, which is a concatenation in text format
2958 of all the man pages except the little individual ones for each function.
2959
2960 7. The NON-UNIX-USE file had not been updated for the different set of source
2961 files that come with release 6. I also added a few comments about the C++
2962 wrapper.
2963
2964
2965 Version 6.0 07-Jun-05
2966 ---------------------
2967
2968 1. Some minor internal re-organization to help with my DFA experiments.
2969
2970 2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
2971 didn't matter for the library itself when fully configured, but did matter
2972 when compiling without UCP support, or within Exim, where the ucp files are
2973 not imported.
2974
2975 3. Refactoring of the library code to split up the various functions into
2976 different source modules. The addition of the new DFA matching code (see
2977 below) to a single monolithic source would have made it really too
2978 unwieldy, quite apart from causing all the code to be include in a
2979 statically linked application, when only some functions are used. This is
2980 relevant even without the DFA addition now that patterns can be compiled in
2981 one application and matched in another.
2982
2983 The downside of splitting up is that there have to be some external
2984 functions and data tables that are used internally in different modules of
2985 the library but which are not part of the API. These have all had their
2986 names changed to start with "_pcre_" so that they are unlikely to clash
2987 with other external names.
2988
2989 4. Added an alternate matching function, pcre_dfa_exec(), which matches using
2990 a different (DFA) algorithm. Although it is slower than the original
2991 function, it does have some advantages for certain types of matching
2992 problem.
2993
2994 5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
2995 including restarting after a partial match.
2996
2997 6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
2998 defined when compiling for Windows was sent to me. I have put it into the
2999 code, though I have no means of testing or verifying it.
3000
3001 7. Added the pcre_refcount() auxiliary function.
3002
3003 8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
3004 match before or at the first newline in the subject string. In pcretest,
3005 the /f option on a pattern can be used to set this.
3006
3007 9. A repeated \w when used in UTF-8 mode with characters greater than 256
3008 would behave wrongly. This has been present in PCRE since release 4.0.
3009
3010 10. A number of changes to the pcregrep command:
3011
3012 (a) Refactored how -x works; insert ^(...)$ instead of setting
3013 PCRE_ANCHORED and checking the length, in preparation for adding
3014 something similar for -w.
3015
3016 (b) Added the -w (match as a word) option.
3017
3018 (c) Refactored the way lines are read and buffered so as to have more
3019 than one at a time available.
3020
3021 (d) Implemented a pcregrep test script.
3022
3023 (e) Added the -M (multiline match) option. This allows patterns to match
3024 over several lines of the subject. The buffering ensures that at least
3025 8K, or the rest of the document (whichever is the shorter) is available
3026 for matching (and similarly the previous 8K for lookbehind assertions).
3027
3028 (f) Changed the --help output so that it now says
3029
3030 -w, --word-regex(p)
3031
3032 instead of two lines, one with "regex" and the other with "regexp"
3033 because that confused at least one person since the short forms are the
3034 same. (This required a bit of code, as the output is generated
3035 automatically from a table. It wasn't just a text change.)
3036
3037 (g) -- can be used to terminate pcregrep options if the next thing isn't an
3038 option but starts with a hyphen. Could be a pattern or a path name
3039 starting with a hyphen, for instance.
3040
3041 (h) "-" can be given as a file name to represent stdin.
3042
3043 (i) When file names are being printed, "(standard input)" is used for
3044 the standard input, for compatibility with GNU grep. Previously
3045 "<stdin>" was used.
3046
3047 (j) The option --label=xxx can be used to supply a name to be used for
3048 stdin when file names are being printed. There is no short form.
3049
3050 (k) Re-factored the options decoding logic because we are going to add
3051 two more options that take data. Such options can now be given in four
3052 different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
3053
3054 (l) Added the -A, -B, and -C options for requesting that lines of context
3055 around matches be printed.
3056
3057 (m) Added the -L option to print the names of files that do not contain
3058 any matching lines, that is, the complement of -l.
3059
3060 (n) The return code is 2 if any file cannot be opened, but pcregrep does
3061 continue to scan other files.
3062
3063 (o) The -s option was incorrectly implemented. For compatibility with other
3064 greps, it now suppresses the error message for a non-existent or non-
3065 accessible file (but not the return code). There is a new option called
3066 -q that suppresses the output of matching lines, which was what -s was
3067 previously doing.
3068
3069 (p) Added --include and --exclude options to specify files for inclusion
3070 and exclusion when recursing.
3071
3072 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
3073 Hopefully, it now does.
3074
3075 12. Missing cast in pcre_study().
3076
3077 13. Added an "uninstall" target to the makefile.
3078
3079 14. Replaced "extern" in the function prototypes in Makefile.in with
3080 "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
3081 world, but is set differently for Windows.
3082
3083 15. Added a second compiling function called pcre_compile2(). The only
3084 difference is that it has an extra argument, which is a pointer to an
3085 integer error code. When there is a compile-time failure, this is set
3086 non-zero, in addition to the error test pointer being set to point to an
3087 error message. The new argument may be NULL if no error number is required
3088 (but then you may as well call pcre_compile(), which is now just a
3089 wrapper). This facility is provided because some applications need a
3090 numeric error indication, but it has also enabled me to tidy up the way
3091 compile-time errors are handled in the POSIX wrapper.
3092
3093 16. Added VPATH=.libs to the makefile; this should help when building with one
3094 prefix path and installing with another. (Or so I'm told by someone who
3095 knows more about this stuff than I do.)
3096
3097 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
3098 passes PCRE_DOTALL to the pcre_compile() function, making the "." character
3099 match everything, including newlines. This is not POSIX-compatible, but
3100 somebody wanted the feature. From pcretest it can be activated by using
3101 both the P and the s flags.
3102
3103 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
3104
3105 19. libpcre.pc was being incorrectly installed as executable.
3106
3107 20. A couple of places in pcretest check for end-of-line by looking for '\n';
3108 it now also looks for '\r' so that it will work unmodified on Windows.
3109
3110 21. Added Google's contributed C++ wrapper to the distribution.
3111
3112 22. Added some untidy missing memory free() calls in pcretest, to keep
3113 Electric Fence happy when testing.
3114
3115
3116
3117 Version 5.0 13-Sep-04
3118 ---------------------
3119
3120 1. Internal change: literal characters are no longer packed up into items
3121 containing multiple characters in a single byte-string. Each character
3122 is now matched using a separate opcode. However, there may be more than one
3123 byte in the character in UTF-8 mode.
3124
3125 2. The pcre_callout_block structure has two new fields: pattern_position and
3126 next_item_length. These contain the offset in the pattern to the next match
3127 item, and its length, respectively.
3128
3129 3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
3130 insertion of callouts before each pattern item. Added the /C option to
3131 pcretest to make use of this.
3132
3133 4. On the advice of a Windows user, the lines
3134
3135 #if defined(_WIN32) || defined(WIN32)
3136 _setmode( _fileno( stdout ), 0x8000 );
3137 #endif /* defined(_WIN32) || defined(WIN32) */
3138
3139 have been added to the source of pcretest. This apparently does useful
3140 magic in relation to line terminators.
3141
3142 5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
3143 for the benefit of those environments where the "b" makes a difference.
3144
3145 6. The icc compiler has the same options as gcc, but "configure" doesn't seem
3146 to know about it. I have put a hack into configure.in that adds in code
3147 to set GCC=yes if CC=icc. This seems to end up at a point in the
3148 generated configure script that is early enough to affect the setting of
3149 compiler options, which is what is needed, but I have no means of testing
3150 whether it really works. (The user who reported this had patched the
3151 generated configure script, which of course I cannot do.)
3152
3153 LATER: After change 22 below (new libtool files), the configure script
3154 seems to know about icc (and also ecc). Therefore, I have commented out
3155 this hack in configure.in.
3156
3157 7. Added support for pkg-config (2 patches were sent in).
3158
3159 8. Negated POSIX character classes that used a combination of internal tables
3160 were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
3161 [[:^ascii]]. Typically, they would match almost any characters. The other
3162 POSIX classes were not broken in this way.
3163
3164 9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
3165 to find the match, as PCRE was deluded into thinking that the match had to
3166 start at the start point or following a newline. The same bug applied to
3167 patterns with negative forward assertions or any backward assertions
3168 preceding ".*" at the start, unless the pattern required a fixed first
3169 character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
3170
3171 10. In UTF-8 mode, when moving forwards in the subject after a failed match
3172 starting at the last subject character, bytes beyond the end of the subject
3173 string were read.
3174
3175 11. Renamed the variable "class" as "classbits" to make life easier for C++
3176 users. (Previously there was a macro definition, but it apparently wasn't
3177 enough.)
3178
3179 12. Added the new field "tables" to the extra data so that tables can be passed
3180 in at exec time, or the internal tables can be re-selected. This allows
3181 a compiled regex to be saved and re-used at a later time by a different
3182 program that might have everything at different addresses.
3183
3184 13. Modified the pcre-config script so that, when run on Solaris, it shows a
3185 -R library as well as a -L library.
3186
3187 14. The debugging options of pcretest (-d on the command line or D on a
3188 pattern) showed incorrect output for anything following an extended class
3189 that contained multibyte characters and which was followed by a quantifier.
3190
3191 15. Added optional support for general category Unicode character properties
3192 via the \p, \P, and \X escapes. Unicode property support implies UTF-8
3193 support. It adds about 90K to the size of the library. The meanings of the
3194 inbuilt class escapes such as \d and \s have NOT been changed.
3195
3196 16. Updated pcredemo.c to include calls to free() to release the memory for the
3197 compiled pattern.
3198
3199 17. The generated file chartables.c was being created in the source directory
3200 instead of in the building directory. This caused the build to fail if the
3201 source directory was different from the building directory, and was
3202 read-only.
3203
3204 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
3205 file. No doubt somebody will tell me if they don't make sense... Also added
3206 Dan Mooney's comments about building on OpenVMS.
3207
3208 19. Added support for partial matching via the PCRE_PARTIAL option for
3209 pcre_exec() and the \P data escape in pcretest.
3210
3211 20. Extended pcretest with 3 new pattern features:
3212
3213 (i) A pattern option of the form ">rest-of-line" causes pcretest to
3214 write the compiled pattern to the file whose name is "rest-of-line".
3215 This is a straight binary dump of the data, with the saved pointer to
3216 the character tables forced to be NULL. The study data, if any, is
3217 written too. After writing, pcretest reads a new pattern.
3218
3219 (ii) If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
3220 compiled pattern from the given file. There must not be any
3221 occurrences of "<" in the file name (pretty unlikely); if there are,
3222 pcretest will instead treat the initial "<" as a pattern delimiter.
3223 After reading in the pattern, pcretest goes on to read data lines as
3224 usual.
3225
3226 (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
3227 and 16-bit fields in a compiled pattern, to simulate a pattern that
3228 was compiled on a host of opposite endianness.
3229
3230 21. The pcre-exec() function can now cope with patterns that were compiled on
3231 hosts of opposite endianness, with this restriction:
3232
3233 As for any compiled expression that is saved and used later, the tables
3234 pointer field cannot be preserved; the extra_data field in the arguments
3235 to pcre_exec() should be used to pass in a tables address if a value
3236 other than the default internal tables were used at compile time.
3237
3238 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
3239 now diagnosed as an error. Previously, most of the time, a negative number
3240 would have been treated as zero, but if in addition "ovector" was passed as
3241 NULL, a crash could occur.
3242
3243 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
3244 new versions from the libtool 1.5 distribution (the last one is a copy of
3245 a file called libtool.m4). This seems to have fixed the need to patch
3246 "configure" to support Darwin 1.3 (which I used to do). However, I still
3247 had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
3248 workstation).
3249
3250 24. Changed the PCRE licence to be the more standard "BSD" licence.
3251
3252
3253 Version 4.5 01-Dec-03
3254 ---------------------
3255
3256 1. There has been some re-arrangement of the code for the match() function so
3257 that it can be compiled in a version that does not call itself recursively.
3258 Instead, it keeps those local variables that need separate instances for
3259 each "recursion" in a frame on the heap, and gets/frees frames whenever it
3260 needs to "recurse". Keeping track of where control must go is done by means
3261 of setjmp/longjmp. The whole thing is implemented by a set of macros that
3262 hide most of the details from the main code, and operates only if
3263 NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
3264 "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
3265 operating.
3266
3267 To make it easier for callers to provide specially tailored get/free
3268 functions for this usage, two new functions, pcre_stack_malloc, and
3269 pcre_stack_free, are used. They are always called in strict stacking order,
3270 and the size of block requested is always the same.
3271
3272 The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
3273 PCRE has been compiled to use the stack or the heap for recursion. The
3274 -C option of pcretest uses this to show which version is compiled.
3275
3276 A new data escape \S, is added to pcretest; it causes the amounts of store
3277 obtained and freed by both kinds of malloc/free at match time to be added
3278 to the output.
3279
3280 2. Changed the locale test to use "fr_FR" instead of "fr" because that's
3281 what's available on my current Linux desktop machine.
3282
3283 3. When matching a UTF-8 string, the test for a valid string at the start has
3284 been extended. If start_offset is not zero, PCRE now checks that it points
3285 to a byte that is the start of a UTF-8 character. If not, it returns
3286 PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
3287 this is necessary because there may be backward assertions in the pattern.
3288 When matching the same subject several times, it may save resources to use
3289 PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
3290
3291 4. The code for checking the validity of UTF-8 strings has been tightened so
3292 that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
3293 containing "overlong sequences".
3294
3295 5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
3296 I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
3297 should have been "&", but it just so happened that all the cases this let
3298 through by mistake were picked up later in the function.
3299
3300 6. I had used a variable called "isblank" - this is a C99 function, causing
3301 some compilers to warn. To avoid this, I renamed it (as "blankclass").
3302
3303 7. Cosmetic: (a) only output another newline at the end of pcretest if it is
3304 prompting; (b) run "./pcretest /dev/null" at the start of the test script
3305 so the version is shown; (c) stop "make test" echoing "./RunTest".
3306
3307 8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
3308
3309 9. The prototype for memmove() for systems that don't have it was using
3310 size_t, but the inclusion of the header that defines size_t was later. I've
3311 moved the #includes for the C headers earlier to avoid this.
3312
3313 10. Added some adjustments to the code to make it easier to compiler on certain
3314 special systems:
3315
3316 (a) Some "const" qualifiers were missing.
3317 (b) Added the macro EXPORT before all exported functions; by default this
3318 is defined to be empty.
3319 (c) Changed the dftables auxiliary program (that builds chartables.c) so
3320 that it reads its output file name as an argument instead of writing
3321 to the standard output and assuming this can be redirected.
3322
3323 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
3324 class containing characters with values greater than 255, PCRE compilation
3325 went into a loop.
3326
3327 12. A recursive reference to a subpattern that was within another subpattern
3328 that had a minimum quantifier of zero caused PCRE to crash. For example,
3329 (x(y(?2))z)? provoked this bug with a subject that got as far as the
3330 recursion. If the recursively-called subpattern itself had a zero repeat,
3331 that was OK.
3332
3333 13. In pcretest, the buffer for reading a data line was set at 30K, but the
3334 buffer into which it was copied (for escape processing) was still set at
3335 1024, so long lines caused crashes.
3336
3337 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
3338 "internal error: code overflow...". This applied to any character class
3339 that was followed by a possessive quantifier.
3340
3341 15. Modified the Makefile to add libpcre.la as a prerequisite for
3342 libpcreposix.la because I was told this is needed for a parallel build to
3343 work.
3344
3345 16. If a pattern that contained .* following optional items at the start was
3346 studied, the wrong optimizing data was generated, leading to matching
3347 errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
3348 matching string must start with a or b or c. The correct conclusion for
3349 this pattern is that a match can start with any character.
3350
3351
3352 Version 4.4 13-Aug-03
3353 ---------------------
3354
3355 1. In UTF-8 mode, a character class containing characters with values between
3356 127 and 255 was not handled correctly if the compiled pattern was studied.
3357 In fixing this, I have also improved the studying algorithm for such
3358 classes (slightly).
3359
3360 2. Three internal functions had redundant arguments passed to them. Removal
3361 might give a very teeny performance improvement.
3362
3363 3. Documentation bug: the value of the capture_top field in a callout is *one
3364 more than* the number of the hightest numbered captured substring.
3365
3366 4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
3367 in incorrectly linking with a previously installed version. They now link
3368 explicitly with libpcre.la.
3369
3370 5. configure.in no longer needs to recognize Cygwin specially.
3371
3372 6. A problem in pcre.in for Windows platforms is fixed.
3373
3374 7. If a pattern was successfully studied, and the -d (or /D) flag was given to
3375 pcretest, it used to include the size of the study block as part of its
3376 output. Unfortunately, the structure contains a field that has a different
3377 size on different hardware architectures. This meant that the tests that
3378 showed this size failed. As the block is currently always of a fixed size,
3379 this information isn't actually particularly useful in pcretest output, so
3380 I have just removed it.
3381
3382 8. Three pre-processor statements accidentally did not start in column 1.
3383 Sadly, there are *still* compilers around that complain, even though
3384 standard C has not required this for well over a decade. Sigh.
3385
3386 9. In pcretest, the code for checking callouts passed small integers in the
3387 callout_data field, which is a void * field. However, some picky compilers
3388 complained about the casts involved for this on 64-bit systems. Now
3389 pcretest passes the address of the small integer instead, which should get
3390 rid of the warnings.
3391
3392 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
3393 both compile and run time, and gives an error if an invalid UTF-8 sequence
3394 is found. There is a option for disabling this check in cases where the
3395 string is known to be correct and/or the maximum performance is wanted.
3396
3397 11. In response to a bug report, I changed one line in Makefile.in from
3398
3399 -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
3400 to
3401 -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
3402
3403 to look similar to other lines, but I have no way of telling whether this
3404 is the right thing to do, as I do not use Windows. No doubt I'll get told
3405 if it's wrong...
3406
3407
3408 Version 4.3 21-May-03
3409 ---------------------
3410
3411 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
3412 Makefile.
3413
3414 2. Some refactoring to improve the quality of the code:
3415
3416 (i) The utf8_table... variables are now declared "const".
3417
3418 (ii) The code for \cx, which used the "case flipping" table to upper case
3419 lower case letters, now just substracts 32. This is ASCII-specific,
3420 but the whole concept of \cx is ASCII-specific, so it seems
3421 reasonable.
3422
3423 (iii) PCRE was using its character types table to recognize decimal and
3424 hexadecimal digits in the pattern. This is silly, because it handles
3425 only 0-9, a-f, and A-F, but the character types table is locale-
3426 specific, which means strange things might happen. A private
3427 table is now used for this - though it costs 256 bytes, a table is
3428 much faster than multiple explicit tests. Of course, the standard
3429 character types table is still used for matching digits in subject
3430 strings against \d.
3431
3432 (iv) Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
3433 ending in _t are). So I've renamed it as ESC_tee.
3434
3435 3. The first argument for regexec() in the POSIX wrapper should have been
3436 defined as "const".
3437
3438 4. Changed pcretest to use malloc() for its buffers so that they can be
3439 Electric Fenced for debugging.
3440
3441 5. There were several places in the code where, in UTF-8 mode, PCRE would try
3442 to read one or more bytes before the start of the subject string. Often this
3443 had no effect on PCRE's behaviour, but in some circumstances it could
3444 provoke a segmentation fault.
3445
3446 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
3447 to try to read one or more bytes before the start of the subject string.
3448
3449 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
3450 UTF-8 support could misbehave in various ways if the subject string
3451 contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
3452 area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
3453 back over UTF-8 characters.)
3454
3455
3456 Version 4.2 14-Apr-03
3457 ---------------------
3458
3459 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
3460
3461 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
3462 [ON_WINDOWS]: new variable, "#" on non-Windows platforms
3463 [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
3464 [WIN_PREFIX]: new variable, "cyg" for Cygwin
3465 * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
3466 and BUILD_EXEEXT
3467 Note: automatic setting of the BUILD variables is not yet working
3468 set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
3469 compile-time but not at link-time
3470 [LINK]: use for linking executables only
3471 make different versions for Windows and non-Windows
3472 [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
3473 libraries
3474 [LINK_FOR_BUILD]: new variable
3475 [OBJEXT]: use throughout
3476 [EXEEXT]: use throughout
3477 <winshared>: new target
3478 <wininstall>: new target
3479 <dftables.o>: use native compiler
3480 <dftables>: use native linker
3481 <install>: handle Windows platform correctly
3482 <clean>: ditto
3483 <check>: ditto
3484 copy DLL to top builddir before testing
3485
3486 As part of these changes, -no-undefined was removed again. This was reported
3487 to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
3488 in any case.
3489
3490 3. Some tidies to get rid of compiler warnings:
3491
3492 . In the match_data structure, match_limit was an unsigned long int, whereas
3493 match_call_count was an int. I've made them both unsigned long ints.
3494
3495 . In pcretest the fact that a const uschar * doesn't automatically cast to
3496 a void * provoked a warning.
3497
3498 . Turning on some more compiler warnings threw up some "shadow" variables
3499 and a few more missing casts.
3500
3501 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
3502 option, a class that contained a single character with a value between 128
3503 and 255 (e.g. /[\xFF]/) caused PCRE to crash.
3504
3505 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
3506 option, a class that contained several characters, but with at least one
3507 whose value was between 128 and 255 caused PCRE to crash.
3508
3509
3510 Version 4.1 12-Mar-03
3511 ---------------------
3512
3513 1. Compiling with gcc -pedantic found a couple of places where casts were
3514 needed, and a string in dftables.c that was longer than standard compilers are
3515 required to support.
3516
3517 2. Compiling with Sun's compiler found a few more places where the code could
3518 be tidied up in order to avoid warnings.
3519
3520 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
3521 first of these names is deprecated in the latest Autoconf in favour of the name
3522 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
3523 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
3524 analogy I have changed it to CFLAGS_FOR_BUILD.
3525
3526 4. Added -no-undefined to the linking command in the Makefile, because this is
3527 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
3528 linking step for the pcreposix library.
3529
3530 5. PCRE was failing to diagnose the case of two named groups with the same
3531 name.
3532
3533 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
3534 literal character that is needed in the subject for a match, and scans along to
3535 ensure that it is present before embarking on the full matching process. This
3536 saves time in cases of nested unlimited repeats that are never going to match.
3537 Problem: the scan can take a lot of time if the subject is very long (e.g.
3538 megabytes), thus penalizing straightforward matches. It is now done only if the
3539 amount of subject to be scanned is less than 1000 bytes.
3540
3541 7. A lesser problem with the same optimization is that it was recording the
3542 first character of an anchored pattern as "needed", thus provoking a search
3543 right along the subject, even when the first match of the pattern was going to
3544 fail. The "needed" character is now not set for anchored patterns, unless it
3545 follows something in the pattern that is of non-fixed length. Thus, it still
3546 fulfils its original purpose of finding quick non-matches in cases of nested
3547 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
3548
3549
3550 Version 4.0 17-Feb-03
3551 ---------------------
3552
3553 1. If a comment in an extended regex that started immediately after a meta-item
3554 extended to the end of string, PCRE compiled incorrect data. This could lead to
3555 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
3556
3557 2. Moved to autoconf 2.53 and libtool 1.4.2.
3558
3559 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
3560 the special perltest8 script is no longer needed - all the tests can be run
3561 from a single perltest script.
3562
3563 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
3564 by \s. It has now been removed in PCRE. This means it isn't recognized as
3565 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
3566 class [:space:] *does* include VT, thereby creating a mess.
3567
3568 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
3569 space and tab.
3570
3571 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
3572 its new features into the main test script, reducing the number of scripts.
3573
3574 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
3575 were backward compatible, and made the (?i) apply to the whole pattern, as if
3576 /i were given. Now it behaves more logically, and applies the option setting
3577 only to what follows. PCRE has been changed to follow suit. However, if it
3578 finds options settings right at the start of the pattern, it extracts them into
3579 the global options, as before. Thus, they show up in the info data.
3580
3581 8. Added support for the \Q...\E escape sequence. Characters in between are
3582 treated as literals. This is slightly different from Perl in that $ and @ are
3583 also handled as literals inside the quotes. In Perl, they will cause variable
3584 interpolation. Note the following examples:
3585
3586 Pattern PCRE matches Perl matches
3587
3588 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
3589 \Qabc\$xyz\E abc\$xyz abc\$xyz
3590 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
3591
3592 For compatibility with Perl, \Q...\E sequences are recognized inside character
3593 classes as well as outside them.
3594
3595 9. Re-organized 3 code statements in pcretest to avoid "overflow in
3596 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
3597 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
3598 signed/unsigned warnings.
3599
3600 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
3601 option for pcretest, so I've replaced it by a simple function that does just
3602 that job.
3603
3604 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
3605 "pcregrep -".
3606
3607 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
3608 Java package. This provides some syntactic sugar for simple cases of what my
3609 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
3610 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
3611 item, you can use this simplified notation. Note that only makes sense with
3612 greedy quantifiers. Consequently, the use of the possessive quantifier forces
3613 greediness, whatever the setting of the PCRE_UNGREEDY option.
3614
3615 13. A change of greediness default within a pattern was not taking effect at
3616 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
3617 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
3618 was abstracted outside.
3619
3620 14. PCRE now supports the \G assertion. It is true when the current matching
3621 position is at the start point of the match. This differs from \A when the
3622 starting offset is non-zero. Used with the /g option of pcretest (or similar
3623 code), it works in the same way as it does for Perl's /g option. If all
3624 alternatives of a regex begin with \G, the expression is anchored to the start
3625 match position, and the "anchored" flag is set in the compiled expression.
3626
3627 15. Some bugs concerning the handling of certain option changes within patterns
3628 have been fixed. These applied to options other than (?ims). For example,
3629 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
3630 been the other way round. Some of this was related to change 7 above.
3631
3632 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
3633 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
3634 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
3635 POSIX classes only within a class (e.g. /[[:alpha:]]/).
3636
3637 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
3638 mode. Unlike ".", it always matches newline, whatever the setting of
3639 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
3640 assertions. Perl allows it, but it doesn't (in general) work because it can't
3641 calculate the length of the lookbehind. At least, that's the case for Perl
3642 5.8.0 - I've been told they are going to document that it doesn't work in
3643 future.
3644
3645 18. Added an error diagnosis for escapes that PCRE does not support: these are
3646 \L, \l, \N, \P, \p, \U, \u, and \X.
3647
3648 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
3649 reading past the end of the pattern in cases such as /[abcd/.
3650
3651 20. PCRE was getting more memory than necessary for patterns with classes that
3652 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
3653
3654 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
3655 compiling PCRE for use with Virtual Pascal.
3656
3657 22. Small fix to the Makefile to make it work properly if the build is done
3658 outside the source tree.
3659
3660 23. Added a new extension: a condition to go with recursion. If a conditional
3661 subpattern starts with (?(R) the "true" branch is used if recursion has
3662 happened, whereas the "false" branch is used only at the top level.
3663
3664 24. When there was a very long string of literal characters (over 255 bytes
3665 without UTF support, over 250 bytes with UTF support), the computation of how
3666 much memory was required could be incorrect, leading to segfaults or other
3667 strange effects.
3668
3669 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
3670 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
3671 there was a subsequent back reference to those brackets. This meant that, for
3672 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
3673 possible to check for precisely this case. All we can do is abandon the
3674 optimization if .* occurs inside capturing brackets when there are any back
3675 references whatsoever. (See below for a better fix that came later.)
3676
3677 26. The handling of the optimization for finding the first character of a
3678 non-anchored pattern, and for finding a character that is required later in the
3679 match were failing in some cases. This didn't break the matching; it just
3680 failed to optimize when it could. The way this is done has been re-implemented.
3681
3682 27. Fixed typo in error message for invalid (?R item (it said "(?p").
3683
3684 28. Added a new feature that provides some of the functionality that Perl
3685 provides with (?{...}). The facility is termed a "callout". The way it is done
3686 in PCRE is for the caller to provide an optional function, by setting
3687 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
3688 global variable. By default it is unset, which disables all calling out. To get
3689 the function called, the regex must include (?C) at appropriate points. This
3690 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
3691 This provides a means of identifying different callout points. When PCRE
3692 reaches such a point in the regex, if pcre_callout has been set, the external
3693 function is called. It is provided with data in a structure called
3694 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
3695 matching continues; if it returns a non-zero value, the match at the current
3696 point fails. However, backtracking will occur if possible. [This was changed
3697 later and other features added - see item 49 below.]
3698
3699 29. pcretest is upgraded to test the callout functionality. It provides a
3700 callout function that displays information. By default, it shows the start of
3701 the match and the current position in the text. There are some new data escapes
3702 to vary what happens:
3703
3704 \C+ in addition, show current contents of captured substrings
3705 \C- do not supply a callout function
3706 \C!n return 1 when callout number n is reached
3707 \C!n!m return 1 when callout number n is reached for the mth time
3708
3709 30. If pcregrep was called with the -l option and just a single file name, it
3710 output "<stdin>" if a match was found, instead of the file name.
3711
3712 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
3713 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
3714 pcre_exec(). This saves a malloc/free per call. The default value of
3715 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
3716 when configuring.
3717
3718 32. The default maximum size of a compiled pattern is 64K. There have been a
3719 few cases of people hitting this limit. The code now uses macros to handle the
3720 storing of links as offsets within the compiled pattern. It defaults to 2-byte
3721 links, but this can be changed to 3 or 4 bytes by --with-link-size when
3722 configuring. Tests 2 and 5 work only with 2-byte links because they output
3723 debugging information about compiled patterns.
3724
3725 33. Internal code re-arrangements:
3726
3727 (a) Moved the debugging function for printing out a compiled regex into
3728 its own source file (printint.c) and used #include to pull it into
3729 pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
3730 separate copies.
3731
3732 (b) Defined the list of op-code names for debugging as a macro in
3733 internal.h so that it is next to the definition of the opcodes.
3734
3735 (c) Defined a table of op-code lengths for simpler skipping along compiled
3736 code. This is again a macro in internal.h so that it is next to the
3737 definition of the opcodes.
3738
3739 34. Added support for recursive calls to individual subpatterns, along the
3740 lines of Robin Houston's patch (but implemented somewhat differently).
3741
3742 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
3743 allow it to read and process whole directories in Win32. This code was
3744 contributed by Lionel Fourquaux; it has not been tested by me.
3745
3746 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
3747 used to name a group. Names consist of alphanumerics and underscores, and must
3748 be unique. Back references use the syntax (?P=name) and recursive calls use
3749 (?P>name) which is a PCRE extension to the Python extension. Groups still have
3750 numbers. The function pcre_fullinfo() can be used after compilation to extract
3751 a name/number map. There are three relevant calls:
3752
3753 PCRE_INFO_NAMEENTRYSIZE yields the size of each entry in the map
3754 PCRE_INFO_NAMECOUNT yields the number of entries
3755 PCRE_INFO_NAMETABLE yields a pointer to the map.
3756
3757 The map is a vector of fixed-size entries. The size of each entry depends on
3758 the length of the longest name used. The first two bytes of each entry are the
3759 group number, most significant byte first. There follows the corresponding
3760 name, zero terminated. The names are in alphabetical order.
3761
3762 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
3763 case instead of 255. Making it the same both with and without UTF-8 support
3764 means that the same test output works with both.
3765
3766 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
3767 calling malloc() with a zero argument.
3768
3769 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
3770 optimization. I've improved things by keeping a bitmap of backreferences with
3771 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
3772 fact referenced, the optimization can be applied. It is unlikely that a
3773 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
3774 the match to follow \n) will appear inside brackets with a number greater than
3775 31, but if it does, any back reference > 31 suppresses the optimization.
3776
3777 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
3778 of disabling numbered capturing parentheses. Any opening parenthesis that is
3779 not followed by ? behaves as if it were followed by ?: but named parentheses
3780 can still be used for capturing (and they will acquire numbers in the usual
3781 way).
3782
3783 41. Redesigned the return codes from the match() function into yes/no/error so
3784 that errors can be passed back from deep inside the nested calls. A malloc
3785 failure while inside a recursive subpattern call now causes the
3786 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
3787
3788 42. It is now possible to set a limit on the number of times the match()
3789 function is called in a call to pcre_exec(). This facility makes it possible to
3790 limit the amount of recursion and backtracking, though not in a directly
3791 obvious way, because the match() function is used in a number of different
3792 circumstances. The count starts from zero for each position in the subject
3793 string (for non-anchored patterns). The default limit is, for compatibility, a
3794 large number, namely 10 000 000. You can change this in two ways:
3795
3796 (a) When configuring PCRE before making, you can use --with-match-limit=n
3797 to set a default value for the compiled library.
3798
3799 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
3800 a different value is set. See 45 below.
3801
3802 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
3803
3804 43. Added a new function pcre_config(int, void *) to enable run-time extraction
3805 of things that can be changed at compile time. The first argument specifies
3806 what is wanted and the second points to where the information is to be placed.
3807 The current list of available information is:
3808
3809 PCRE_CONFIG_UTF8
3810
3811 The output is an integer that is set to one if UTF-8 support is available;
3812 otherwise it is set to zero.
3813
3814 PCRE_CONFIG_NEWLINE
3815
3816 The output is an integer that it set to the value of the code that is used for
3817 newline. It is either LF (10) or CR (13).
3818
3819 PCRE_CONFIG_LINK_SIZE
3820
3821 The output is an integer that contains the number of bytes used for internal
3822 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
3823
3824 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
3825
3826 The output is an integer that contains the threshold above which the POSIX
3827 interface uses malloc() for output vectors. See item 31 above.
3828
3829 PCRE_CONFIG_MATCH_LIMIT
3830
3831 The output is an unsigned integer that contains the default limit of the number
3832 of match() calls in a pcre_exec() execution. See 42 above.
3833
3834 44. pcretest has been upgraded by the addition of the -C option. This causes it
3835 to extract all the available output from the new pcre_config() function, and to
3836 output it. The program then exits immediately.
3837
3838 45. A need has arisen to pass over additional data with calls to pcre_exec() in
3839 order to support additional features. One way would have been to define
3840 pcre_exec2() (for example) with extra arguments, but this would not have been
3841 extensible, and would also have required all calls to the original function to
3842 be mapped to the new one. Instead, I have chosen to extend the mechanism that
3843 is used for passing in "extra" data from pcre_study().
3844
3845 The pcre_extra structure is now exposed and defined in pcre.h. It currently
3846 contains the following fields:
3847
3848 flags a bitmap indicating which of the following fields are set
3849 study_data opaque data from pcre_study()
3850 match_limit a way of specifying a limit on match() calls for a specific
3851 call to pcre_exec()
3852 callout_data data for callouts (see 49 below)
3853
3854 The flag bits are also defined in pcre.h, and are
3855
3856 PCRE_EXTRA_STUDY_DATA
3857 PCRE_EXTRA_MATCH_LIMIT
3858 PCRE_EXTRA_CALLOUT_DATA
3859
3860 The pcre_study() function now returns one of these new pcre_extra blocks, with
3861 the actual study data pointed to by the study_data field, and the
3862 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
3863 before. That is, this change is entirely upwards-compatible and requires no
3864 change to existing code.
3865
3866 If you want to pass in additional data to pcre_exec(), you can either place it
3867 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
3868 block.
3869
3870 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
3871 data string contains the escape sequence \M, pcretest calls pcre_exec() several
3872 times with different match limits, until it finds the minimum value needed for
3873 pcre_exec() to complete. The value is then output. This can be instructive; for
3874 most simple matches the number is quite small, but for pathological cases it
3875 gets very large very quickly.
3876
3877 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
3878 returns the size of the data block pointed to by the study_data field in a
3879 pcre_extra block, that is, the value that was passed as the argument to
3880 pcre_malloc() when PCRE was getting memory in which to place the information
3881 created by pcre_study(). The fourth argument should point to a size_t variable.
3882 pcretest has been extended so that this information is shown after a successful
3883 pcre_study() call when information about the compiled regex is being displayed.
3884
3885 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
3886 because what follows is always an absolute path. (Later: it turns out that this
3887 is more than cosmetic for MinGW, because it doesn't like empty path
3888 components.)
3889
3890 49. Some changes have been made to the callout feature (see 28 above):
3891
3892 (i) A callout function now has three choices for what it returns:
3893
3894 0 => success, carry on matching
3895 > 0 => failure at this point, but backtrack if possible
3896 < 0 => serious error, return this value from pcre_exec()
3897
3898 Negative values should normally be chosen from the set of PCRE_ERROR_xxx
3899 values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
3900 "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
3901 use by callout functions. It will never be used by PCRE itself.
3902
3903 (ii) The pcre_extra structure (see 45 above) has a void * field called
3904 callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
3905 pcre_callout_block structure has a field of the same name. The contents of
3906 the field passed in the pcre_extra structure are passed to the callout
3907 function in the corresponding field in the callout block. This makes it
3908 easier to use the same callout-containing regex from multiple threads. For
3909 testing, the pcretest program has a new data escape
3910
3911 \C*n pass the number n (may be negative) as callout_data
3912
3913 If the callout function in pcretest receives a non-zero value as
3914 callout_data, it returns that value.
3915
3916 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
3917 there were some redundant $(CFLAGS) in commands that are now specified as
3918 $(LINK), which already includes $(CFLAGS).
3919
3920 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
3921 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
3922 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
3923 one-byte characters throughout. Note that case-insensitive matching applies
3924 only to characters whose values are less than 256. PCRE doesn't support the
3925 notion of cases for higher-valued characters.
3926
3927 (i) A character class whose characters are all within 0-255 is handled as
3928 a bit map, and the map is inverted for negative classes. Previously, a
3929 character > 255 always failed to match such a class; however it should
3930 match if the class was a negative one (e.g. [^ab]). This has been fixed.
3931
3932 (ii) A negated character class with a single character < 255 is coded as
3933 "not this character" (OP_NOT). This wasn't working properly when the test
3934 character was multibyte, either singly or repeated.
3935
3936 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
3937 mode, for example: \x{100}{2,3}.
3938
3939 (iv) The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
3940 singly or repeated) now correctly test multibyte characters. However,
3941 PCRE doesn't recognize any characters with values greater than 255 as
3942 digits, spaces, or word characters. Such characters always match \D, \S,
3943 and \W, and never match \d, \s, or \w.
3944
3945 (v) Classes may now contain characters and character ranges with values
3946 greater than 255. For example: [ab\x{100}-\x{400}].
3947
3948 (vi) pcregrep now has a --utf-8 option (synonym -u) which makes it call
3949 PCRE in UTF-8 mode.
3950
3951 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
3952 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
3953 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
3954 value.)
3955
3956 53. The single man page has become too large. I have therefore split it up into
3957 a number of separate man pages. These also give rise to individual HTML pages;
3958 these are now put in a separate directory, and there is an index.html page that
3959 lists them all. Some hyperlinking between the pages has been installed.
3960
3961 54. Added convenience functions for handling named capturing parentheses.
3962
3963 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
3964 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
3965 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
3966 are faulted.
3967
3968 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
3969 calling configure. These values are used when compiling the dftables.c program
3970 which is run to generate the source of the default character tables. They
3971 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
3972 you will need to set these values.
3973
3974 57. Updated the building process for Windows DLL, as provided by Fred Cox.
3975
3976
3977 Version 3.9 02-Jan-02
3978 ---------------------
3979
3980 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
3981
3982 2. If --disable-static was given, the building process failed when trying to
3983 build pcretest and pcregrep. (For some reason it was using libtool to compile
3984 them, which is not right, as they aren't part of the library.)
3985
3986
3987 Version 3.8 18-Dec-01
3988 ---------------------
3989
3990 1. The experimental UTF-8 code was completely screwed up. It was packing the
3991 bytes in the wrong order. How dumb can you get?
3992
3993
3994 Version 3.7 29-Oct-01
3995 ---------------------
3996
3997 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
3998 This caused pcretest, when used on the test data, to segfault. Unfortunately,
3999 this didn't happen under Solaris 8, where I normally test things.
4000
4001 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
4002 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
4003 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
4004 this fix an hour or so after the initial 3.7 release.)
4005
4006
4007 Version 3.6 23-Oct-01
4008 ---------------------
4009
4010 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
4011 offsets passed as NULL with zero offset count.
4012
4013 2. The config.guess and config.sub files had not been updated when I moved to
4014 the latest autoconf.
4015
4016
4017 Version 3.5 15-Aug-01
4018 ---------------------
4019
4020 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
4021 had been forgotten.
4022
4023 2. By using declared but undefined structures, we can avoid using "void"
4024 definitions in pcre.h while keeping the internal definitions of the structures
4025 private.
4026
4027 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
4028 user point of view, this means that both static and shared libraries are built
4029 by default, but this can be individually controlled. More of the work of
4030 handling this static/shared cases is now inside libtool instead of PCRE's make
4031 file.
4032
4033 4. The pcretest utility is now installed along with pcregrep because it is
4034 useful for users (to test regexs) and by doing this, it automatically gets
4035 relinked by libtool. The documentation has been turned into a man page, so
4036 there are now .1, .txt, and .html versions in /doc.
4037
4038 5. Upgrades to pcregrep:
4039 (i) Added long-form option names like gnu grep.
4040 (ii) Added --help to list all options with an explanatory phrase.
4041 (iii) Added -r, --recursive to recurse into sub-directories.
4042 (iv) Added -f, --file to read patterns from a file.
4043
4044 6. pcre_exec() was referring to its "code" argument before testing that
4045 argument for NULL (and giving an error if it was NULL).
4046
4047 7. Upgraded Makefile.in to allow for compiling in a different directory from
4048 the source directory.
4049
4050 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
4051 options bits, the pointer it was passed was to an int instead of to an unsigned
4052 long int. This mattered only on 64-bit systems.
4053
4054 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
4055 generated) instead of pcre.in, which it its source. Also made the same change
4056 in several of the .c files.
4057
4058 10. A new release of gcc defines printf() as a macro, which broke pcretest
4059 because it had an ifdef in the middle of a string argument for printf(). Fixed
4060 by using separate calls to printf().
4061
4062 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
4063 script, to force use of CR or LF instead of \n in the source. On non-Unix
4064 systems, the value can be set in config.h.
4065
4066 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
4067 absolute limit. Changed the text of the error message to make this clear, and
4068 likewise updated the man page.
4069
4070 13. The limit of 99 on the number of capturing subpatterns has been removed.
4071 The new limit is 65535, which I hope will not be a "real" limit.
4072
4073
4074 Version 3.4 22-Aug-00
4075 ---------------------
4076
4077 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
4078
4079 2. Diagnose condition (?(0) as an error instead of crashing on matching.
4080
4081
4082 Version 3.3 01-Aug-00
4083 ---------------------
4084
4085 1. If an octal character was given, but the value was greater than \377, it
4086 was not getting masked to the least significant bits, as documented. This could
4087 lead to crashes in some systems.
4088
4089 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
4090 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
4091
4092 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
4093 These just pass their arguments on to (pcre_free)(), but they are provided
4094 because some uses of PCRE bind it to non-C systems that can call its functions,
4095 but cannot call free() or pcre_free() directly.
4096
4097 4. Add "make test" as a synonym for "make check". Corrected some comments in
4098 the Makefile.
4099
4100 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
4101 Makefile.
4102
4103 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
4104 command called pgrep for grepping around the active processes.
4105
4106 7. Added the beginnings of support for UTF-8 character strings.
4107
4108 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
4109 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
4110 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
4111 out for the ar command.)
4112
4113
4114 Version 3.2 12-May-00
4115 ---------------------
4116
4117 This is purely a bug fixing release.
4118
4119 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
4120 of ZA. This was just one example of several cases that could provoke this bug,
4121 which was introduced by change 9 of version 2.00. The code for breaking
4122 infinite loops after an iteration that matches an empty string was't working
4123 correctly.
4124
4125 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
4126 when matched against abbab (for example). After matching an empty string, it
4127 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
4128 caused it to match further down the string than it should.
4129
4130 3. The code contained an inclusion of sys/types.h. It isn't clear why this
4131 was there because it doesn't seem to be needed, and it causes trouble on some
4132 systems, as it is not a Standard C header. It has been removed.
4133
4134 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
4135 were reported on the Macintosh. The changes were from
4136
4137 while ((c = *(++ptr)) != 0 && c != '\n');
4138 to
4139 while ((c = *(++ptr)) != 0 && c != '\n') ;
4140
4141 Totally extraordinary, but if that's what it takes...
4142
4143 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
4144 available. Added HAVE_BCOPY and an autoconf test for it; if neither
4145 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
4146 assumes the way PCRE uses memmove() (always moving upwards).
4147
4148 6. PCRE is being used in one environment where strchr() is not available. There
4149 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
4150 faster code anyway.
4151
4152
4153 Version 3.1 09-Feb-00
4154 ---------------------
4155
4156 The only change in this release is the fixing of some bugs in Makefile.in for
4157 the "install" target:
4158
4159 (1) It was failing to install pcreposix.h.
4160
4161 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
4162
4163
4164 Version 3.0 01-Feb-00
4165 ---------------------
4166
4167 1. Add support for the /+ modifier to perltest (to output $` like it does in
4168 pcretest).
4169
4170 2. Add support for the /g modifier to perltest.
4171
4172 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
4173 matches null strings.
4174
4175 4. Fix perltest so that it doesn't do unwanted things when fed an empty
4176 pattern. Perl treats empty patterns specially - it reuses the most recent
4177 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
4178 effect.
4179
4180 5. The POSIX interface was broken in that it was just handing over the POSIX
4181 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
4182 required a bigger vector, with some working space on the end. This means that
4183 the POSIX wrapper now has to get and free some memory, and copy the results.
4184
4185 6. Added some simple autoconf support, placing the test data and the
4186 documentation in separate directories, re-organizing some of the
4187 information files, and making it build pcre-config (a GNU standard). Also added
4188 libtool support for building PCRE as a shared library, which is now the
4189 default.
4190
4191 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
4192 09 are not valid octal constants. Single digits will be used for minor values
4193 less than 10.
4194
4195 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
4196 existing programs that set these in the POSIX interface can use PCRE without
4197 modification.
4198
4199 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
4200 return all that pcre_info() returns, plus additional data. The pcre_info()
4201 function is retained for compatibility, but is considered to be obsolete.
4202
4203 10. Added experimental recursion feature (?R) to handle one common case that
4204 Perl 5.6 will be able to do with (?p{...}).
4205
4206 11. Added support for POSIX character classes like [:alpha:], which Perl is
4207 adopting.
4208
4209
4210 Version 2.08 31-Aug-99
4211 ----------------------
4212
4213 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
4214 trying to match at the startoffset position, but instead was moving forward to
4215 the next newline as if a previous match had failed.
4216
4217 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
4218 and could get into a loop if a null string was matched other than at the start
4219 of the subject.
4220
4221 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
4222 be distinguished at compile time, and for completeness also added PCRE_DATE.
4223
4224 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
4225 in GnuWin32 environments.
4226
4227
4228 Version 2.07 29-Jul-99
4229 ----------------------
4230
4231 1. The documentation is now supplied in plain text form and HTML as well as in
4232 the form of man page sources.
4233
4234 2. C++ compilers don't like assigning (void *) values to other pointer types.
4235 In particular this affects malloc(). Although there is no problem in Standard
4236 C, I've put in casts to keep C++ compilers happy.
4237
4238 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
4239 should be (const char *).
4240
4241 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
4242 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
4243 However, I haven't made this a standard facility. The documentation doesn't
4244 mention it, and the Makefile doesn't support it.
4245
4246 5. The Makefile now contains an "install" target, with editable destinations at
4247 the top of the file. The pcretest program is not installed.
4248
4249 6. pgrep -V now gives the PCRE version number and date.
4250
4251 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
4252 causing the entire string to be ignored, instead of just the last character.
4253
4254 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
4255 non-matching string, it can take a very, very long time, even for strings of
4256 quite modest length, because of the nested recursion. PCRE now does better in
4257 some of these cases. It does this by remembering the last required literal
4258 character in the pattern, and pre-searching the subject to ensure it is present
4259 before running the real match. In other words, it applies a heuristic to detect
4260 some types of certain failure quickly, and in the above example, if presented
4261 with a string that has no trailing " it gives "no match" very quickly.
4262
4263 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
4264 other alternatives are tried instead.
4265
4266
4267 Version 2.06 09-Jun-99
4268 ----------------------
4269
4270 1. Change pcretest's output for amount of store used to show just the code
4271 space, because the remainder (the data block) varies in size between 32-bit and
4272 64-bit systems.
4273
4274 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
4275 start matching at. This allows lookbehinds to work when searching for multiple
4276 occurrences in a string.
4277
4278 3. Added additional options to pcretest for testing multiple occurrences:
4279
4280 /+ outputs the rest of the string that follows a match
4281 /g loops for multiple occurrences, using the new startoffset argument
4282 /G loops for multiple occurrences by passing an incremented pointer
4283
4284 4. PCRE wasn't doing the "first character" optimization for patterns starting
4285 with \b or \B, though it was doing it for other lookbehind assertions. That is,
4286 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
4287 the letter 'x'. On long subject strings, this gives a significant speed-up.
4288
4289
4290 Version 2.05 21-Apr-99
4291 ----------------------
4292
4293 1. Changed the type of magic_number from int to long int so that it works
4294 properly on 16-bit systems.
4295
4296 2. Fixed a bug which caused patterns starting with .* not to work correctly
4297 when the subject string contained newline characters. PCRE was assuming
4298 anchoring for such patterns in all cases, which is not correct because .* will
4299 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
4300 DOTALL is set at top level; otherwise it knows that patterns starting with .*
4301 must be retried after every newline in the subject.
4302
4303
4304 Version 2.04 18-Feb-99
4305 ----------------------
4306
4307 1. For parenthesized subpatterns with repeats whose minimum was zero, the
4308 computation of the store needed to hold the pattern was incorrect (too large).
4309 If such patterns were nested a few deep, this could multiply and become a real
4310 problem.
4311
4312 2. Added /M option to pcretest to show the memory requirement of a specific
4313 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
4314
4315 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
4316 compiled in such a way that the backtracking after subsequent failure was
4317 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
4318 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
4319
4320
4321 Version 2.03 02-Feb-99
4322 ----------------------
4323
4324 1. Fixed typo and small mistake in man page.
4325
4326 2. Added 4th condition (GPL supersedes if conflict) and created separate
4327 LICENCE file containing the conditions.
4328
4329 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
4330 Perl, that is the internal \ allows the delimiter to be included in the
4331 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
4332 the final delimiter, add \ to the end of the pattern (to test the error).
4333
4334 4. Added the convenience functions for extracting substrings after a successful
4335 match. Updated pcretest to make it able to test these functions.
4336
4337
4338 Version 2.02 14-Jan-99
4339 ----------------------
4340
4341 1. Initialized the working variables associated with each extraction so that
4342 their saving and restoring doesn't refer to uninitialized store.
4343
4344 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
4345 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
4346 fix the problem.
4347
4348 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
4349 calls, and wasn't printing the correct value for compiling calls. Increased the
4350 default value of LOOPREPEAT, and the number of significant figures in the
4351 times.
4352
4353 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
4354
4355 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
4356 a building problem on Windows NT with a FAT file system.
4357
4358
4359 Version 2.01 21-Oct-98
4360 ----------------------
4361
4362 1. Changed the API for pcre_compile() to allow for the provision of a pointer
4363 to character tables built by pcre_maketables() in the current locale. If NULL
4364 is passed, the default tables are used.
4365
4366
4367 Version 2.00 24-Sep-98
4368 ----------------------
4369
4370 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
4371 it any more.
4372
4373 2. Allow quantification of (?>) groups, and make it work correctly.
4374
4375 3. The first character computation wasn't working for (?>) groups.
4376
4377 4. Correct the implementation of \Z (it is permitted to match on the \n at the
4378 end of the subject) and add 5.005's \z, which really does match only at the
4379 very end of the subject.
4380
4381 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
4382
4383 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
4384 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
4385 localized options. All options to pcre_study() were also removed.
4386
4387 7. Add other new features from 5.005:
4388
4389 $(?<= positive lookbehind
4390 $(?<! negative lookbehind
4391 (?imsx-imsx) added the unsetting capability
4392 such a setting is global if at outer level; local otherwise
4393 (?imsx-imsx:) non-capturing groups with option setting
4394 (?(cond)re|re) conditional pattern matching
4395
4396 A backreference to itself in a repeated group matches the previous
4397 captured string.
4398
4399 8. General tidying up of studying (both automatic and via "study")
4400 consequential on the addition of new assertions.
4401
4402 9. As in 5.005, unlimited repeated groups that could match an empty substring
4403 are no longer faulted at compile time. Instead, the loop is forcibly broken at
4404 runtime if any iteration does actually match an empty substring.
4405
4406 10. Include the RunTest script in the distribution.
4407
4408 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
4409 discrepancies, some of which were old and were also with respect to 5.004. They
4410 have now been fixed.
4411
4412
4413 Version 1.09 28-Apr-98
4414 ----------------------
4415
4416 1. A negated single character class followed by a quantifier with a minimum
4417 value of one (e.g. [^x]{1,6} ) was not compiled correctly. This could lead to
4418 program crashes, or just wrong answers. This did not apply to negated classes
4419 containing more than one character, or to minima other than one.
4420
4421
4422 Version 1.08 27-Mar-98
4423 ----------------------
4424
4425 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
4426
4427 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
4428 latter must appear before anything that relies on it in the pattern.
4429
4430
4431 Version 1.07 16-Feb-98
4432 ----------------------
4433
4434 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
4435 repeat of a potentially empty string).
4436
4437
4438 Version 1.06 23-Jan-98
4439 ----------------------
4440
4441 1. Added Markus Oberhumer's little patches for C++.
4442
4443 2. Literal strings longer than 255 characters were broken.
4444
4445
4446 Version 1.05 23-Dec-97
4447 ----------------------
4448
4449 1. Negated character classes containing more than one character were failing if
4450 PCRE_CASELESS was set at run time.
4451
4452
4453 Version 1.04 19-Dec-97
4454 ----------------------
4455
4456 1. Corrected the man page, where some "const" qualifiers had been omitted.
4457
4458 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
4459 input syntax.
4460
4461 3. Fixed memory leak which occurred when a regex with back references was
4462 matched with an offsets vector that wasn't big enough. The temporary memory
4463 that is used in this case wasn't being freed if the match failed.
4464
4465 4. Tidied pcretest to ensure it frees memory that it gets.
4466
4467 5. Temporary memory was being obtained in the case where the passed offsets
4468 vector was exactly big enough.
4469
4470 6. Corrected definition of offsetof() from change 5 below.
4471
4472 7. I had screwed up change 6 below and broken the rules for the use of
4473 setjmp(). Now fixed.
4474
4475
4476 Version 1.03 18-Dec-97
4477 ----------------------
4478
4479 1. A erroneous regex with a missing opening parenthesis was correctly
4480 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
4481 on some systems.
4482
4483 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
4484 it was reported that one broken compiler failed on the former because "code" is
4485 also an independent variable.
4486
4487 3. The erroneous regex a[]b caused an array overrun reference.
4488
4489 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
4490 fail on data ending with that character. (It was going on too far, and checking
4491 the next character, typically a binary zero.) This was specific to the
4492 optimized code for single-character negative classes.
4493
4494 5. Added a contributed patch from the TIN world which does the following:
4495
4496 + Add an undef for memmove, in case the the system defines a macro for it.
4497
4498 + Add a definition of offsetof(), in case there isn't one. (I don't know
4499 the reason behind this - offsetof() is part of the ANSI standard - but
4500 it does no harm).
4501
4502 + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
4503 most of the places where whitespace preceded '#'. I have given up and
4504 allowed the remaining 2 cases to be at the margin.
4505
4506 + Rename some variables in pcre to eliminate shadowing. This seems very
4507 pedantic, but does no harm, of course.
4508
4509 6. Moved the call to setjmp() into its own function, to get rid of warnings
4510 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
4511
4512 7. Constructs such as \d{8,} were compiling into the equivalent of
4513 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
4514 outcome, but in this particular case used more store than had been allocated,
4515 which caused the bug to be discovered because it threw up an internal error.
4516
4517 8. The debugging code in both pcre and pcretest for outputting the compiled
4518 form of a regex was going wrong in the case of back references followed by
4519 curly-bracketed repeats.
4520
4521
4522 Version 1.02 12-Dec-97
4523 ----------------------
4524
4525 1. Typos in pcre.3 and comments in the source fixed.
4526
4527 2. Applied a contributed patch to get rid of places where it used to remove
4528 'const' from variables, and fixed some signed/unsigned and uninitialized
4529 variable warnings.
4530
4531 3. Added the "runtest" target to Makefile.
4532
4533 4. Set default compiler flag to -O2 rather than just -O.
4534
4535
4536 Version 1.01 19-Nov-97
4537 ----------------------
4538
4539 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
4540 like /([ab]*)*/, that is, for classes with more than one character in them.
4541
4542 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
4543 as /((?>a*))*/ (a PCRE_EXTRA facility).
4544
4545
4546 Version 1.00 18-Nov-97
4547 ----------------------
4548
4549 1. Added compile-time macros to support systems such as SunOS4 which don't have
4550 memmove() or strerror() but have other things that can be used instead.
4551
4552 2. Arranged that "make clean" removes the executables.
4553
4554
4555 Version 0.99 27-Oct-97
4556 ----------------------
4557
4558 1. Fixed bug in code for optimizing classes with only one character. It was
4559 initializing a 32-byte map regardless, which could cause it to run off the end
4560 of the memory it had got.
4561
4562 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
4563
4564
4565 Version 0.98 22-Oct-97
4566 ----------------------
4567
4568 1. Fixed bug in code for handling temporary memory usage when there are more
4569 back references than supplied space in the ovector. This could cause segfaults.
4570
4571
4572 Version 0.97 21-Oct-97
4573 ----------------------
4574
4575 1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
4576
4577 2. Optimized negated single characters not to use a bit map.
4578
4579 3. Brought error texts together as macro definitions; clarified some of them;
4580 fixed one that was wrong - it said "range out of order" when it meant "invalid
4581 escape sequence".
4582
4583 4. Changed some char * arguments to const char *.
4584
4585 5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
4586
4587 6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
4588 pcretest.
4589
4590
4591 Version 0.96 16-Oct-97
4592 ----------------------
4593
4594 1. Added a simple "pgrep" utility to the distribution.
4595
4596 2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
4597 unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
4598 where "ddd" means "one or more decimal digits".
4599
4600 3. Fixed serious bug. If a pattern had a back reference, but the call to
4601 pcre_exec() didn't supply a large enough ovector to record the related
4602 identifying subpattern, the match always failed. PCRE now remembers the number
4603 of the largest back reference, and gets some temporary memory in which to save
4604 the offsets during matching if necessary, in order to ensure that
4605 backreferences always work.
4606
4607 4. Increased the compatibility with Perl in a number of ways:
4608
4609 (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
4610 to request this handling. The option can be set at compile or exec time.
4611
4612 (b) $ matches before a terminating newline by default; an option
4613 PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
4614 mode). The option can be set at compile or exec time.
4615
4616 (c) The handling of \ followed by a digit other than 0 is now supposed to be
4617 the same as Perl's. If the decimal number it represents is less than 10
4618 or there aren't that many previous left capturing parentheses, an octal
4619 escape is read. Inside a character class, it's always an octal escape,
4620 even if it is a single digit.
4621
4622 (d) An escaped but undefined alphabetic character is taken as a literal,
4623 unless PCRE_EXTRA is set. Currently this just reserves the remaining
4624 escapes.
4625
4626 (e) {0} is now permitted. (The previous item is removed from the compiled
4627 pattern).
4628
4629 5. Changed all the names of code files so that the basic parts are no longer
4630 than 10 characters, and abolished the teeny "globals.c" file.
4631
4632 6. Changed the handling of character classes; they are now done with a 32-byte
4633 bit map always.
4634
4635 7. Added the -d and /D options to pcretest to make it possible to look at the
4636 internals of compilation without having to recompile pcre.
4637
4638
4639 Version 0.95 23-Sep-97
4640 ----------------------
4641
4642 1. Fixed bug in pre-pass concerning escaped "normal" characters such as \x5c or
4643 \x20 at the start of a run of normal characters. These were being treated as
4644 real characters, instead of the source characters being re-checked.
4645
4646
4647 Version 0.94 18-Sep-97
4648 ----------------------
4649
4650 1. The functions are now thread-safe, with the caveat that the global variables
4651 containing pointers to malloc() and free() or alternative functions are the
4652 same for all threads.
4653
4654 2. Get pcre_study() to generate a bitmap of initial characters for non-
4655 anchored patterns when this is possible, and use it if passed to pcre_exec().
4656
4657
4658 Version 0.93 15-Sep-97
4659 ----------------------
4660
4661 1. /(b)|(:+)/ was computing an incorrect first character.
4662
4663 2. Add pcre_study() to the API and the passing of pcre_extra to pcre_exec(),
4664 but not actually doing anything yet.
4665
4666 3. Treat "-" characters in classes that cannot be part of ranges as literals,
4667 as Perl does (e.g. [-az] or [az-]).
4668
4669 4. Set the anchored flag if a branch starts with .* or .*? because that tests
4670 all possible positions.
4671
4672 5. Split up into different modules to avoid including unneeded functions in a
4673 compiled binary. However, compile and exec are still in one module. The "study"
4674 function is split off.
4675
4676 6. The character tables are now in a separate module whose source is generated
4677 by an auxiliary program - but can then be edited by hand if required. There are
4678 now no calls to isalnum(), isspace(), isdigit(), isxdigit(), tolower() or
4679 toupper() in the code.
4680
4681 7. Turn the malloc/free funtions variables into pcre_malloc and pcre_free and
4682 make them global. Abolish the function for setting them, as the caller can now
4683 set them directly.
4684
4685
4686 Version 0.92 11-Sep-97
4687 ----------------------
4688
4689 1. A repeat with a fixed maximum and a minimum of 1 for an ordinary character
4690 (e.g. /a{1,3}/) was broken (I mis-optimized it).
4691
4692 2. Caseless matching was not working in character classes if the characters in
4693 the pattern were in upper case.
4694
4695 3. Make ranges like [W-c] work in the same way as Perl for caseless matching.
4696
4697 4. Make PCRE_ANCHORED public and accept as a compile option.
4698
4699 5. Add an options word to pcre_exec() and accept PCRE_ANCHORED and
4700 PCRE_CASELESS at run time. Add escapes \A and \I to pcretest to cause it to
4701 pass them.
4702
4703 6. Give an error if bad option bits passed at compile or run time.
4704
4705 7. Add PCRE_MULTILINE at compile and exec time, and (?m) as well. Add \M to
4706 pcretest to cause it to pass that flag.
4707
4708 8. Add pcre_info(), to get the number of identifying subpatterns, the stored
4709 options, and the first character, if set.
4710
4711 9. Recognize C+ or C{n,m} where n >= 1 as providing a fixed starting character.
4712
4713
4714 Version 0.91 10-Sep-97
4715 ----------------------
4716
4717 1. PCRE was failing to diagnose unlimited repeats of subpatterns that could
4718 match the empty string as in /(a*)*/. It was looping and ultimately crashing.
4719
4720 2. PCRE was looping on encountering an indefinitely repeated back reference to
4721 a subpattern that had matched an empty string, e.g. /(a|)\1*/. It now does what
4722 Perl does - treats the match as successful.
4723
4724 ****

Properties

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

  ViewVC Help
Powered by ViewVC 1.1.5