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

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log


Revision 741 - (show annotations)
Mon Oct 31 09:31:46 2011 UTC (3 years, 6 months ago) by zherczeg
File size: 217775 byte(s)
Error occurred while calculating annotation data.
Supporting OP_NCREF, OP_RREF and OP_NRREF by the JIT compiler
1 ChangeLog for PCRE
2 ------------------
3
4 Version 8.21
5 ------------
6
7 1. Updating the JIT compiler.
8
9 2. JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
10 are added as well.
11
12
13 Version 8.20 21-Oct-2011
14 ------------------------
15
16 1. Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
17 a POSIX class. After further experiments with Perl, which convinced me that
18 Perl has bugs and confusions, a closing square bracket is no longer allowed
19 in a POSIX name. This bug also affected patterns with classes that started
20 with full stops.
21
22 2. If a pattern such as /(a)b|ac/ is matched against "ac", there is no
23 captured substring, but while checking the failing first alternative,
24 substring 1 is temporarily captured. If the output vector supplied to
25 pcre_exec() was not big enough for this capture, the yield of the function
26 was still zero ("insufficient space for captured substrings"). This cannot
27 be totally fixed without adding another stack variable, which seems a lot
28 of expense for a edge case. However, I have improved the situation in cases
29 such as /(a)(b)x|abc/ matched against "abc", where the return code
30 indicates that fewer than the maximum number of slots in the ovector have
31 been set.
32
33 3. Related to (2) above: when there are more back references in a pattern than
34 slots in the output vector, pcre_exec() uses temporary memory during
35 matching, and copies in the captures as far as possible afterwards. It was
36 using the entire output vector, but this conflicts with the specification
37 that only 2/3 is used for passing back captured substrings. Now it uses
38 only the first 2/3, for compatibility. This is, of course, another edge
39 case.
40
41 4. Zoltan Herczeg's just-in-time compiler support has been integrated into the
42 main code base, and can be used by building with --enable-jit. When this is
43 done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
44 runtime --no-jit option is given.
45
46 5. When the number of matches in a pcre_dfa_exec() run exactly filled the
47 ovector, the return from the function was zero, implying that there were
48 other matches that did not fit. The correct "exactly full" value is now
49 returned.
50
51 6. If a subpattern that was called recursively or as a subroutine contained
52 (*PRUNE) or any other control that caused it to give a non-standard return,
53 invalid errors such as "Error -26 (nested recursion at the same subject
54 position)" or even infinite loops could occur.
55
56 7. If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
57 computing the minimum length on reaching *ACCEPT, and so ended up with the
58 wrong value of 1 rather than 0. Further investigation indicates that
59 computing a minimum subject length in the presence of *ACCEPT is difficult
60 (think back references, subroutine calls), and so I have changed the code
61 so that no minimum is registered for a pattern that contains *ACCEPT.
62
63 8. If (*THEN) was present in the first (true) branch of a conditional group,
64 it was not handled as intended. [But see 16 below.]
65
66 9. Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
67 Sheri Pierce.
68
69 10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
70 the first byte in a match must be "a".
71
72 11. Change 17 for 8.13 increased the recursion depth for patterns like
73 /a(?:.)*?a/ drastically. I've improved things by remembering whether a
74 pattern contains any instances of (*THEN). If it does not, the old
75 optimizations are restored. It would be nice to do this on a per-group
76 basis, but at the moment that is not feasible.
77
78 12. In some environments, the output of pcretest -C is CRLF terminated. This
79 broke RunTest's code that checks for the link size. A single white space
80 character after the value is now allowed for.
81
82 13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
83 For "fr", it uses the Windows-specific input and output files.
84
85 14. If (*THEN) appeared in a group that was called recursively or as a
86 subroutine, it did not work as intended. [But see next item.]
87
88 15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
89 pattern fragments (but not containing any | characters). If A and B are
90 matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
91 was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
92 D. In other words, Perl considers parentheses that do not contain any |
93 characters to be part of a surrounding alternative, whereas PCRE was
94 treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
95 differently. PCRE now behaves in the same way as Perl, except in the case
96 of subroutine/recursion calls such as (?1) which have in any case always
97 been different (but PCRE had them first :-).
98
99 16. Related to 15 above: Perl does not treat the | in a conditional group as
100 creating alternatives. Such a group is treated in the same way as an
101 ordinary group without any | characters when processing (*THEN). PCRE has
102 been changed to match Perl's behaviour.
103
104 17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
105 RunGrepTest script failed.
106
107 18. Change 22 for version 13 caused atomic groups to use more stack. This is
108 inevitable for groups that contain captures, but it can lead to a lot of
109 stack use in large patterns. The old behaviour has been restored for atomic
110 groups that do not contain any capturing parentheses.
111
112 19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
113 suppress the check for a minimum subject length at run time. (If it was
114 given to pcre_exec() or pcre_dfa_exec() it did work.)
115
116 20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
117 fail to work when decoding hex characters in data strings in EBCDIC
118 environments.
119
120 21. It appears that in at least one Mac OS environment, the isxdigit() function
121 is implemented as a macro that evaluates to its argument more than once,
122 contravening the C 90 Standard (I haven't checked a later standard). There
123 was an instance in pcretest which caused it to go wrong when processing
124 \x{...} escapes in subject strings. The has been rewritten to avoid using
125 things like p++ in the argument of isxdigit().
126
127
128 Version 8.13 16-Aug-2011
129 ------------------------
130
131 1. The Unicode data tables have been updated to Unicode 6.0.0.
132
133 2. Two minor typos in pcre_internal.h have been fixed.
134
135 3. Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
136 pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
137 in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
138
139 4. There were a number of related bugs in the code for matching backrefences
140 caselessly in UTF-8 mode when codes for the characters concerned were
141 different numbers of bytes. For example, U+023A and U+2C65 are an upper
142 and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
143 (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
144 code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
145 2-byte code at the end of the subject (it thought there wasn't enough data
146 left).
147
148 5. Comprehensive information about what went wrong is now returned by
149 pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
150 as the output vector has at least 2 elements. The offset of the start of
151 the failing character and a reason code are placed in the vector.
152
153 6. When the UTF-8 string check fails for pcre_compile(), the offset that is
154 now returned is for the first byte of the failing character, instead of the
155 last byte inspected. This is an incompatible change, but I hope it is small
156 enough not to be a problem. It makes the returned offset consistent with
157 pcre_exec() and pcre_dfa_exec().
158
159 7. pcretest now gives a text phrase as well as the error number when
160 pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
161 failure, the offset and reason code are output.
162
163 8. When \R was used with a maximizing quantifier it failed to skip backwards
164 over a \r\n pair if the subsequent match failed. Instead, it just skipped
165 back over a single character (\n). This seems wrong (because it treated the
166 two characters as a single entity when going forwards), conflicts with the
167 documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
168 behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
169 has been changed.
170
171 9. Some internal refactoring has changed the processing so that the handling
172 of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
173 time (the PCRE_DOTALL option was changed this way some time ago: version
174 7.7 change 16). This has made it possible to abolish the OP_OPT op code,
175 which was always a bit of a fudge. It also means that there is one less
176 argument for the match() function, which reduces its stack requirements
177 slightly. This change also fixes an incompatibility with Perl: the pattern
178 (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
179
180 10. More internal refactoring has drastically reduced the number of recursive
181 calls to match() for possessively repeated groups such as (abc)++ when
182 using pcre_exec().
183
184 11. While implementing 10, a number of bugs in the handling of groups were
185 discovered and fixed:
186
187 (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
188 (a|)*(?1) gave a compile-time internal error.
189 ((a|)+)+ did not notice that the outer group could match an empty string.
190 (^a|^)+ was not marked as anchored.
191 (.*a|.*)+ was not marked as matching at start or after a newline.
192
193 12. Yet more internal refactoring has removed another argument from the match()
194 function. Special calls to this function are now indicated by setting a
195 value in a variable in the "match data" data block.
196
197 13. Be more explicit in pcre_study() instead of relying on "default" for
198 opcodes that mean there is no starting character; this means that when new
199 ones are added and accidentally left out of pcre_study(), testing should
200 pick them up.
201
202 14. The -s option of pcretest has been documented for ages as being an old
203 synonym of -m (show memory usage). I have changed it to mean "force study
204 for every regex", that is, assume /S for every regex. This is similar to -i
205 and -d etc. It's slightly incompatible, but I'm hoping nobody is still
206 using it. It makes it easier to run collections of tests with and without
207 study enabled, and thereby test pcre_study() more easily. All the standard
208 tests are now run with and without -s (but some patterns can be marked as
209 "never study" - see 20 below).
210
211 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
212 restoration of the capturing data to the outer values was not happening
213 correctly.
214
215 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
216 empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
217 pattern had matched an empty string, and so incorrectly returned a no
218 match.
219
220 17. There was optimizing code for the last branch of non-capturing parentheses,
221 and also for the obeyed branch of a conditional subexpression, which used
222 tail recursion to cut down on stack usage. Unfortunately, now that there is
223 the possibility of (*THEN) occurring in these branches, tail recursion is
224 no longer possible because the return has to be checked for (*THEN). These
225 two optimizations have therefore been removed. [But see 8.20/11 above.]
226
227 18. If a pattern containing \R was studied, it was assumed that \R always
228 matched two bytes, thus causing the minimum subject length to be
229 incorrectly computed because \R can also match just one byte.
230
231 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
232 was incorrectly computed.
233
234 20. If /S is present twice on a test pattern in pcretest input, it now
235 *disables* studying, thereby overriding the use of -s on the command line
236 (see 14 above). This is necessary for one or two tests to keep the output
237 identical in both cases.
238
239 21. When (*ACCEPT) was used in an assertion that matched an empty string and
240 PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
241
242 22. When an atomic group that contained a capturing parenthesis was
243 successfully matched, but the branch in which it appeared failed, the
244 capturing was not being forgotten if a higher numbered group was later
245 captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
246 group 1 to "a", when in fact it should be unset. This applied to multi-
247 branched capturing and non-capturing groups, repeated or not, and also to
248 positive assertions (capturing in negative assertions does not happen
249 in PCRE) and also to nested atomic groups.
250
251 23. Add the ++ qualifier feature to pcretest, to show the remainder of the
252 subject after a captured substring, to make it easier to tell which of a
253 number of identical substrings has been captured.
254
255 24. The way atomic groups are processed by pcre_exec() has been changed so that
256 if they are repeated, backtracking one repetition now resets captured
257 values correctly. For example, if ((?>(a+)b)+aabab) is matched against
258 "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
259 "aaa". Previously, it would have been "a". As part of this code
260 refactoring, the way recursive calls are handled has also been changed.
261
262 25. If an assertion condition captured any substrings, they were not passed
263 back unless some other capturing happened later. For example, if
264 (?(?=(a))a) was matched against "a", no capturing was returned.
265
266 26. When studying a pattern that contained subroutine calls or assertions,
267 the code for finding the minimum length of a possible match was handling
268 direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
269 group 1 called group 2 while simultaneously a separate group 2 called group
270 1). A stack overflow occurred in this case. I have fixed this by limiting
271 the recursion depth to 10.
272
273 27. Updated RunTest.bat in the distribution to the version supplied by Tom
274 Fortmann. This supports explicit test numbers on the command line, and has
275 argument validation and error reporting.
276
277 28. An instance of \X with an unlimited repeat could fail if at any point the
278 first character it looked at was a mark character.
279
280 29. Some minor code refactoring concerning Unicode properties and scripts
281 should reduce the stack requirement of match() slightly.
282
283 30. Added the '=' option to pcretest to check the setting of unused capturing
284 slots at the end of the pattern, which are documented as being -1, but are
285 not included in the return count.
286
287 31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
288 compiled something random. Now it gives a compile-time error (as does
289 Perl).
290
291 32. A *MARK encountered during the processing of a positive assertion is now
292 recorded and passed back (compatible with Perl).
293
294 33. If --only-matching or --colour was set on a pcregrep call whose pattern
295 had alternative anchored branches, the search for a second match in a line
296 was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
297 matched the line "0102" twice. The same bug affected patterns that started
298 with a backwards assertion. For example /\b01|\b02/ also matched "0102"
299 twice.
300
301 34. Previously, PCRE did not allow quantification of assertions. However, Perl
302 does, and because of capturing effects, quantifying parenthesized
303 assertions may at times be useful. Quantifiers are now allowed for
304 parenthesized assertions.
305
306 35. A minor code tidy in pcre_compile() when checking options for \R usage.
307
308 36. \g was being checked for fancy things in a character class, when it should
309 just be a literal "g".
310
311 37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
312 appearance of a nested POSIX class supersedes an apparent external class.
313 For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
314 unescaped square brackets may also appear as part of class names. For
315 example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
316 more like Perl. (But see 8.20/1 above.)
317
318 38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
319 was because it thought it was \N{name}, which is not supported).
320
321 39. Add minix to OS list not supporting the -S option in pcretest.
322
323 40. PCRE tries to detect cases of infinite recursion at compile time, but it
324 cannot analyze patterns in sufficient detail to catch mutual recursions
325 such as ((?1))((?2)). There is now a runtime test that gives an error if a
326 subgroup is called recursively as a subpattern for a second time at the
327 same position in the subject string. In previous releases this might have
328 been caught by the recursion limit, or it might have run out of stack.
329
330 41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
331 happen only once. PCRE was, however incorrectly giving a compile time error
332 "recursive call could loop indefinitely" because it cannot analyze the
333 pattern in sufficient detail. The compile time test no longer happens when
334 PCRE is compiling a conditional subpattern, but actual runaway loops are
335 now caught at runtime (see 40 above).
336
337 42. It seems that Perl allows any characters other than a closing parenthesis
338 to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
339 has been changed to be the same.
340
341 43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
342 as not to get warnings when autogen.sh is called. Also changed
343 AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
344
345 44. To help people who use pcregrep to scan files containing exceedingly long
346 lines, the following changes have been made:
347
348 (a) The default value of the buffer size parameter has been increased from
349 8K to 20K. (The actual buffer used is three times this size.)
350
351 (b) The default can be changed by ./configure --with-pcregrep-bufsize when
352 PCRE is built.
353
354 (c) A --buffer-size=n option has been added to pcregrep, to allow the size
355 to be set at run time.
356
357 (d) Numerical values in pcregrep options can be followed by K or M, for
358 example --buffer-size=50K.
359
360 (e) If a line being scanned overflows pcregrep's buffer, an error is now
361 given and the return code is set to 2.
362
363 45. Add a pointer to the latest mark to the callout data block.
364
365 46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
366 partial match of an empty string instead of no match. This was specific to
367 the use of ".".
368
369 47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
370 complete match instead of a partial match. This bug was dependent on both
371 the PCRE_UTF8 and PCRE_DOTALL options being set.
372
373 48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
374 starting byte set, because \b was not being ignored.
375
376
377 Version 8.12 15-Jan-2011
378 ------------------------
379
380 1. Fixed some typos in the markup of the man pages, and wrote a script that
381 checks for such things as part of the documentation building process.
382
383 2. On a big-endian 64-bit system, pcregrep did not correctly process the
384 --match-limit and --recursion-limit options (added for 8.11). In
385 particular, this made one of the standard tests fail. (The integer value
386 went into the wrong half of a long int.)
387
388 3. If the --colour option was given to pcregrep with -v (invert match), it
389 did strange things, either producing crazy output, or crashing. It should,
390 of course, ignore a request for colour when reporting lines that do not
391 match.
392
393 4. Another pcregrep bug caused similar problems if --colour was specified with
394 -M (multiline) and the pattern match finished with a line ending.
395
396 5. In pcregrep, when a pattern that ended with a literal newline sequence was
397 matched in multiline mode, the following line was shown as part of the
398 match. This seems wrong, so I have changed it.
399
400 6. Another pcregrep bug in multiline mode, when --colour was specified, caused
401 the check for further matches in the same line (so they could be coloured)
402 to overrun the end of the current line. If another match was found, it was
403 incorrectly shown (and then shown again when found in the next line).
404
405 7. If pcregrep was compiled under Windows, there was a reference to the
406 function pcregrep_exit() before it was defined. I am assuming this was
407 the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
408 reported by a user. I've moved the definition above the reference.
409
410
411 Version 8.11 10-Dec-2010
412 ------------------------
413
414 1. (*THEN) was not working properly if there were untried alternatives prior
415 to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
416 backtracked to try for "b" instead of moving to the next alternative branch
417 at the same level (in this case, to look for "c"). The Perl documentation
418 is clear that when (*THEN) is backtracked onto, it goes to the "next
419 alternative in the innermost enclosing group".
420
421 2. (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
422 such as (A(*COMMIT)B(*THEN)C|D) any failure after matching A should
423 result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
424 (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
425 (*THEN).
426
427 3. If \s appeared in a character class, it removed the VT character from
428 the class, even if it had been included by some previous item, for example
429 in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
430 of \s, but is part of the POSIX "space" class.)
431
432 4. A partial match never returns an empty string (because you can always
433 match an empty string at the end of the subject); however the checking for
434 an empty string was starting at the "start of match" point. This has been
435 changed to the "earliest inspected character" point, because the returned
436 data for a partial match starts at this character. This means that, for
437 example, /(?<=abc)def/ gives a partial match for the subject "abc"
438 (previously it gave "no match").
439
440 5. Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
441 of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
442 previously a full match would be given. However, setting PCRE_PARTIAL_HARD
443 has an implication that the given string is incomplete (because a partial
444 match is preferred over a full match). For this reason, these items now
445 give a partial match in this situation. [Aside: previously, the one case
446 /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
447 match rather than a full match, which was wrong by the old rules, but is
448 now correct.]
449
450 6. There was a bug in the handling of #-introduced comments, recognized when
451 PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
452 If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
453 UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
454 scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
455 but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
456 places in pcre_compile().
457
458 7. Related to (6) above, when pcre_compile() was skipping #-introduced
459 comments when looking ahead for named forward references to subpatterns,
460 the only newline sequence it recognized was NL. It now handles newlines
461 according to the set newline convention.
462
463 8. SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
464 former, but used strtoul(), whereas pcretest avoided strtoul() but did not
465 cater for a lack of strerror(). These oversights have been fixed.
466
467 9. Added --match-limit and --recursion-limit to pcregrep.
468
469 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
470
471 11. When the -o option was used, pcregrep was setting a return code of 1, even
472 when matches were found, and --line-buffered was not being honoured.
473
474 12. Added an optional parentheses number to the -o and --only-matching options
475 of pcregrep.
476
477 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
478 can match an empty string. The code to do it in pcretest and pcredemo
479 needed fixing:
480
481 (a) When the newline convention was "crlf", pcretest got it wrong, skipping
482 only one byte after an empty string match just before CRLF (this case
483 just got forgotten; "any" and "anycrlf" were OK).
484
485 (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
486 mode when an empty string match preceded an ASCII character followed by
487 a non-ASCII character. (The code for advancing by one character rather
488 than one byte was nonsense.)
489
490 (c) The pcredemo.c sample program did not have any code at all to handle
491 the cases when CRLF is a valid newline sequence.
492
493 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
494 as a starting offset was within the subject string. There is now a new
495 error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
496 negative or greater than the length of the string. In order to test this,
497 pcretest is extended to allow the setting of negative starting offsets.
498
499 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
500 starting offset points to the beginning of a UTF-8 character was
501 unnecessarily clumsy. I tidied it up.
502
503 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
504 bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
505
506 17. Nobody had reported that the --include_dir option, which was added in
507 release 7.7 should have been called --include-dir (hyphen, not underscore)
508 for compatibility with GNU grep. I have changed it to --include-dir, but
509 left --include_dir as an undocumented synonym, and the same for
510 --exclude-dir, though that is not available in GNU grep, at least as of
511 release 2.5.4.
512
513 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
514 characters from a string of bytes) have been redefined so as not to use
515 loops, in order to improve performance in some environments. At the same
516 time, I abstracted some of the common code into auxiliary macros to save
517 repetition (this should not affect the compiled code).
518
519 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
520 compile-time error is now given if \c is not followed by an ASCII
521 character, that is, a byte less than 128. (In EBCDIC mode, the code is
522 different, and any byte value is allowed.)
523
524 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
525 START_OPTIMIZE option, which is now allowed at compile time - but just
526 passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
527 to pcregrep and other applications that have no direct access to PCRE
528 options. The new /Y option in pcretest sets this option when calling
529 pcre_compile().
530
531 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
532 back references. Groups containing recursive back references were forced to
533 be atomic by that change, but in the case of named groups, the amount of
534 memory required was incorrectly computed, leading to "Failed: internal
535 error: code overflow". This has been fixed.
536
537 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
538 pcretest.c, to avoid build problems in some Borland environments.
539
540
541 Version 8.10 25-Jun-2010
542 ------------------------
543
544 1. Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
545 THEN.
546
547 2. (*ACCEPT) was not working when inside an atomic group.
548
549 3. Inside a character class, \B is treated as a literal by default, but
550 faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
551 causes the error). The code is unchanged, but I tidied the documentation.
552
553 4. Inside a character class, PCRE always treated \R and \X as literals,
554 whereas Perl faults them if its -w option is set. I have changed PCRE so
555 that it faults them when PCRE_EXTRA is set.
556
557 5. Added support for \N, which always matches any character other than
558 newline. (It is the same as "." when PCRE_DOTALL is not set.)
559
560 6. When compiling pcregrep with newer versions of gcc which may have
561 FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
562 declared with attribute warn_unused_result" were given. Just casting the
563 result to (void) does not stop the warnings; a more elaborate fudge is
564 needed. I've used a macro to implement this.
565
566 7. Minor change to pcretest.c to avoid a compiler warning.
567
568 8. Added four artifical Unicode properties to help with an option to make
569 \s etc use properties (see next item). The new properties are: Xan
570 (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
571
572 9. Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
573 use Unicode properties. (*UCP) at the start of a pattern can be used to set
574 this option. Modified pcretest to add /W to test this facility. Added
575 REG_UCP to make it available via the POSIX interface.
576
577 10. Added --line-buffered to pcregrep.
578
579 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
580 studied, and the match started with a letter with a code point greater than
581 127 whose first byte was different to the first byte of the other case of
582 the letter, the other case of this starting letter was not recognized
583 (#976).
584
585 12. If a pattern that was studied started with a repeated Unicode property
586 test, for example, \p{Nd}+, there was the theoretical possibility of
587 setting up an incorrect bitmap of starting bytes, but fortunately it could
588 not have actually happened in practice until change 8 above was made (it
589 added property types that matched character-matching opcodes).
590
591 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
592 possible starting bytes for non-anchored patterns.
593
594 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
595 \R, and also a number of cases that involve Unicode properties, both
596 explicit and implicit when PCRE_UCP is set.
597
598 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
599 input, it could crash or give wrong results if characters with values
600 greater than 0xc0 were present in the subject string. (Detail: it assumed
601 UTF-8 input when processing these items.)
602
603 16. Added a lot of (int) casts to avoid compiler warnings in systems where
604 size_t is 64-bit (#991).
605
606 17. Added a check for running out of memory when PCRE is compiled with
607 --disable-stack-for-recursion (#990).
608
609 18. If the last data line in a file for pcretest does not have a newline on
610 the end, a newline was missing in the output.
611
612 19. The default pcre_chartables.c file recognizes only ASCII characters (values
613 less than 128) in its various bitmaps. However, there is a facility for
614 generating tables according to the current locale when PCRE is compiled. It
615 turns out that in some environments, 0x85 and 0xa0, which are Unicode space
616 characters, are recognized by isspace() and therefore were getting set in
617 these tables, and indeed these tables seem to approximate to ISO 8859. This
618 caused a problem in UTF-8 mode when pcre_study() was used to create a list
619 of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
620 which of course cannot start UTF-8 characters. I have changed the code so
621 that only real ASCII characters (less than 128) and the correct starting
622 bytes for UTF-8 encodings are set for characters greater than 127 when in
623 UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
624 altogether.)
625
626 20. Added the /T option to pcretest so as to be able to run tests with non-
627 standard character tables, thus making it possible to include the tests
628 used for 19 above in the standard set of tests.
629
630 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
631 reference to a subpattern the other side of a comment that contains an
632 opening parenthesis caused either an internal compiling error, or a
633 reference to the wrong subpattern.
634
635
636 Version 8.02 19-Mar-2010
637 ------------------------
638
639 1. The Unicode data tables have been updated to Unicode 5.2.0.
640
641 2. Added the option --libs-cpp to pcre-config, but only when C++ support is
642 configured.
643
644 3. Updated the licensing terms in the pcregexp.pas file, as agreed with the
645 original author of that file, following a query about its status.
646
647 4. On systems that do not have stdint.h (e.g. Solaris), check for and include
648 inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
649
650 5. A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
651 quantifier applied to a forward-referencing subroutine call, could compile
652 incorrect code or give the error "internal error: previously-checked
653 referenced subpattern not found".
654
655 6. Both MS Visual Studio and Symbian OS have problems with initializing
656 variables to point to external functions. For these systems, therefore,
657 pcre_malloc etc. are now initialized to local functions that call the
658 relevant global functions.
659
660 7. There were two entries missing in the vectors called coptable and poptable
661 in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
662 I've fixed the data, and added a kludgy way of testing at compile time that
663 the lengths are correct (equal to the number of opcodes).
664
665 8. Following on from 7, I added a similar kludge to check the length of the
666 eint vector in pcreposix.c.
667
668 9. Error texts for pcre_compile() are held as one long string to avoid too
669 much relocation at load time. To find a text, the string is searched,
670 counting zeros. There was no check for running off the end of the string,
671 which could happen if a new error number was added without updating the
672 string.
673
674 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
675
676 11. \K was not working if it appeared in an atomic group or in a group that
677 was called as a "subroutine", or in an assertion. Perl 5.11 documents that
678 \K is "not well defined" if used in an assertion. PCRE now accepts it if
679 the assertion is positive, but not if it is negative.
680
681 12. Change 11 fortuitously reduced the size of the stack frame used in the
682 "match()" function of pcre_exec.c by one pointer. Forthcoming
683 implementation of support for (*MARK) will need an extra pointer on the
684 stack; I have reserved it now, so that the stack frame size does not
685 decrease.
686
687 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
688 item in branch that calls a recursion is a subroutine call - as in the
689 second branch in the above example - was incorrectly given the compile-
690 time error "recursive call could loop indefinitely" because pcre_compile()
691 was not correctly checking the subroutine for matching a non-empty string.
692
693 14. The checks for overrunning compiling workspace could trigger after an
694 overrun had occurred. This is a "should never occur" error, but it can be
695 triggered by pathological patterns such as hundreds of nested parentheses.
696 The checks now trigger 100 bytes before the end of the workspace.
697
698 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
699
700
701 Version 8.01 19-Jan-2010
702 ------------------------
703
704 1. If a pattern contained a conditional subpattern with only one branch (in
705 particular, this includes all (*DEFINE) patterns), a call to pcre_study()
706 computed the wrong minimum data length (which is of course zero for such
707 subpatterns). This could cause incorrect "no match" results.
708
709 2. For patterns such as (?i)a(?-i)b|c where an option setting at the start of
710 the pattern is reset in the first branch, pcre_compile() failed with
711 "internal error: code overflow at offset...". This happened only when
712 the reset was to the original external option setting. (An optimization
713 abstracts leading options settings into an external setting, which was the
714 cause of this.)
715
716 3. A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
717 of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
718 assertion pattern did not match (meaning that the assertion was true), it
719 was incorrectly treated as false if the SKIP had been reached during the
720 matching. This also applied to assertions used as conditions.
721
722 4. If an item that is not supported by pcre_dfa_exec() was encountered in an
723 assertion subpattern, including such a pattern used as a condition,
724 unpredictable results occurred, instead of the error return
725 PCRE_ERROR_DFA_UITEM.
726
727 5. The C++ GlobalReplace function was not working like Perl for the special
728 situation when an empty string is matched. It now does the fancy magic
729 stuff that is necessary.
730
731 6. In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
732 removed. (These were left over from very, very early versions of PCRE.)
733
734 7. Some cosmetic changes to the code to make life easier when compiling it
735 as part of something else:
736
737 (a) Change DEBUG to PCRE_DEBUG.
738
739 (b) In pcre_compile(), rename the member of the "branch_chain" structure
740 called "current" as "current_branch", to prevent a collision with the
741 Linux macro when compiled as a kernel module.
742
743 (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
744 prevent a collision with the Linux macro when compiled as a kernel
745 module.
746
747 8. In pcre_compile() there are some checks for integer overflows that used to
748 cast potentially large values to (double). This has been changed to that
749 when building, a check for int64_t is made, and if it is found, it is used
750 instead, thus avoiding the use of floating point arithmetic. (There is no
751 other use of FP in PCRE.) If int64_t is not found, the fallback is to
752 double.
753
754 9. Added two casts to avoid signed/unsigned warnings from VS Studio Express
755 2005 (difference between two addresses compared to an unsigned value).
756
757 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
758 custom one, because of the following reported problem in Windows:
759
760 - libbz2 uses the Pascal calling convention (WINAPI) for the functions
761 under Win32.
762 - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
763 therefore missing the function definition.
764 - The compiler thus generates a "C" signature for the test function.
765 - The linker fails to find the "C" function.
766 - PCRE fails to configure if asked to do so against libbz2.
767
768 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
769 messages were output:
770
771 Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
772 rerunning libtoolize, to keep the correct libtool macros in-tree.
773 Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
774
775 I have done both of these things.
776
777 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
778 most of the time, it *can* run out if it is given a pattern that contains a
779 runaway infinite recursion. I updated the discussion in the pcrestack man
780 page.
781
782 13. Now that we have gone to the x.xx style of version numbers, the minor
783 version may start with zero. Using 08 or 09 is a bad idea because users
784 might check the value of PCRE_MINOR in their code, and 08 or 09 may be
785 interpreted as invalid octal numbers. I've updated the previous comment in
786 configure.ac, and also added a check that gives an error if 08 or 09 are
787 used.
788
789 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
790 causing partial matching to fail when the end of the subject matched \W
791 in a UTF-8 pattern where \W was quantified with a minimum of 3.
792
793 15. There were some discrepancies between the declarations in pcre_internal.h
794 of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
795 their definitions. The declarations used "const uschar *" and the
796 definitions used USPTR. Even though USPTR is normally defined as "const
797 unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
798 reported that: "This difference in casting confuses some C++ compilers, for
799 example, SunCC recognizes above declarations as different functions and
800 generates broken code for hbpcre." I have changed the declarations to use
801 USPTR.
802
803 16. GNU libtool is named differently on some systems. The autogen.sh script now
804 tries several variants such as glibtoolize (MacOSX) and libtoolize1x
805 (FreeBSD).
806
807 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
808 (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
809 comment: "Figure out how to create a longlong from a string: strtoll and
810 equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
811 instance, but it only takes 2 args instead of 3!"
812
813 18. A subtle bug concerned with back references has been fixed by a change of
814 specification, with a corresponding code fix. A pattern such as
815 ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
816 refers, was giving matches when it shouldn't. For example, xa=xaaa would
817 match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
818 same bug. Such groups have to be quantified to be useful, or contained
819 inside another quantified group. (If there's no repetition, the reference
820 can never match.) The problem arises because, having left the group and
821 moved on to the rest of the pattern, a later failure that backtracks into
822 the group uses the captured value from the final iteration of the group
823 rather than the correct earlier one. I have fixed this in PCRE by forcing
824 any group that contains a reference to itself to be an atomic group; that
825 is, there cannot be any backtracking into it once it has completed. This is
826 similar to recursive and subroutine calls.
827
828
829 Version 8.00 19-Oct-09
830 ----------------------
831
832 1. The table for translating pcre_compile() error codes into POSIX error codes
833 was out-of-date, and there was no check on the pcre_compile() error code
834 being within the table. This could lead to an OK return being given in
835 error.
836
837 2. Changed the call to open a subject file in pcregrep from fopen(pathname,
838 "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
839 in a Windows environment.
840
841 3. The pcregrep --count option prints the count for each file even when it is
842 zero, as does GNU grep. However, pcregrep was also printing all files when
843 --files-with-matches was added. Now, when both options are given, it prints
844 counts only for those files that have at least one match. (GNU grep just
845 prints the file name in this circumstance, but including the count seems
846 more useful - otherwise, why use --count?) Also ensured that the
847 combination -clh just lists non-zero counts, with no names.
848
849 4. The long form of the pcregrep -F option was incorrectly implemented as
850 --fixed_strings instead of --fixed-strings. This is an incompatible change,
851 but it seems right to fix it, and I didn't think it was worth preserving
852 the old behaviour.
853
854 5. The command line items --regex=pattern and --regexp=pattern were not
855 recognized by pcregrep, which required --regex pattern or --regexp pattern
856 (with a space rather than an '='). The man page documented the '=' forms,
857 which are compatible with GNU grep; these now work.
858
859 6. No libpcreposix.pc file was created for pkg-config; there was just
860 libpcre.pc and libpcrecpp.pc. The omission has been rectified.
861
862 7. Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
863 when UCP support is not needed, by modifying the Python script that
864 generates it from Unicode data files. This should not matter if the module
865 is correctly used as a library, but I received one complaint about 50K of
866 unwanted data. My guess is that the person linked everything into his
867 program rather than using a library. Anyway, it does no harm.
868
869 8. A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
870 was a minimum greater than 1 for a wide character in a possessive
871 repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
872 which had an unlimited repeat of a nested, fixed maximum repeat of a wide
873 character. Chaos in the form of incorrect output or a compiling loop could
874 result.
875
876 9. The restrictions on what a pattern can contain when partial matching is
877 requested for pcre_exec() have been removed. All patterns can now be
878 partially matched by this function. In addition, if there are at least two
879 slots in the offset vector, the offset of the earliest inspected character
880 for the match and the offset of the end of the subject are set in them when
881 PCRE_ERROR_PARTIAL is returned.
882
883 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
884 synonymous with PCRE_PARTIAL, for backwards compatibility, and
885 PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
886 and may be more useful for multi-segment matching.
887
888 11. Partial matching with pcre_exec() is now more intuitive. A partial match
889 used to be given if ever the end of the subject was reached; now it is
890 given only if matching could not proceed because another character was
891 needed. This makes a difference in some odd cases such as Z(*FAIL) with the
892 string "Z", which now yields "no match" instead of "partial match". In the
893 case of pcre_dfa_exec(), "no match" is given if every matching path for the
894 final character ended with (*FAIL).
895
896 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
897 if the pattern had a "must contain" character that was already found in the
898 earlier partial match, unless partial matching was again requested. For
899 example, with the pattern /dog.(body)?/, the "must contain" character is
900 "g". If the first part-match was for the string "dog", restarting with
901 "sbody" failed. This bug has been fixed.
902
903 13. The string returned by pcre_dfa_exec() after a partial match has been
904 changed so that it starts at the first inspected character rather than the
905 first character of the match. This makes a difference only if the pattern
906 starts with a lookbehind assertion or \b or \B (\K is not supported by
907 pcre_dfa_exec()). It's an incompatible change, but it makes the two
908 matching functions compatible, and I think it's the right thing to do.
909
910 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
911 so that the demonstration program is easily available in environments where
912 PCRE has not been installed from source.
913
914 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
915 libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
916 library.
917
918 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
919 It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
920 is not the first non-POSIX option to be added. Clearly some people find
921 these options useful.
922
923 17. If a caller to the POSIX matching function regexec() passes a non-zero
924 value for nmatch with a NULL value for pmatch, the value of
925 nmatch is forced to zero.
926
927 18. RunGrepTest did not have a test for the availability of the -u option of
928 the diff command, as RunTest does. It now checks in the same way as
929 RunTest, and also checks for the -b option.
930
931 19. If an odd number of negated classes containing just a single character
932 interposed, within parentheses, between a forward reference to a named
933 subpattern and the definition of the subpattern, compilation crashed with
934 an internal error, complaining that it could not find the referenced
935 subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
936 [The bug was that it was starting one character too far in when skipping
937 over the character class, thus treating the ] as data rather than
938 terminating the class. This meant it could skip too much.]
939
940 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
941 /g option in pcretest when the pattern contains \K, which makes it possible
942 to have an empty string match not at the start, even when the pattern is
943 anchored. Updated pcretest and pcredemo to use this option.
944
945 21. If the maximum number of capturing subpatterns in a recursion was greater
946 than the maximum at the outer level, the higher number was returned, but
947 with unset values at the outer level. The correct (outer level) value is
948 now given.
949
950 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
951 PCRE did not set those parentheses (unlike Perl). I have now found a way to
952 make it do so. The string so far is captured, making this feature
953 compatible with Perl.
954
955 23. The tests have been re-organized, adding tests 11 and 12, to make it
956 possible to check the Perl 5.10 features against Perl 5.10.
957
958 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
959 pattern matches a fixed length string. PCRE did not allow this; now it
960 does. Neither allows recursion.
961
962 25. I finally figured out how to implement a request to provide the minimum
963 length of subject string that was needed in order to match a given pattern.
964 (It was back references and recursion that I had previously got hung up
965 on.) This code has now been added to pcre_study(); it finds a lower bound
966 to the length of subject needed. It is not necessarily the greatest lower
967 bound, but using it to avoid searching strings that are too short does give
968 some useful speed-ups. The value is available to calling programs via
969 pcre_fullinfo().
970
971 26. While implementing 25, I discovered to my embarrassment that pcretest had
972 not been passing the result of pcre_study() to pcre_dfa_exec(), so the
973 study optimizations had never been tested with that matching function.
974 Oops. What is worse, even when it was passed study data, there was a bug in
975 pcre_dfa_exec() that meant it never actually used it. Double oops. There
976 were also very few tests of studied patterns with pcre_dfa_exec().
977
978 27. If (?| is used to create subpatterns with duplicate numbers, they are now
979 allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
980 on the other side of the coin, they are no longer allowed to have different
981 names, because these cannot be distinguished in PCRE, and this has caused
982 confusion. (This is a difference from Perl.)
983
984 28. When duplicate subpattern names are present (necessarily with different
985 numbers, as required by 27 above), and a test is made by name in a
986 conditional pattern, either for a subpattern having been matched, or for
987 recursion in such a pattern, all the associated numbered subpatterns are
988 tested, and the overall condition is true if the condition is true for any
989 one of them. This is the way Perl works, and is also more like the way
990 testing by number works.
991
992
993 Version 7.9 11-Apr-09
994 ---------------------
995
996 1. When building with support for bzlib/zlib (pcregrep) and/or readline
997 (pcretest), all targets were linked against these libraries. This included
998 libpcre, libpcreposix, and libpcrecpp, even though they do not use these
999 libraries. This caused unwanted dependencies to be created. This problem
1000 has been fixed, and now only pcregrep is linked with bzlib/zlib and only
1001 pcretest is linked with readline.
1002
1003 2. The "typedef int BOOL" in pcre_internal.h that was included inside the
1004 "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
1005 moved outside it again, because FALSE and TRUE are already defined in AIX,
1006 but BOOL is not.
1007
1008 3. The pcre_config() function was treating the PCRE_MATCH_LIMIT and
1009 PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
1010
1011 4. The pcregrep documentation said spaces were inserted as well as colons (or
1012 hyphens) following file names and line numbers when outputting matching
1013 lines. This is not true; no spaces are inserted. I have also clarified the
1014 wording for the --colour (or --color) option.
1015
1016 5. In pcregrep, when --colour was used with -o, the list of matching strings
1017 was not coloured; this is different to GNU grep, so I have changed it to be
1018 the same.
1019
1020 6. When --colo(u)r was used in pcregrep, only the first matching substring in
1021 each matching line was coloured. Now it goes on to look for further matches
1022 of any of the test patterns, which is the same behaviour as GNU grep.
1023
1024 7. A pattern that could match an empty string could cause pcregrep to loop; it
1025 doesn't make sense to accept an empty string match in pcregrep, so I have
1026 locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
1027 seems to be how GNU grep behaves.
1028
1029 8. The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
1030 start or after a newline", because the conditional assertion was not being
1031 correctly handled. The rule now is that both the assertion and what follows
1032 in the first alternative must satisfy the test.
1033
1034 9. If auto-callout was enabled in a pattern with a conditional group whose
1035 condition was an assertion, PCRE could crash during matching, both with
1036 pcre_exec() and pcre_dfa_exec().
1037
1038 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
1039 used for matching.
1040
1041 11. Unicode property support in character classes was not working for
1042 characters (bytes) greater than 127 when not in UTF-8 mode.
1043
1044 12. Added the -M command line option to pcretest.
1045
1046 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
1047
1048 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
1049
1050 16. Added comments and documentation about mis-use of no_arg in the C++
1051 wrapper.
1052
1053 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
1054 from Martin Jerabek that uses macro names for all relevant character and
1055 string constants.
1056
1057 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
1058 SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
1059 SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
1060 these, but not everybody uses configure.
1061
1062 19. A conditional group that had only one branch was not being correctly
1063 recognized as an item that could match an empty string. This meant that an
1064 enclosing group might also not be so recognized, causing infinite looping
1065 (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
1066 with the subject "ab", where knowledge that the repeated group can match
1067 nothing is needed in order to break the loop.
1068
1069 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
1070 exec(), but without supplying a callout function, matching went wrong.
1071
1072 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
1073 leak if the size of the offset vector was greater than 30. When the vector
1074 is smaller, the saved offsets during recursion go onto a local stack
1075 vector, but for larger vectors malloc() is used. It was failing to free
1076 when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
1077 error, in fact).
1078
1079 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
1080 heapframe that is used only when UTF-8 support is enabled. This caused no
1081 problem, but was untidy.
1082
1083 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
1084 CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
1085 included within another project.
1086
1087 24. Steven Van Ingelgem's patches to add more options to the CMake support,
1088 slightly modified by me:
1089
1090 (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
1091 not building pcregrep.
1092
1093 (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
1094 if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
1095
1096 25. Forward references, both numeric and by name, in patterns that made use of
1097 duplicate group numbers, could behave incorrectly or give incorrect errors,
1098 because when scanning forward to find the reference group, PCRE was not
1099 taking into account the duplicate group numbers. A pattern such as
1100 ^X(?3)(a)(?|(b)|(q))(Y) is an example.
1101
1102 26. Changed a few more instances of "const unsigned char *" to USPTR, making
1103 the feature of a custom pointer more persuasive (as requested by a user).
1104
1105 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
1106 pcretest.c, inside #ifndefs, because it seems they are sometimes already
1107 pre-defined.
1108
1109 28. Added support for (*UTF8) at the start of a pattern.
1110
1111 29. Arrange for flags added by the "release type" setting in CMake to be shown
1112 in the configuration summary.
1113
1114
1115 Version 7.8 05-Sep-08
1116 ---------------------
1117
1118 1. Replaced UCP searching code with optimized version as implemented for Ad
1119 Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
1120 stage table and inline lookup instead of a function, giving speed ups of 2
1121 to 5 times on some simple patterns that I tested. Permission was given to
1122 distribute the MultiStage2.py script that generates the tables (it's not in
1123 the tarball, but is in the Subversion repository).
1124
1125 2. Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
1126 scripts.
1127
1128 3. Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
1129 a group with a zero qualifier. The result of the study could be incorrect,
1130 or the function might crash, depending on the pattern.
1131
1132 4. Caseless matching was not working for non-ASCII characters in back
1133 references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
1134 It now works when Unicode Property Support is available.
1135
1136 5. In pcretest, an escape such as \x{de} in the data was always generating
1137 a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
1138 non-UTF-8 mode. If the value is greater than 255, it gives a warning about
1139 truncation.
1140
1141 6. Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
1142
1143 7. Added two (int) casts to pcregrep when printing the difference of two
1144 pointers, in case they are 64-bit values.
1145
1146 8. Added comments about Mac OS X stack usage to the pcrestack man page and to
1147 test 2 if it fails.
1148
1149 9. Added PCRE_CALL_CONVENTION just before the names of all exported functions,
1150 and a #define of that name to empty if it is not externally set. This is to
1151 allow users of MSVC to set it if necessary.
1152
1153 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
1154 the convenience functions in the pcre_get.c source file.
1155
1156 11. An option change at the start of a pattern that had top-level alternatives
1157 could cause overwriting and/or a crash. This command provoked a crash in
1158 some environments:
1159
1160 printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
1161
1162 This potential security problem was recorded as CVE-2008-2371.
1163
1164 12. For a pattern where the match had to start at the beginning or immediately
1165 after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
1166 pcre_dfa_exec() could read past the end of the passed subject if there was
1167 no match. To help with detecting such bugs (e.g. with valgrind), I modified
1168 pcretest so that it places the subject at the end of its malloc-ed buffer.
1169
1170 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
1171 exec() might read past the end of the data buffer in UTF-8 mode.
1172
1173 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
1174 the data contained the byte 0x85 as part of a UTF-8 character within its
1175 first line. This applied both to normal and DFA matching.
1176
1177 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
1178 /^[^d]*?$/8 failed to match "abc".
1179
1180 16. Added a missing copyright notice to pcrecpp_internal.h.
1181
1182 17. Make it more clear in the documentation that values returned from
1183 pcre_exec() in ovector are byte offsets, not character counts.
1184
1185 18. Tidied a few places to stop certain compilers from issuing warnings.
1186
1187 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
1188 supplied by Stefan Weber. I made a further small update for 7.8 because
1189 there is a change of source arrangements: the pcre_searchfuncs.c module is
1190 replaced by pcre_ucd.c.
1191
1192
1193 Version 7.7 07-May-08
1194 ---------------------
1195
1196 1. Applied Craig's patch to sort out a long long problem: "If we can't convert
1197 a string to a long long, pretend we don't even have a long long." This is
1198 done by checking for the strtoq, strtoll, and _strtoi64 functions.
1199
1200 2. Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
1201 pre-7.6 versions, which defined a global no_arg variable instead of putting
1202 it in the RE class. (See also #8 below.)
1203
1204 3. Remove a line of dead code, identified by coverity and reported by Nuno
1205 Lopes.
1206
1207 4. Fixed two related pcregrep bugs involving -r with --include or --exclude:
1208
1209 (1) The include/exclude patterns were being applied to the whole pathnames
1210 of files, instead of just to the final components.
1211
1212 (2) If there was more than one level of directory, the subdirectories were
1213 skipped unless they satisfied the include/exclude conditions. This is
1214 inconsistent with GNU grep (and could even be seen as contrary to the
1215 pcregrep specification - which I improved to make it absolutely clear).
1216 The action now is always to scan all levels of directory, and just
1217 apply the include/exclude patterns to regular files.
1218
1219 5. Added the --include_dir and --exclude_dir patterns to pcregrep, and used
1220 --exclude_dir in the tests to avoid scanning .svn directories.
1221
1222 6. Applied Craig's patch to the QuoteMeta function so that it escapes the
1223 NUL character as backslash + 0 rather than backslash + NUL, because PCRE
1224 doesn't support NULs in patterns.
1225
1226 7. Added some missing "const"s to declarations of static tables in
1227 pcre_compile.c and pcre_dfa_exec.c.
1228
1229 8. Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
1230 caused by fix #2 above. (Subsequently also a second patch to fix the
1231 first patch. And a third patch - this was a messy problem.)
1232
1233 9. Applied Craig's patch to remove the use of push_back().
1234
1235 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
1236 matching function regexec().
1237
1238 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
1239 which, however, unlike Perl's \g{...}, are subroutine calls, not back
1240 references. PCRE supports relative numbers with this syntax (I don't think
1241 Oniguruma does).
1242
1243 12. Previously, a group with a zero repeat such as (...){0} was completely
1244 omitted from the compiled regex. However, this means that if the group
1245 was called as a subroutine from elsewhere in the pattern, things went wrong
1246 (an internal error was given). Such groups are now left in the compiled
1247 pattern, with a new opcode that causes them to be skipped at execution
1248 time.
1249
1250 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
1251 to the way PCRE behaves:
1252
1253 (a) A lone ] character is dis-allowed (Perl treats it as data).
1254
1255 (b) A back reference to an unmatched subpattern matches an empty string
1256 (Perl fails the current match path).
1257
1258 (c) A data ] in a character class must be notated as \] because if the
1259 first data character in a class is ], it defines an empty class. (In
1260 Perl it is not possible to have an empty class.) The empty class []
1261 never matches; it forces failure and is equivalent to (*FAIL) or (?!).
1262 The negative empty class [^] matches any one character, independently
1263 of the DOTALL setting.
1264
1265 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
1266 non-existent subpattern following a character class starting with ']' and
1267 containing () gave an internal compiling error instead of "reference to
1268 non-existent subpattern". Fortunately, when the pattern did exist, the
1269 compiled code was correct. (When scanning forwards to check for the
1270 existencd of the subpattern, it was treating the data ']' as terminating
1271 the class, so got the count wrong. When actually compiling, the reference
1272 was subsequently set up correctly.)
1273
1274 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
1275 it was being rejected as not supported by pcre_dfa_exec(), even though
1276 other assertions are supported. I have made pcre_dfa_exec() support
1277 (*FAIL).
1278
1279 16. The implementation of 13c above involved the invention of a new opcode,
1280 OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
1281 cannot be changed at match time, I realized I could make a small
1282 improvement to matching performance by compiling OP_ALLANY instead of
1283 OP_ANY for "." when DOTALL was set, and then removing the runtime tests
1284 on the OP_ANY path.
1285
1286 17. Compiling pcretest on Windows with readline support failed without the
1287 following two fixes: (1) Make the unistd.h include conditional on
1288 HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
1289
1290 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
1291 ncurses library to be included for pcretest when ReadLine support is
1292 requested, but also to allow for it to be overridden. This patch came from
1293 Daniel Bergström.
1294
1295 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
1296 as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
1297 any errors with the current Unicode tables. Thanks to Peter Kankowski for
1298 spotting this.
1299
1300
1301 Version 7.6 28-Jan-08
1302 ---------------------
1303
1304 1. A character class containing a very large number of characters with
1305 codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
1306 overflow.
1307
1308 2. Patch to cut out the "long long" test in pcrecpp_unittest when
1309 HAVE_LONG_LONG is not defined.
1310
1311 3. Applied Christian Ehrlicher's patch to update the CMake build files to
1312 bring them up to date and include new features. This patch includes:
1313
1314 - Fixed PH's badly added libz and libbz2 support.
1315 - Fixed a problem with static linking.
1316 - Added pcredemo. [But later removed - see 7 below.]
1317 - Fixed dftables problem and added an option.
1318 - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
1319 HAVE_LONG_LONG.
1320 - Added readline support for pcretest.
1321 - Added an listing of the option settings after cmake has run.
1322
1323 4. A user submitted a patch to Makefile that makes it easy to create
1324 "pcre.dll" under mingw when using Configure/Make. I added stuff to
1325 Makefile.am that cause it to include this special target, without
1326 affecting anything else. Note that the same mingw target plus all
1327 the other distribution libraries and programs are now supported
1328 when configuring with CMake (see 6 below) instead of with
1329 Configure/Make.
1330
1331 5. Applied Craig's patch that moves no_arg into the RE class in the C++ code.
1332 This is an attempt to solve the reported problem "pcrecpp::no_arg is not
1333 exported in the Windows port". It has not yet been confirmed that the patch
1334 solves the problem, but it does no harm.
1335
1336 6. Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
1337 NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
1338 with CMake, and also correct the comment about stack recursion.
1339
1340 7. Remove the automatic building of pcredemo from the ./configure system and
1341 from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
1342 of a program that users should build themselves after PCRE is installed, so
1343 building it automatically is not really right. What is more, it gave
1344 trouble in some build environments.
1345
1346 8. Further tidies to CMakeLists.txt from Sheri and Christian.
1347
1348
1349 Version 7.5 10-Jan-08
1350 ---------------------
1351
1352 1. Applied a patch from Craig: "This patch makes it possible to 'ignore'
1353 values in parens when parsing an RE using the C++ wrapper."
1354
1355 2. Negative specials like \S did not work in character classes in UTF-8 mode.
1356 Characters greater than 255 were excluded from the class instead of being
1357 included.
1358
1359 3. The same bug as (2) above applied to negated POSIX classes such as
1360 [:^space:].
1361
1362 4. PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
1363 defined or documented. It seems to have been a typo for PCRE_STATIC, so
1364 I have changed it.
1365
1366 5. The construct (?&) was not diagnosed as a syntax error (it referenced the
1367 first named subpattern) and a construct such as (?&a) would reference the
1368 first named subpattern whose name started with "a" (in other words, the
1369 length check was missing). Both these problems are fixed. "Subpattern name
1370 expected" is now given for (?&) (a zero-length name), and this patch also
1371 makes it give the same error for \k'' (previously it complained that that
1372 was a reference to a non-existent subpattern).
1373
1374 6. The erroneous patterns (?+-a) and (?-+a) give different error messages;
1375 this is right because (?- can be followed by option settings as well as by
1376 digits. I have, however, made the messages clearer.
1377
1378 7. Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
1379 than the number used in the conditional) now cause a compile-time error.
1380 This is actually not compatible with Perl, which accepts such patterns, but
1381 treats the conditional as always being FALSE (as PCRE used to), but it
1382 seems to me that giving a diagnostic is better.
1383
1384 8. Change "alphameric" to the more common word "alphanumeric" in comments
1385 and messages.
1386
1387 9. Fix two occurrences of "backslash" in comments that should have been
1388 "backspace".
1389
1390 10. Remove two redundant lines of code that can never be obeyed (their function
1391 was moved elsewhere).
1392
1393 11. The program that makes PCRE's Unicode character property table had a bug
1394 which caused it to generate incorrect table entries for sequences of
1395 characters that have the same character type, but are in different scripts.
1396 It amalgamated them into a single range, with the script of the first of
1397 them. In other words, some characters were in the wrong script. There were
1398 thirteen such cases, affecting characters in the following ranges:
1399
1400 U+002b0 - U+002c1
1401 U+0060c - U+0060d
1402 U+0061e - U+00612
1403 U+0064b - U+0065e
1404 U+0074d - U+0076d
1405 U+01800 - U+01805
1406 U+01d00 - U+01d77
1407 U+01d9b - U+01dbf
1408 U+0200b - U+0200f
1409 U+030fc - U+030fe
1410 U+03260 - U+0327f
1411 U+0fb46 - U+0fbb1
1412 U+10450 - U+1049d
1413
1414 12. The -o option (show only the matching part of a line) for pcregrep was not
1415 compatible with GNU grep in that, if there was more than one match in a
1416 line, it showed only the first of them. It now behaves in the same way as
1417 GNU grep.
1418
1419 13. If the -o and -v options were combined for pcregrep, it printed a blank
1420 line for every non-matching line. GNU grep prints nothing, and pcregrep now
1421 does the same. The return code can be used to tell if there were any
1422 non-matching lines.
1423
1424 14. Added --file-offsets and --line-offsets to pcregrep.
1425
1426 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
1427 infinitely looping recursion. The bug was that positive lookaheads were not
1428 being skipped when checking for a possible empty match (negative lookaheads
1429 and both kinds of lookbehind were skipped).
1430
1431 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
1432 inclusion of <windows.h> to before rather than after the definition of
1433 INVALID_FILE_ATTRIBUTES (patch from David Byron).
1434
1435 17. Specifying a possessive quantifier with a specific limit for a Unicode
1436 character property caused pcre_compile() to compile bad code, which led at
1437 runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
1438 are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
1439 caused the error; without that there was no problem.
1440
1441 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
1442
1443 19. Added --enable-pcretest-libreadline.
1444
1445 20. In pcrecpp.cc, the variable 'count' was incremented twice in
1446 RE::GlobalReplace(). As a result, the number of replacements returned was
1447 double what it should be. I removed one of the increments, but Craig sent a
1448 later patch that removed the other one (the right fix) and added unit tests
1449 that check the return values (which was not done before).
1450
1451 21. Several CMake things:
1452
1453 (1) Arranged that, when cmake is used on Unix, the libraries end up with
1454 the names libpcre and libpcreposix, not just pcre and pcreposix.
1455
1456 (2) The above change means that pcretest and pcregrep are now correctly
1457 linked with the newly-built libraries, not previously installed ones.
1458
1459 (3) Added PCRE_SUPPORT_LIBREADLINE, PCRE_SUPPORT_LIBZ, PCRE_SUPPORT_LIBBZ2.
1460
1461 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
1462 crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
1463 UTF-8 newline character). The key issue is that the pattern starts .*;
1464 this means that the match must be either at the beginning, or after a
1465 newline. The bug was in the code for advancing after a failed match and
1466 checking that the new position followed a newline. It was not taking
1467 account of UTF-8 characters correctly.
1468
1469 23. PCRE was behaving differently from Perl in the way it recognized POSIX
1470 character classes. PCRE was not treating the sequence [:...:] as a
1471 character class unless the ... were all letters. Perl, however, seems to
1472 allow any characters between [: and :], though of course it rejects as
1473 unknown any "names" that contain non-letters, because all the known class
1474 names consist only of letters. Thus, Perl gives an error for [[:1234:]],
1475 for example, whereas PCRE did not - it did not recognize a POSIX character
1476 class. This seemed a bit dangerous, so the code has been changed to be
1477 closer to Perl. The behaviour is not identical to Perl, because PCRE will
1478 diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
1479 treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
1480 Perl does, and where it didn't before.
1481
1482 24. Rewrite so as to remove the single use of %n from pcregrep because in some
1483 Windows environments %n is disabled by default.
1484
1485
1486 Version 7.4 21-Sep-07
1487 ---------------------
1488
1489 1. Change 7.3/28 was implemented for classes by looking at the bitmap. This
1490 means that a class such as [\s] counted as "explicit reference to CR or
1491 LF". That isn't really right - the whole point of the change was to try to
1492 help when there was an actual mention of one of the two characters. So now
1493 the change happens only if \r or \n (or a literal CR or LF) character is
1494 encountered.
1495
1496 2. The 32-bit options word was also used for 6 internal flags, but the numbers
1497 of both had grown to the point where there were only 3 bits left.
1498 Fortunately, there was spare space in the data structure, and so I have
1499 moved the internal flags into a new 16-bit field to free up more option
1500 bits.
1501
1502 3. The appearance of (?J) at the start of a pattern set the DUPNAMES option,
1503 but did not set the internal JCHANGED flag - either of these is enough to
1504 control the way the "get" function works - but the PCRE_INFO_JCHANGED
1505 facility is supposed to tell if (?J) was ever used, so now (?J) at the
1506 start sets both bits.
1507
1508 4. Added options (at build time, compile time, exec time) to change \R from
1509 matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
1510
1511 5. doc/pcresyntax.html was missing from the distribution.
1512
1513 6. Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
1514 compatibility, even though it is no longer used.
1515
1516 7. Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
1517 strtoull to pcrecpp.cc to select the available functions in WIN32 when the
1518 windows.h file is present (where different names are used). [This was
1519 reversed later after testing - see 16 below.]
1520
1521 8. Changed all #include <config.h> to #include "config.h". There were also
1522 some further <pcre.h> cases that I changed to "pcre.h".
1523
1524 9. When pcregrep was used with the --colour option, it missed the line ending
1525 sequence off the lines that it output.
1526
1527 10. It was pointed out to me that arrays of string pointers cause lots of
1528 relocations when a shared library is dynamically loaded. A technique of
1529 using a single long string with a table of offsets can drastically reduce
1530 these. I have refactored PCRE in four places to do this. The result is
1531 dramatic:
1532
1533 Originally: 290
1534 After changing UCP table: 187
1535 After changing error message table: 43
1536 After changing table of "verbs" 36
1537 After changing table of Posix names 22
1538
1539 Thanks to the folks working on Gregex for glib for this insight.
1540
1541 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
1542 unicode-properties was also set.
1543
1544 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
1545
1546 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
1547 checked only for CRLF.
1548
1549 14. Added casts to pcretest.c to avoid compiler warnings.
1550
1551 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
1552
1553 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
1554 and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
1555 entirely. This removes changes made in 7 above.
1556
1557 17. The CMake files have been updated, and there is now more information about
1558 building with CMake in the NON-UNIX-USE document.
1559
1560
1561 Version 7.3 28-Aug-07
1562 ---------------------
1563
1564 1. In the rejigging of the build system that eventually resulted in 7.1, the
1565 line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
1566 brackets there is not right, since it causes compilers to look for an
1567 installed pcre.h, not the version that is in the source that is being
1568 compiled (which of course may be different). I have changed it back to:
1569
1570 #include "pcre.h"
1571
1572 I have a vague recollection that the change was concerned with compiling in
1573 different directories, but in the new build system, that is taken care of
1574 by the VPATH setting the Makefile.
1575
1576 2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
1577 when the subject happened to end in the byte 0x85 (e.g. if the last
1578 character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
1579 characters but of course it shouldn't be taken as a newline when it is part
1580 of another character. The bug was that, for an unlimited repeat of . in
1581 not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
1582 characters when looking for a newline.
1583
1584 3. A small performance improvement in the DOTALL UTF-8 mode .* case.
1585
1586 4. Debugging: adjusted the names of opcodes for different kinds of parentheses
1587 in debug output.
1588
1589 5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
1590 long printing in the pcrecpp unittest when running under MinGW.
1591
1592 6. ESC_K was left out of the EBCDIC table.
1593
1594 7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
1595 parentheses; I made it 1000, which seemed large enough. Unfortunately, the
1596 limit also applies to "virtual nesting" when a pattern is recursive, and in
1597 this case 1000 isn't so big. I have been able to remove this limit at the
1598 expense of backing off one optimization in certain circumstances. Normally,
1599 when pcre_exec() would call its internal match() function recursively and
1600 immediately return the result unconditionally, it uses a "tail recursion"
1601 feature to save stack. However, when a subpattern that can match an empty
1602 string has an unlimited repetition quantifier, it no longer makes this
1603 optimization. That gives it a stack frame in which to save the data for
1604 checking that an empty string has been matched. Previously this was taken
1605 from the 1000-entry workspace that had been reserved. So now there is no
1606 explicit limit, but more stack is used.
1607
1608 8. Applied Daniel's patches to solve problems with the import/export magic
1609 syntax that is required for Windows, and which was going wrong for the
1610 pcreposix and pcrecpp parts of the library. These were overlooked when this
1611 problem was solved for the main library.
1612
1613 9. There were some crude static tests to avoid integer overflow when computing
1614 the size of patterns that contain repeated groups with explicit upper
1615 limits. As the maximum quantifier is 65535, the maximum group length was
1616 set at 30,000 so that the product of these two numbers did not overflow a
1617 32-bit integer. However, it turns out that people want to use groups that
1618 are longer than 30,000 bytes (though not repeat them that many times).
1619 Change 7.0/17 (the refactoring of the way the pattern size is computed) has
1620 made it possible to implement the integer overflow checks in a much more
1621 dynamic way, which I have now done. The artificial limitation on group
1622 length has been removed - we now have only the limit on the total length of
1623 the compiled pattern, which depends on the LINK_SIZE setting.
1624
1625 10. Fixed a bug in the documentation for get/copy named substring when
1626 duplicate names are permitted. If none of the named substrings are set, the
1627 functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
1628 empty string.
1629
1630 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
1631 instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
1632 because the ] is interpreted as the first data character and the
1633 terminating ] is not found. PCRE has been made compatible with Perl in this
1634 regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
1635 cause memory overwriting.
1636
1637 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
1638 string has been matched (to stop an infinite loop). It was not recognizing
1639 a conditional subpattern that could match an empty string if that
1640 subpattern was within another subpattern. For example, it looped when
1641 trying to match (((?(1)X|))*) but it was OK with ((?(1)X|)*) where the
1642 condition was not nested. This bug has been fixed.
1643
1644 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
1645 past the start of the subject in the presence of bytes with the top bit
1646 set, for example "\x8aBCD".
1647
1648 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
1649 (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
1650
1651 14. Optimized (?!) to (*FAIL).
1652
1653 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
1654 This restricts code points to be within the range 0 to 0x10FFFF, excluding
1655 the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
1656 full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
1657 does: it's just the validity check that is more restrictive.
1658
1659 16. Inserted checks for integer overflows during escape sequence (backslash)
1660 processing, and also fixed erroneous offset values for syntax errors during
1661 backslash processing.
1662
1663 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
1664 for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
1665
1666 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
1667 caused an overrun.
1668
1669 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
1670 something other than just ASCII characters) inside a group that had an
1671 unlimited repeat caused a loop at compile time (while checking to see
1672 whether the group could match an empty string).
1673
1674 20. Debugging a pattern containing \p or \P could cause a crash. For example,
1675 [\P{Any}] did so. (Error in the code for printing property names.)
1676
1677 21. An orphan \E inside a character class could cause a crash.
1678
1679 22. A repeated capturing bracket such as (A)? could cause a wild memory
1680 reference during compilation.
1681
1682 23. There are several functions in pcre_compile() that scan along a compiled
1683 expression for various reasons (e.g. to see if it's fixed length for look
1684 behind). There were bugs in these functions when a repeated \p or \P was
1685 present in the pattern. These operators have additional parameters compared
1686 with \d, etc, and these were not being taken into account when moving along
1687 the compiled data. Specifically:
1688
1689 (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
1690 length.
1691
1692 (b) An item such as \pL+ within a repeated group could cause crashes or
1693 loops.
1694
1695 (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
1696 "reference to non-existent subpattern" error.
1697
1698 (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
1699
1700 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
1701 characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
1702
1703 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
1704
1705 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
1706 character were causing crashes (broken optimization).
1707
1708 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
1709 \p or \P) caused a compile-time loop.
1710
1711 28. More problems have arisen in unanchored patterns when CRLF is a valid line
1712 break. For example, the unstudied pattern [\r\n]A does not match the string
1713 "\r\nA" because change 7.0/46 below moves the current point on by two
1714 characters after failing to match at the start. However, the pattern \nA
1715 *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
1716 the same is true. There doesn't seem any very clean way out of this, but
1717 what I have chosen to do makes the common cases work: PCRE now takes note
1718 of whether there can be an explicit match for \r or \n anywhere in the
1719 pattern, and if so, 7.0/46 no longer applies. As part of this change,
1720 there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
1721 pattern has explicit CR or LF references.
1722
1723 29. Added (*CR) etc for changing newline setting at start of pattern.
1724
1725
1726 Version 7.2 19-Jun-07
1727 ---------------------
1728
1729 1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
1730 which is apparently normally available under Windows.
1731
1732 2. Re-jig the pcregrep tests with different newline settings in an attempt
1733 to make them independent of the local environment's newline setting.
1734
1735 3. Add code to configure.ac to remove -g from the CFLAGS default settings.
1736
1737 4. Some of the "internals" tests were previously cut out when the link size
1738 was not 2, because the output contained actual offsets. The recent new
1739 "Z" feature of pcretest means that these can be cut out, making the tests
1740 usable with all link sizes.
1741
1742 5. Implemented Stan Switzer's goto replacement for longjmp() when not using
1743 stack recursion. This gives a massive performance boost under BSD, but just
1744 a small improvement under Linux. However, it saves one field in the frame
1745 in all cases.
1746
1747 6. Added more features from the forthcoming Perl 5.10:
1748
1749 (a) (?-n) (where n is a string of digits) is a relative subroutine or
1750 recursion call. It refers to the nth most recently opened parentheses.
1751
1752 (b) (?+n) is also a relative subroutine call; it refers to the nth next
1753 to be opened parentheses.
1754
1755 (c) Conditions that refer to capturing parentheses can be specified
1756 relatively, for example, (?(-2)... or (?(+3)...
1757
1758 (d) \K resets the start of the current match so that everything before
1759 is not part of it.
1760
1761 (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
1762
1763 (f) \g{name} is another synonym - part of Perl 5.10's unification of
1764 reference syntax.
1765
1766 (g) (?| introduces a group in which the numbering of parentheses in each
1767 alternative starts with the same number.
1768
1769 (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
1770
1771 7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
1772 PCRE_INFO_JCHANGED.
1773
1774 8. A pattern such as (.*(.)?)* caused pcre_exec() to fail by either not
1775 terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
1776 for detecting groups that can match an empty string.
1777
1778 9. A pattern with a very large number of alternatives (more than several
1779 hundred) was running out of internal workspace during the pre-compile
1780 phase, where pcre_compile() figures out how much memory will be needed. A
1781 bit of new cunning has reduced the workspace needed for groups with
1782 alternatives. The 1000-alternative test pattern now uses 12 bytes of
1783 workspace instead of running out of the 4096 that are available.
1784
1785 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
1786
1787 11. Applied patch from Google to remove an optimization that didn't quite work.
1788 The report of the bug said:
1789
1790 pcrecpp::RE("a*").FullMatch("aaa") matches, while
1791 pcrecpp::RE("a*?").FullMatch("aaa") does not, and
1792 pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
1793
1794 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
1795 it matched the wrong number of bytes.
1796
1797
1798 Version 7.1 24-Apr-07
1799 ---------------------
1800
1801 1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
1802 that is more "standard", making use of automake and other Autotools. There
1803 is some re-arrangement of the files and adjustment of comments consequent
1804 on this.
1805
1806 2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
1807 for recursive directory scanning broke on some systems because the files
1808 are not scanned in any specific order and on different systems the order
1809 was different. A call to "sort" has been inserted into RunGrepTest for the
1810 approprate test as a short-term fix. In the longer term there may be an
1811 alternative.
1812
1813 3. I had an email from Eric Raymond about problems translating some of PCRE's
1814 man pages to HTML (despite the fact that I distribute HTML pages, some
1815 people do their own conversions for various reasons). The problems
1816 concerned the use of low-level troff macros .br and .in. I have therefore
1817 removed all such uses from the man pages (some were redundant, some could
1818 be replaced by .nf/.fi pairs). The 132html script that I use to generate
1819 HTML has been updated to handle .nf/.fi and to complain if it encounters
1820 .br or .in.
1821
1822 4. Updated comments in configure.ac that get placed in config.h.in and also
1823 arranged for config.h to be included in the distribution, with the name
1824 config.h.generic, for the benefit of those who have to compile without
1825 Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
1826
1827 5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
1828 Weber: (1) pcre_internal.h was missing some function renames; (2) updated
1829 makevp.bat for the current PCRE, using the additional files
1830 makevp_c.txt, makevp_l.txt, and pcregexp.pas.
1831
1832 6. A Windows user reported a minor discrepancy with test 2, which turned out
1833 to be caused by a trailing space on an input line that had got lost in his
1834 copy. The trailing space was an accident, so I've just removed it.
1835
1836 7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
1837 that is needed.
1838
1839 8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
1840 as "const" (a) because they are and (b) because it helps the PHP
1841 maintainers who have recently made a script to detect big data structures
1842 in the php code that should be moved to the .rodata section. I remembered
1843 to update Builducptable as well, so it won't revert if ucptable.h is ever
1844 re-created.
1845
1846 9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
1847 pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
1848 order to be able to cut out the UTF-8 tables in the latter when UTF-8
1849 support is not required. This saves 1.5-2K of code, which is important in
1850 some applications.
1851
1852 Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
1853 so as not to refer to the tables, even though these functions will never be
1854 called when UTF-8 support is disabled. Otherwise there are problems with a
1855 shared library.
1856
1857 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
1858
1859 (a) It was defining its arguments as char * instead of void *.
1860
1861 (b) It was assuming that all moves were upwards in memory; this was true
1862 a long time ago when I wrote it, but is no longer the case.
1863
1864 The emulated memove() is provided for those environments that have neither
1865 memmove() nor bcopy(). I didn't think anyone used it these days, but that
1866 is clearly not the case, as these two bugs were recently reported.
1867
1868 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
1869 and Detrail to create the HTML documentation, the .txt form of the man
1870 pages, and it removes trailing spaces from listed files. It also creates
1871 pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
1872 case, it wraps all the #defines with #ifndefs. This script should be run
1873 before "make dist".
1874
1875 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
1876 with Unicode property support.
1877
1878 (a) For a maximizing quantifier, if the two different cases of the
1879 character were of different lengths in their UTF-8 codings (there are
1880 some cases like this - I found 11), and the matching function had to
1881 back up over a mixture of the two cases, it incorrectly assumed they
1882 were both the same length.
1883
1884 (b) When PCRE was configured to use the heap rather than the stack for
1885 recursion during matching, it was not correctly preserving the data for
1886 the other case of a UTF-8 character when checking ahead for a match
1887 while processing a minimizing repeat. If the check also involved
1888 matching a wide character, but failed, corruption could cause an
1889 erroneous result when trying to check for a repeat of the original
1890 character.
1891
1892 13. Some tidying changes to the testing mechanism:
1893
1894 (a) The RunTest script now detects the internal link size and whether there
1895 is UTF-8 and UCP support by running ./pcretest -C instead of relying on
1896 values substituted by "configure". (The RunGrepTest script already did
1897 this for UTF-8.) The configure.ac script no longer substitutes the
1898 relevant variables.
1899
1900 (b) The debugging options /B and /D in pcretest show the compiled bytecode
1901 with length and offset values. This means that the output is different
1902 for different internal link sizes. Test 2 is skipped for link sizes
1903 other than 2 because of this, bypassing the problem. Unfortunately,
1904 there was also a test in test 3 (the locale tests) that used /B and
1905 failed for link sizes other than 2. Rather than cut the whole test out,
1906 I have added a new /Z option to pcretest that replaces the length and
1907 offset values with spaces. This is now used to make test 3 independent
1908 of link size. (Test 2 will be tidied up later.)
1909
1910 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
1911 segmentation fault instead of returning the appropriate error message.
1912
1913 15. In multiline mode when the newline sequence was set to "any", the pattern
1914 ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
1915 This doesn't seem right; it now treats the CRLF combination as the line
1916 ending, and so does not match in that case. It's only a pattern such as ^$
1917 that would hit this one: something like ^ABC$ would have failed after \r
1918 and then tried again after \r\n.
1919
1920 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
1921 in an attempt to make files that differ only in their line terminators
1922 compare equal. This works on Linux.
1923
1924 17. Under certain error circumstances pcregrep might try to free random memory
1925 as it exited. This is now fixed, thanks to valgrind.
1926
1927 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
1928 "abc\r\n\r\n", it found an unwanted second match after the second \r. This
1929 was because its rules for how to advance for /g after matching an empty
1930 string at the end of a line did not allow for this case. They now check for
1931 it specially.
1932
1933 20. pcretest is supposed to handle patterns and data of any length, by
1934 extending its buffers when necessary. It was getting this wrong when the
1935 buffer for a data line had to be extended.
1936
1937 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
1938 CRLF as a newline sequence.
1939
1940 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
1941 out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
1942 I have nevertheless tidied it up.
1943
1944 23. Added some casts to kill warnings from HP-UX ia64 compiler.
1945
1946 24. Added a man page for pcre-config.
1947
1948
1949 Version 7.0 19-Dec-06
1950 ---------------------
1951
1952 1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
1953 moving to gcc 4.1.1.
1954
1955 2. The -S option for pcretest uses setrlimit(); I had omitted to #include
1956 sys/time.h, which is documented as needed for this function. It doesn't
1957 seem to matter on Linux, but it showed up on some releases of OS X.
1958
1959 3. It seems that there are systems where bytes whose values are greater than
1960 127 match isprint() in the "C" locale. The "C" locale should be the
1961 default when a C program starts up. In most systems, only ASCII printing
1962 characters match isprint(). This difference caused the output from pcretest
1963 to vary, making some of the tests fail. I have changed pcretest so that:
1964
1965 (a) When it is outputting text in the compiled version of a pattern, bytes
1966 other than 32-126 are always shown as hex escapes.
1967
1968 (b) When it is outputting text that is a matched part of a subject string,
1969 it does the same, unless a different locale has been set for the match
1970 (using the /L modifier). In this case, it uses isprint() to decide.
1971
1972 4. Fixed a major bug that caused incorrect computation of the amount of memory
1973 required for a compiled pattern when options that changed within the
1974 pattern affected the logic of the preliminary scan that determines the
1975 length. The relevant options are -x, and -i in UTF-8 mode. The result was
1976 that the computed length was too small. The symptoms of this bug were
1977 either the PCRE error "internal error: code overflow" from pcre_compile(),
1978 or a glibc crash with a message such as "pcretest: free(): invalid next
1979 size (fast)". Examples of patterns that provoked this bug (shown in
1980 pcretest format) are:
1981
1982 /(?-x: )/x
1983 /(?x)(?-x: \s*#\s*)/
1984 /((?i)[\x{c0}])/8
1985 /(?i:[\x{c0}])/8
1986
1987 HOWEVER: Change 17 below makes this fix obsolete as the memory computation
1988 is now done differently.
1989
1990 5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
1991 wrapper classes; (b) implement a new function in the C++ scanner that is
1992 more efficient than the old way of doing things because it avoids levels of
1993 recursion in the regex matching; (c) add a paragraph to the documentation
1994 for the FullMatch() function.
1995
1996 6. The escape sequence \n was being treated as whatever was defined as
1997 "newline". Not only was this contrary to the documentation, which states
1998 that \n is character 10 (hex 0A), but it also went horribly wrong when
1999 "newline" was defined as CRLF. This has been fixed.
2000
2001 7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
2002 was being set to -1 for the "end of line" case (supposedly a value that no
2003 character can have). Though this value is never used (the check for end of
2004 line is "zero bytes in current character"), it caused compiler complaints.
2005 I've changed it to 0xffffffff.
2006
2007 8. In pcre_version.c, the version string was being built by a sequence of
2008 C macros that, in the event of PCRE_PRERELEASE being defined as an empty
2009 string (as it is for production releases) called a macro with an empty
2010 argument. The C standard says the result of this is undefined. The gcc
2011 compiler treats it as an empty string (which was what was wanted) but it is
2012 reported that Visual C gives an error. The source has been hacked around to
2013 avoid this problem.
2014
2015 9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
2016 builds of pcretest, and changed the call to _setmode() to use _O_BINARY
2017 instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
2018 of them did).
2019
2020 10. Originally, pcretest opened its input and output without "b"; then I was
2021 told that "b" was needed in some environments, so it was added for release
2022 5.0 to both the input and output. (It makes no difference on Unix-like
2023 systems.) Later I was told that it is wrong for the input on Windows. I've
2024 now abstracted the modes into two macros, to make it easier to fiddle with
2025 them, and removed "b" from the input mode under Windows.
2026
2027 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
2028
2029 12. Added -help and --help to pcretest as an official way of being reminded
2030 of the options.
2031
2032 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
2033 and pcrecpp.cc because they annoy compilers at high warning levels.
2034
2035 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
2036
2037 15. Fixed an occurrence of == in configure.ac that should have been = (shell
2038 scripts are not C programs :-) and which was not noticed because it works
2039 on Linux.
2040
2041 16. pcretest is supposed to handle any length of pattern and data line (as one
2042 line or as a continued sequence of lines) by extending its input buffer if
2043 necessary. This feature was broken for very long pattern lines, leading to
2044 a string of junk being passed to pcre_compile() if the pattern was longer
2045 than about 50K.
2046
2047 17. I have done a major re-factoring of the way pcre_compile() computes the
2048 amount of memory needed for a compiled pattern. Previously, there was code
2049 that made a preliminary scan of the pattern in order to do this. That was
2050 OK when PCRE was new, but as the facilities have expanded, it has become
2051 harder and harder to keep it in step with the real compile phase, and there
2052 have been a number of bugs (see for example, 4 above). I have now found a
2053 cunning way of running the real compile function in a "fake" mode that
2054 enables it to compute how much memory it would need, while actually only
2055 ever using a few hundred bytes of working memory and without too many
2056 tests of the mode. This should make future maintenance and development
2057 easier. A side effect of this work is that the limit of 200 on the nesting
2058 depth of parentheses has been removed (though this was never a serious
2059 limitation, I suspect). However, there is a downside: pcre_compile() now
2060 runs more slowly than before (30% or more, depending on the pattern). I
2061 hope this isn't a big issue. There is no effect on runtime performance.
2062
2063 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
2064 newline (only possible for the last line of a file) and it was a
2065 pattern that set a locale (followed by /Lsomething), pcretest crashed.
2066
2067 19. Added additional timing features to pcretest. (1) The -tm option now times
2068 matching only, not compiling. (2) Both -t and -tm can be followed, as a
2069 separate command line item, by a number that specifies the number of
2070 repeats to use when timing. The default is 50000; this gives better
2071 precision, but takes uncomfortably long for very large patterns.
2072
2073 20. Extended pcre_study() to be more clever in cases where a branch of a
2074 subpattern has no definite first character. For example, (a*|b*)[cd] would
2075 previously give no result from pcre_study(). Now it recognizes that the
2076 first character must be a, b, c, or d.
2077
2078 21. There was an incorrect error "recursive call could loop indefinitely" if
2079 a subpattern (or the entire pattern) that was being tested for matching an
2080 empty string contained only one non-empty item after a nested subpattern.
2081 For example, the pattern (?>\x{100}*)\d(?R) provoked this error
2082 incorrectly, because the \d was being skipped in the check.
2083
2084 22. The pcretest program now has a new pattern option /B and a command line
2085 option -b, which is equivalent to adding /B to every pattern. This causes
2086 it to show the compiled bytecode, without the additional information that
2087 -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
2088 is the same as /B/I).
2089
2090 23. A new optimization is now able automatically to treat some sequences such
2091 as a*b as a*+b. More specifically, if something simple (such as a character
2092 or a simple class like \d) has an unlimited quantifier, and is followed by
2093 something that cannot possibly match the quantified thing, the quantifier
2094 is automatically "possessified".
2095
2096 24. A recursive reference to a subpattern whose number was greater than 39
2097 went wrong under certain circumstances in UTF-8 mode. This bug could also
2098 have affected the operation of pcre_study().
2099
2100 25. Realized that a little bit of performance could be had by replacing
2101 (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
2102
2103 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
2104
2105 27. Possessive quantifiers such as a++ were previously implemented by turning
2106 them into atomic groups such as ($>a+). Now they have their own opcodes,
2107 which improves performance. This includes the automatically created ones
2108 from 23 above.
2109
2110 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
2111 lookahead was broken if it was not anchored. PCRE was mistakenly expecting
2112 the first matched character to be a colon. This applied both to named and
2113 numbered groups.
2114
2115 29. The ucpinternal.h header file was missing its idempotency #ifdef.
2116
2117 30. I was sent a "project" file called libpcre.a.dev which I understand makes
2118 building PCRE on Windows easier, so I have included it in the distribution.
2119
2120 31. There is now a check in pcretest against a ridiculously large number being
2121 returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
2122 loop, the loop is abandoned.
2123
2124 32. Forward references to subpatterns in conditions such as (?(2)...) where
2125 subpattern 2 is defined later cause pcre_compile() to search forwards in
2126 the pattern for the relevant set of parentheses. This search went wrong
2127 when there were unescaped parentheses in a character class, parentheses
2128 escaped with \Q...\E, or parentheses in a #-comment in /x mode.
2129
2130 33. "Subroutine" calls and backreferences were previously restricted to
2131 referencing subpatterns earlier in the regex. This restriction has now
2132 been removed.
2133
2134 34. Added a number of extra features that are going to be in Perl 5.10. On the
2135 whole, these are just syntactic alternatives for features that PCRE had
2136 previously implemented using the Python syntax or my own invention. The
2137 other formats are all retained for compatibility.
2138
2139 (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
2140 as (?P<name>...). The new forms, as well as being in Perl 5.10, are
2141 also .NET compatible.
2142
2143 (b) A recursion or subroutine call to a named group can now be defined as
2144 (?&name) as well as (?P>name).
2145
2146 (c) A backreference to a named group can now be defined as \k<name> or
2147 \k'name' as well as (?P=name). The new forms, as well as being in Perl
2148 5.10, are also .NET compatible.
2149
2150 (d) A conditional reference to a named group can now use the syntax
2151 (?(<name>) or (?('name') as well as (?(name).
2152
2153 (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
2154 groups (named and numbered) that are never evaluated inline, but can be
2155 called as "subroutines" from elsewhere. In effect, the DEFINE condition
2156 is always false. There may be only one alternative in such a group.
2157
2158 (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
2159 as the simple (?(R). The condition is true only if the most recent
2160 recursion is that of the given number or name. It does not search out
2161 through the entire recursion stack.
2162
2163 (g) The escape \gN or \g{N} has been added, where N is a positive or
2164 negative number, specifying an absolute or relative reference.
2165
2166 35. Tidied to get rid of some further signed/unsigned compiler warnings and
2167 some "unreachable code" warnings.
2168
2169 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
2170 things, this adds five new scripts.
2171
2172 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
2173 There were also incompatibilities regarding the handling of \Q..\E inside
2174 character classes, for example with patterns like [\Qa\E-\Qz\E] where the
2175 hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
2176
2177 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
2178 matches an empty string, and forcibly breaks the loop. There were bugs in
2179 this code in non-simple cases. For a pattern such as ^(a()*)* matched
2180 against aaaa the result was just "a" rather than "aaaa", for example. Two
2181 separate and independent bugs (that affected different cases) have been
2182 fixed.
2183
2184 39. Refactored the code to abolish the use of different opcodes for small
2185 capturing bracket numbers. This is a tidy that I avoided doing when I
2186 removed the limit on the number of capturing brackets for 3.5 back in 2001.
2187 The new approach is not only tidier, it makes it possible to reduce the
2188 memory needed to fix the previous bug (38).
2189
2190 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
2191 sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
2192 processing dot, circumflex, or dollar metacharacters, or #-comments in /x
2193 mode.
2194
2195 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
2196 report.
2197
2198 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
2199 copy construction and assignment in the C++ wrapper.
2200
2201 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
2202 couple of bugs that could have given wrong results in the "--newline=crlf"
2203 case.
2204
2205 44. Added a number of casts and did some reorganization of signed/unsigned int
2206 variables following suggestions from Dair Grant. Also renamed the variable
2207 "this" as "item" because it is a C++ keyword.
2208
2209 45. Arranged for dftables to add
2210
2211 #include "pcre_internal.h"
2212
2213 to pcre_chartables.c because without it, gcc 4.x may remove the array
2214 definition from the final binary if PCRE is built into a static library and
2215 dead code stripping is activated.
2216
2217 46. For an unanchored pattern, if a match attempt fails at the start of a
2218 newline sequence, and the newline setting is CRLF or ANY, and the next two
2219 characters are CRLF, advance by two characters instead of one.
2220
2221
2222 Version 6.7 04-Jul-06
2223 ---------------------
2224
2225 1. In order to handle tests when input lines are enormously long, pcretest has
2226 been re-factored so that it automatically extends its buffers when
2227 necessary. The code is crude, but this _is_ just a test program. The
2228 default size has been increased from 32K to 50K.
2229
2230 2. The code in pcre_study() was using the value of the re argument before
2231 testing it for NULL. (Of course, in any sensible call of the function, it
2232 won't be NULL.)
2233
2234 3. The memmove() emulation function in pcre_internal.h, which is used on
2235 systems that lack both memmove() and bcopy() - that is, hardly ever -
2236 was missing a "static" storage class specifier.
2237
2238 4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
2239 containing an extended class (one that cannot be represented by a bitmap
2240 because it contains high-valued characters or Unicode property items, e.g.
2241 [\pZ]). Almost always one would set UTF-8 mode when processing such a
2242 pattern, but PCRE should not loop if you do not (it no longer does).
2243 [Detail: two cases were found: (a) a repeated subpattern containing an
2244 extended class; (b) a recursive reference to a subpattern that followed a
2245 previous extended class. It wasn't skipping over the extended class
2246 correctly when UTF-8 mode was not set.]
2247
2248 5. A negated single-character class was not being recognized as fixed-length
2249 in lookbehind assertions such as (?<=[^f]), leading to an incorrect
2250 compile error "lookbehind assertion is not fixed length".
2251
2252 6. The RunPerlTest auxiliary script was showing an unexpected difference
2253 between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
2254 write a Perl script that can interpret lines of an input file either as
2255 byte characters or as UTF-8, which is what "perltest" was being required to
2256 do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
2257 can't do is switch easily at run time between having the "use utf8;" pragma
2258 or not. In the end, I fudged it by using the RunPerlTest script to insert
2259 "use utf8;" explicitly for the UTF-8 tests.
2260
2261 7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
2262 the end of the subject string, contrary to the documentation and to what
2263 Perl does. This was true of both matching functions. Now it matches only at
2264 the start of the subject and immediately after *internal* newlines.
2265
2266 8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
2267 a pointer to an int instead of a pointer to an unsigned long int. This
2268 caused problems on 64-bit systems.
2269
2270 9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
2271 instance of the 'standard' template library not being so standard".
2272
2273 10. There was no check on the number of named subpatterns nor the maximum
2274 length of a subpattern name. The product of these values is used to compute
2275 the size of the memory block for a compiled pattern. By supplying a very
2276 long subpattern name and a large number of named subpatterns, the size
2277 computation could be caused to overflow. This is now prevented by limiting
2278 the length of names to 32 characters, and the number of named subpatterns
2279 to 10,000.
2280
2281 11. Subpatterns that are repeated with specific counts have to be replicated in
2282 the compiled pattern. The size of memory for this was computed from the
2283 length of the subpattern and the repeat count. The latter is limited to
2284 65535, but there was no limit on the former, meaning that integer overflow
2285 could in principle occur. The compiled length of a repeated subpattern is
2286 now limited to 30,000 bytes in order to prevent this.
2287
2288 12. Added the optional facility to have named substrings with the same name.
2289
2290 13. Added the ability to use a named substring as a condition, using the
2291 Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
2292 are numbers (not recommended). Forward references are permitted.
2293
2294 14. Added forward references in named backreferences (if you see what I mean).
2295
2296 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
2297 pattern could run off the end of the subject. For example, the pattern
2298 "(?s)(.{1,5})"8 did this with the subject "ab".
2299
2300 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
2301 PCRE_CASELESS was set when matching characters that were quantified with ?
2302 or *.
2303
2304 17. A character class other than a single negated character that had a minimum
2305 but no maximum quantifier - for example [ab]{6,} - was not handled
2306 correctly by pce_dfa_exec(). It would match only one character.
2307
2308 18. A valid (though odd) pattern that looked like a POSIX character
2309 class but used an invalid character after [ (for example [[,abc,]]) caused
2310 pcre_compile() to give the error "Failed: internal error: code overflow" or
2311 in some cases to crash with a glibc free() error. This could even happen if
2312 the pattern terminated after [[ but there just happened to be a sequence of
2313 letters, a binary zero, and a closing ] in the memory that followed.
2314
2315 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
2316 over the years. Originally (before any Unicode support), just the bottom 8
2317 bits were taken. Thus, for example, \500 really meant \100. Nowadays the
2318 output from "man perlunicode" includes this:
2319
2320 The regular expression compiler produces polymorphic opcodes. That
2321 is, the pattern adapts to the data and automatically switches to
2322 the Unicode character scheme when presented with Unicode data--or
2323 instead uses a traditional byte scheme when presented with byte
2324 data.
2325
2326 Sadly, a wide octal escape does not cause a switch, and in a string with
2327 no other multibyte characters, these octal escapes are treated as before.
2328 Thus, in Perl, the pattern /\500/ actually matches \100 but the pattern
2329 /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
2330 Unicode string.
2331
2332 I have not perpetrated such confusion in PCRE. Up till now, it took just
2333 the bottom 8 bits, as in old Perl. I have now made octal escapes with
2334 values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
2335 translate to the appropriate multibyte character.
2336
2337 29. Applied some refactoring to reduce the number of warnings from Microsoft
2338 and Borland compilers. This has included removing the fudge introduced
2339 seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
2340 a warning about an unused variable.
2341
2342 21. PCRE has not included VT (character 0x0b) in the set of whitespace
2343 characters since release 4.0, because Perl (from release 5.004) does not.
2344 [Or at least, is documented not to: some releases seem to be in conflict
2345 with the documentation.] However, when a pattern was studied with
2346 pcre_study() and all its branches started with \s, PCRE still included VT
2347 as a possible starting character. Of course, this did no harm; it just
2348 caused an unnecessary match attempt.
2349
2350 22. Removed a now-redundant internal flag bit that recorded the fact that case
2351 dependency changed within the pattern. This was once needed for "required
2352 byte" processing, but is no longer used. This recovers a now-scarce options
2353 bit. Also moved the least significant internal flag bit to the most-
2354 significant bit of the word, which was not previously used (hangover from
2355 the days when it was an int rather than a uint) to free up another bit for
2356 the future.
2357
2358 23. Added support for CRLF line endings as well as CR and LF. As well as the
2359 default being selectable at build time, it can now be changed at runtime
2360 via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
2361 specify that it is scanning data with non-default line endings.
2362
2363 24. Changed the definition of CXXLINK to make it agree with the definition of
2364 LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
2365
2366 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
2367 recursions. This makes a big different to stack usage for some patterns.
2368
2369 26. If a subpattern containing a named recursion or subroutine reference such
2370 as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
2371 the space required for the compiled pattern went wrong and gave too small a
2372 value. Depending on the environment, this could lead to "Failed: internal
2373 error: code overflow at offset 49" or "glibc detected double free or
2374 corruption" errors.
2375
2376 27. Applied patches from Google (a) to support the new newline modes and (b) to
2377 advance over multibyte UTF-8 characters in GlobalReplace.
2378
2379 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
2380 difference for some implementation of PCRE in some Windows version.
2381
2382 29. Added some extra testing facilities to pcretest:
2383
2384 \q<number> in a data line sets the "match limit" value
2385 \Q<number> in a data line sets the "match recursion limt" value
2386 -S <number> sets the stack size, where <number> is in megabytes
2387
2388 The -S option isn't available for Windows.
2389
2390
2391 Version 6.6 06-Feb-06
2392 ---------------------
2393
2394 1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
2395 in pcreposix.h. I have copied the definition from pcre.h.
2396
2397 2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
2398 because pcre.h is no longer a built file.
2399
2400 3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
2401 not normally included in the compiled code.
2402
2403
2404 Version 6.5 01-Feb-06
2405 ---------------------
2406
2407 1. When using the partial match feature with pcre_dfa_exec(), it was not
2408 anchoring the second and subsequent partial matches at the new starting
2409 point. This could lead to incorrect results. For example, with the pattern
2410 /1234/, partially matching against "123" and then "a4" gave a match.
2411
2412 2. Changes to pcregrep:
2413
2414 (a) All non-match returns from pcre_exec() were being treated as failures
2415 to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
2416 error message is output. Some extra information is given for the
2417 PCRE_ERROR_MATCHLIMIT and PCRE_ERROR_RECURSIONLIMIT errors, which are
2418 probably the only errors that are likely to be caused by users (by
2419 specifying a regex that has nested indefinite repeats, for instance).
2420 If there are more than 20 of these errors, pcregrep is abandoned.
2421
2422 (b) A binary zero was treated as data while matching, but terminated the
2423 output line if it was written out. This has been fixed: binary zeroes
2424 are now no different to any other data bytes.
2425
2426 (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
2427 used to set a locale for matching. The --locale=xxxx long option has
2428 been added (no short equivalent) to specify a locale explicitly on the
2429 pcregrep command, overriding the environment variables.
2430
2431 (d) When -B was used with -n, some line numbers in the output were one less
2432 than they should have been.
2433
2434 (e) Added the -o (--only-matching) option.
2435
2436 (f) If -A or -C was used with -c (count only), some lines of context were
2437 accidentally printed for the final match.
2438
2439 (g) Added the -H (--with-filename) option.
2440
2441 (h) The combination of options -rh failed to suppress file names for files
2442 that were found from directory arguments.
2443
2444 (i) Added the -D (--devices) and -d (--directories) options.
2445
2446 (j) Added the -F (--fixed-strings) option.
2447
2448 (k) Allow "-" to be used as a file name for -f as well as for a data file.
2449
2450 (l) Added the --colo(u)r option.
2451
2452 (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
2453 is not present by default.
2454
2455 3. A nasty bug was discovered in the handling of recursive patterns, that is,
2456 items such as (?R) or (?1), when the recursion could match a number of
2457 alternatives. If it matched one of the alternatives, but subsequently,
2458 outside the recursion, there was a failure, the code tried to back up into
2459 the recursion. However, because of the way PCRE is implemented, this is not
2460 possible, and the result was an incorrect result from the match.
2461
2462 In order to prevent this happening, the specification of recursion has
2463 been changed so that all such subpatterns are automatically treated as
2464 atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
2465
2466 4. I had overlooked the fact that, in some locales, there are characters for
2467 which isalpha() is true but neither isupper() nor islower() are true. In
2468 the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
2469 and ordfeminine) are like this. This affected the treatment of \w and \W
2470 when they appeared in character classes, but not when they appeared outside
2471 a character class. The bit map for "word" characters is now created
2472 separately from the results of isalnum() instead of just taking it from the
2473 upper, lower, and digit maps. (Plus the underscore character, of course.)
2474
2475 5. The above bug also affected the handling of POSIX character classes such as
2476 [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
2477 permanent tables. Instead, the bit maps for such a class were previously
2478 created as the appropriate unions of the upper, lower, and digit bitmaps.
2479 Now they are created by subtraction from the [[:word:]] class, which has
2480 its own bitmap.
2481
2482 6. The [[:blank:]] character class matches horizontal, but not vertical space.
2483 It is created by subtracting the vertical space characters (\x09, \x0a,
2484 \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
2485 subtraction was done in the overall bitmap for a character class, meaning
2486 that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
2487 be recognized. This bug has been fixed.
2488
2489 7. Patches from the folks at Google:
2490
2491 (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
2492 real life, but is still worth protecting against".
2493
2494 (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
2495 regular expressions".
2496
2497 (c) pcre_scanner.cc: avoid use of std::count() because not all systems
2498 have it.
2499
2500 (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
2501 "configure" and the latter not, in order to fix a problem somebody had
2502 with compiling the Arg class on HP-UX.
2503
2504 (e) Improve the error-handling of the C++ wrapper a little bit.
2505
2506 (f) New tests for checking recursion limiting.
2507
2508 8. The pcre_memmove() function, which is used only if the environment does not
2509 have a standard memmove() function (and is therefore rarely compiled),
2510 contained two bugs: (a) use of int instead of size_t, and (b) it was not
2511 returning a result (though PCRE never actually uses the result).
2512
2513 9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
2514 large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
2515 returned instead of calling malloc() with an overflowing number that would
2516 most likely cause subsequent chaos.
2517
2518 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
2519
2520 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
2521 with this option is matched, the nmatch and pmatch options of regexec() are
2522 ignored.
2523
2524 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
2525 provided in case anyone wants to the the POSIX interface with UTF-8
2526 strings.
2527
2528 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
2529 C++ linking (needed for some HP-UX environments).
2530
2531 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
2532 (unused parameter) and in the pcre_printint() function (omitted "default"
2533 switch label when the default is to do nothing).
2534
2535 15. Added some code to make it possible, when PCRE is compiled as a C++
2536 library, to replace subject pointers for pcre_exec() with a smart pointer
2537 class, thus making it possible to process discontinuous strings.
2538
2539 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
2540 much the same function. They were added by different people who were trying
2541 to make PCRE easy to compile on non-Unix systems. It has been suggested
2542 that PCRE_EXPORT be abolished now that there is more automatic apparatus
2543 for compiling on Windows systems. I have therefore replaced it with
2544 PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
2545 defaults to "extern" for C or "extern C" for C++, which works fine on
2546 Unix-like systems. It is now possible to override the value of PCRE_DATA_
2547 SCOPE with something explicit in config.h. In addition:
2548
2549 (a) pcreposix.h still had just "extern" instead of either of these macros;
2550 I have replaced it with PCRE_DATA_SCOPE.
2551
2552 (b) Functions such as _pcre_xclass(), which are internal to the library,
2553 but external in the C sense, all had PCRE_EXPORT in their definitions.
2554 This is apparently wrong for the Windows case, so I have removed it.
2555 (It makes no difference on Unix-like systems.)
2556
2557 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
2558 of recursive calls to match(). This is different to MATCH_LIMIT because
2559 that limits the total number of calls to match(), not all of which increase
2560 the depth of recursion. Limiting the recursion depth limits the amount of
2561 stack (or heap if NO_RECURSE is set) that is used. The default can be set
2562 when PCRE is compiled, and changed at run time. A patch from Google adds
2563 this functionality to the C++ interface.
2564
2565 18. Changes to the handling of Unicode character properties:
2566
2567 (a) Updated the table to Unicode 4.1.0.
2568
2569 (b) Recognize characters that are not in the table as "Cn" (undefined).
2570
2571 (c) I revised the way the table is implemented to a much improved format
2572 which includes recognition of ranges. It now supports the ranges that
2573 are defined in UnicodeData.txt, and it also amalgamates other
2574 characters into ranges. This has reduced the number of entries in the
2575 table from around 16,000 to around 3,000, thus reducing its size
2576 considerably. I realized I did not need to use a tree structure after
2577 all - a binary chop search is just as efficient. Having reduced the
2578 number of entries, I extended their size from 6 bytes to 8 bytes to
2579 allow for more data.
2580
2581 (d) Added support for Unicode script names via properties such as \p{Han}.
2582
2583 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
2584 matching that character.
2585
2586 20. When matching a repeated Unicode property with a minimum greater than zero,
2587 (for example \pL{2,}), PCRE could look past the end of the subject if it
2588 reached it while seeking the minimum number of characters. This could
2589 happen only if some of the characters were more than one byte long, because
2590 there is a check for at least the minimum number of bytes.
2591
2592 21. Refactored the implementation of \p and \P so as to be more general, to
2593 allow for more different types of property in future. This has changed the
2594 compiled form incompatibly. Anybody with saved compiled patterns that use
2595 \p or \P will have to recompile them.
2596
2597 22. Added "Any" and "L&" to the supported property types.
2598
2599 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
2600 but give a compile time error if the value is greater than 0xff.
2601
2602 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
2603 accidentally not being installed or uninstalled.
2604
2605 25. The pcre.h file was built from pcre.h.in, but the only changes that were
2606 made were to insert the current release number. This seemed silly, because
2607 it made things harder for people building PCRE on systems that don't run
2608 "configure". I have turned pcre.h into a distributed file, no longer built
2609 by "configure", with the version identification directly included. There is
2610 no longer a pcre.h.in file.
2611
2612 However, this change necessitated a change to the pcre-config script as
2613 well. It is built from pcre-config.in, and one of the substitutions was the
2614 release number. I have updated configure.ac so that ./configure now finds
2615 the release number by grepping pcre.h.
2616
2617 26. Added the ability to run the tests under valgrind.
2618
2619
2620 Version 6.4 05-Sep-05
2621 ---------------------
2622
2623 1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
2624 "--" to be printed when multiple files were scanned, even when none of the
2625 -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
2626 consider it to be a bug, and have restored the previous behaviour.
2627
2628 2. A couple of code tidies to get rid of compiler warnings.
2629
2630 3. The pcretest program used to cheat by referring to symbols in the library
2631 whose names begin with _pcre_. These are internal symbols that are not
2632 really supposed to be visible externally, and in some environments it is
2633 possible to suppress them. The cheating is now confined to including
2634 certain files from the library's source, which is a bit cleaner.
2635
2636 4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
2637 file's purpose clearer.
2638
2639 5. Reorganized pcre_ucp_findchar().
2640
2641
2642 Version 6.3 15-Aug-05
2643 ---------------------
2644
2645 1. The file libpcre.pc.in did not have general read permission in the tarball.
2646
2647 2. There were some problems when building without C++ support:
2648
2649 (a) If C++ support was not built, "make install" and "make test" still
2650 tried to test it.
2651
2652 (b) There were problems when the value of CXX was explicitly set. Some
2653 changes have been made to try to fix these, and ...
2654
2655 (c) --disable-cpp can now be used to explicitly disable C++ support.
2656
2657 (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
2658 backslash in a target when C++ was disabled. This confuses some
2659 versions of "make", apparently. Using an intermediate variable solves
2660 this. (Same for CPP_LOBJ.)
2661
2662 3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
2663 (non-Windows) now includes $(CFLAGS) because these flags are sometimes
2664 necessary on certain architectures.
2665
2666 4. Added a setting of -export-symbols-regex to the link command to remove
2667 those symbols that are exported in the C sense, but actually are local
2668 within the library, and not documented. Their names all begin with
2669 "_pcre_". This is not a perfect job, because (a) we have to except some
2670 symbols that pcretest ("illegally") uses, and (b) the facility isn't always
2671 available (and never for static libraries). I have made a note to try to
2672 find a way round (a) in the future.
2673
2674
2675 Version 6.2 01-Aug-05
2676 ---------------------
2677
2678 1. There was no test for integer overflow of quantifier values. A construction
2679 such as {1111111111111111} would give undefined results. What is worse, if
2680 a minimum quantifier for a parenthesized subpattern overflowed and became
2681 negative, the calculation of the memory size went wrong. This could have
2682 led to memory overwriting.
2683
2684 2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
2685
2686 3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
2687 operating environments where this matters.
2688
2689 4. Applied Giuseppe Maxia's patch to add additional features for controlling
2690 PCRE options from within the C++ wrapper.
2691
2692 5. Named capturing subpatterns were not being correctly counted when a pattern
2693 was compiled. This caused two problems: (a) If there were more than 100
2694 such subpatterns, the calculation of the memory needed for the whole
2695 compiled pattern went wrong, leading to an overflow error. (b) Numerical
2696 back references of the form \12, where the number was greater than 9, were
2697 not recognized as back references, even though there were sufficient
2698 previous subpatterns.
2699
2700 6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
2701 versions of gcc, e.g. 2.95.4.
2702
2703
2704 Version 6.1 21-Jun-05
2705 ---------------------
2706
2707 1. There was one reference to the variable "posix" in pcretest.c that was not
2708 surrounded by "#if !defined NOPOSIX".
2709
2710 2. Make it possible to compile pcretest without DFA support, UTF8 support, or
2711 the cross-check on the old pcre_info() function, for the benefit of the
2712 cut-down version of PCRE that is currently imported into Exim.
2713
2714 3. A (silly) pattern starting with (?i)(?-i) caused an internal space
2715 allocation error. I've done the easy fix, which wastes 2 bytes for sensible
2716 patterns that start (?i) but I don't think that matters. The use of (?i) is
2717 just an example; this all applies to the other options as well.
2718
2719 4. Since libtool seems to echo the compile commands it is issuing, the output
2720 from "make" can be reduced a bit by putting "@" in front of each libtool
2721 compile command.
2722
2723 5. Patch from the folks at Google for configure.in to be a bit more thorough
2724 in checking for a suitable C++ installation before trying to compile the
2725 C++ stuff. This should fix a reported problem when a compiler was present,
2726 but no suitable headers.
2727
2728 6. The man pages all had just "PCRE" as their title. I have changed them to
2729 be the relevant file name. I have also arranged that these names are
2730 retained in the file doc/pcre.txt, which is a concatenation in text format
2731 of all the man pages except the little individual ones for each function.
2732
2733 7. The NON-UNIX-USE file had not been updated for the different set of source
2734 files that come with release 6. I also added a few comments about the C++
2735 wrapper.
2736
2737
2738 Version 6.0 07-Jun-05
2739 ---------------------
2740
2741 1. Some minor internal re-organization to help with my DFA experiments.
2742
2743 2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
2744 didn't matter for the library itself when fully configured, but did matter
2745 when compiling without UCP support, or within Exim, where the ucp files are
2746 not imported.
2747
2748 3. Refactoring of the library code to split up the various functions into
2749 different source modules. The addition of the new DFA matching code (see
2750 below) to a single monolithic source would have made it really too
2751 unwieldy, quite apart from causing all the code to be include in a
2752 statically linked application, when only some functions are used. This is
2753 relevant even without the DFA addition now that patterns can be compiled in
2754 one application and matched in another.
2755
2756 The downside of splitting up is that there have to be some external
2757 functions and data tables that are used internally in different modules of
2758 the library but which are not part of the API. These have all had their
2759 names changed to start with "_pcre_" so that they are unlikely to clash
2760 with other external names.
2761
2762 4. Added an alternate matching function, pcre_dfa_exec(), which matches using
2763 a different (DFA) algorithm. Although it is slower than the original
2764 function, it does have some advantages for certain types of matching
2765 problem.
2766
2767 5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
2768 including restarting after a partial match.
2769
2770 6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
2771 defined when compiling for Windows was sent to me. I have put it into the
2772 code, though I have no means of testing or verifying it.
2773
2774 7. Added the pcre_refcount() auxiliary function.
2775
2776 8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
2777 match before or at the first newline in the subject string. In pcretest,
2778 the /f option on a pattern can be used to set this.
2779
2780 9. A repeated \w when used in UTF-8 mode with characters greater than 256
2781 would behave wrongly. This has been present in PCRE since release 4.0.
2782
2783 10. A number of changes to the pcregrep command:
2784
2785 (a) Refactored how -x works; insert ^(...)$ instead of setting
2786 PCRE_ANCHORED and checking the length, in preparation for adding
2787 something similar for -w.
2788
2789 (b) Added the -w (match as a word) option.
2790
2791 (c) Refactored the way lines are read and buffered so as to have more
2792 than one at a time available.
2793
2794 (d) Implemented a pcregrep test script.
2795
2796 (e) Added the -M (multiline match) option. This allows patterns to match
2797 over several lines of the subject. The buffering ensures that at least
2798 8K, or the rest of the document (whichever is the shorter) is available
2799 for matching (and similarly the previous 8K for lookbehind assertions).
2800
2801 (f) Changed the --help output so that it now says
2802
2803 -w, --word-regex(p)
2804
2805 instead of two lines, one with "regex" and the other with "regexp"
2806 because that confused at least one person since the short forms are the
2807 same. (This required a bit of code, as the output is generated
2808 automatically from a table. It wasn't just a text change.)
2809
2810 (g) -- can be used to terminate pcregrep options if the next thing isn't an
2811 option but starts with a hyphen. Could be a pattern or a path name
2812 starting with a hyphen, for instance.
2813
2814 (h) "-" can be given as a file name to represent stdin.
2815
2816 (i) When file names are being printed, "(standard input)" is used for
2817 the standard input, for compatibility with GNU grep. Previously
2818 "<stdin>" was used.
2819
2820 (j) The option --label=xxx can be used to supply a name to be used for
2821 stdin when file names are being printed. There is no short form.
2822
2823 (k) Re-factored the options decoding logic because we are going to add
2824 two more options that take data. Such options can now be given in four
2825 different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
2826
2827 (l) Added the -A, -B, and -C options for requesting that lines of context
2828 around matches be printed.
2829
2830 (m) Added the -L option to print the names of files that do not contain
2831 any matching lines, that is, the complement of -l.
2832
2833 (n) The return code is 2 if any file cannot be opened, but pcregrep does
2834 continue to scan other files.
2835
2836 (o) The -s option was incorrectly implemented. For compatibility with other
2837 greps, it now suppresses the error message for a non-existent or non-
2838 accessible file (but not the return code). There is a new option called
2839 -q that suppresses the output of matching lines, which was what -s was
2840 previously doing.
2841
2842 (p) Added --include and --exclude options to specify files for inclusion
2843 and exclusion when recursing.
2844
2845 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
2846 Hopefully, it now does.
2847
2848 12. Missing cast in pcre_study().
2849
2850 13. Added an "uninstall" target to the makefile.
2851
2852 14. Replaced "extern" in the function prototypes in Makefile.in with
2853 "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
2854 world, but is set differently for Windows.
2855
2856 15. Added a second compiling function called pcre_compile2(). The only
2857 difference is that it has an extra argument, which is a pointer to an
2858 integer error code. When there is a compile-time failure, this is set
2859 non-zero, in addition to the error test pointer being set to point to an
2860 error message. The new argument may be NULL if no error number is required
2861 (but then you may as well call pcre_compile(), which is now just a
2862 wrapper). This facility is provided because some applications need a
2863 numeric error indication, but it has also enabled me to tidy up the way
2864 compile-time errors are handled in the POSIX wrapper.
2865
2866 16. Added VPATH=.libs to the makefile; this should help when building with one
2867 prefix path and installing with another. (Or so I'm told by someone who
2868 knows more about this stuff than I do.)
2869
2870 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
2871 passes PCRE_DOTALL to the pcre_compile() function, making the "." character
2872 match everything, including newlines. This is not POSIX-compatible, but
2873 somebody wanted the feature. From pcretest it can be activated by using
2874 both the P and the s flags.
2875
2876 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
2877
2878 19. libpcre.pc was being incorrectly installed as executable.
2879
2880 20. A couple of places in pcretest check for end-of-line by looking for '\n';
2881 it now also looks for '\r' so that it will work unmodified on Windows.
2882
2883 21. Added Google's contributed C++ wrapper to the distribution.
2884
2885 22. Added some untidy missing memory free() calls in pcretest, to keep
2886 Electric Fence happy when testing.
2887
2888
2889
2890 Version 5.0 13-Sep-04
2891 ---------------------
2892
2893 1. Internal change: literal characters are no longer packed up into items
2894 containing multiple characters in a single byte-string. Each character
2895 is now matched using a separate opcode. However, there may be more than one
2896 byte in the character in UTF-8 mode.
2897
2898 2. The pcre_callout_block structure has two new fields: pattern_position and
2899 next_item_length. These contain the offset in the pattern to the next match
2900 item, and its length, respectively.
2901
2902 3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
2903 insertion of callouts before each pattern item. Added the /C option to
2904 pcretest to make use of this.
2905
2906 4. On the advice of a Windows user, the lines
2907
2908 #if defined(_WIN32) || defined(WIN32)
2909 _setmode( _fileno( stdout ), 0x8000 );
2910 #endif /* defined(_WIN32) || defined(WIN32) */
2911
2912 have been added to the source of pcretest. This apparently does useful
2913 magic in relation to line terminators.
2914
2915 5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
2916 for the benefit of those environments where the "b" makes a difference.
2917
2918 6. The icc compiler has the same options as gcc, but "configure" doesn't seem
2919 to know about it. I have put a hack into configure.in that adds in code
2920 to set GCC=yes if CC=icc. This seems to end up at a point in the
2921 generated configure script that is early enough to affect the setting of
2922 compiler options, which is what is needed, but I have no means of testing
2923 whether it really works. (The user who reported this had patched the
2924 generated configure script, which of course I cannot do.)
2925
2926 LATER: After change 22 below (new libtool files), the configure script
2927 seems to know about icc (and also ecc). Therefore, I have commented out
2928 this hack in configure.in.
2929
2930 7. Added support for pkg-config (2 patches were sent in).
2931
2932 8. Negated POSIX character classes that used a combination of internal tables
2933 were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
2934 [[:^ascii]]. Typically, they would match almost any characters. The other
2935 POSIX classes were not broken in this way.
2936
2937 9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
2938 to find the match, as PCRE was deluded into thinking that the match had to
2939 start at the start point or following a newline. The same bug applied to
2940 patterns with negative forward assertions or any backward assertions
2941 preceding ".*" at the start, unless the pattern required a fixed first
2942 character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
2943
2944 10. In UTF-8 mode, when moving forwards in the subject after a failed match
2945 starting at the last subject character, bytes beyond the end of the subject
2946 string were read.
2947
2948 11. Renamed the variable "class" as "classbits" to make life easier for C++
2949 users. (Previously there was a macro definition, but it apparently wasn't
2950 enough.)
2951
2952 12. Added the new field "tables" to the extra data so that tables can be passed
2953 in at exec time, or the internal tables can be re-selected. This allows
2954 a compiled regex to be saved and re-used at a later time by a different
2955 program that might have everything at different addresses.
2956
2957 13. Modified the pcre-config script so that, when run on Solaris, it shows a
2958 -R library as well as a -L library.
2959
2960 14. The debugging options of pcretest (-d on the command line or D on a
2961 pattern) showed incorrect output for anything following an extended class
2962 that contained multibyte characters and which was followed by a quantifier.
2963
2964 15. Added optional support for general category Unicode character properties
2965 via the \p, \P, and \X escapes. Unicode property support implies UTF-8
2966 support. It adds about 90K to the size of the library. The meanings of the
2967 inbuilt class escapes such as \d and \s have NOT been changed.
2968
2969 16. Updated pcredemo.c to include calls to free() to release the memory for the
2970 compiled pattern.
2971
2972 17. The generated file chartables.c was being created in the source directory
2973 instead of in the building directory. This caused the build to fail if the
2974 source directory was different from the building directory, and was
2975 read-only.
2976
2977 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
2978 file. No doubt somebody will tell me if they don't make sense... Also added
2979 Dan Mooney's comments about building on OpenVMS.
2980
2981 19. Added support for partial matching via the PCRE_PARTIAL option for
2982 pcre_exec() and the \P data escape in pcretest.
2983
2984 20. Extended pcretest with 3 new pattern features:
2985
2986 (i) A pattern option of the form ">rest-of-line" causes pcretest to
2987 write the compiled pattern to the file whose name is "rest-of-line".
2988 This is a straight binary dump of the data, with the saved pointer to
2989 the character tables forced to be NULL. The study data, if any, is
2990 written too. After writing, pcretest reads a new pattern.
2991
2992 (ii) If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
2993 compiled pattern from the given file. There must not be any
2994 occurrences of "<" in the file name (pretty unlikely); if there are,
2995 pcretest will instead treat the initial "<" as a pattern delimiter.
2996 After reading in the pattern, pcretest goes on to read data lines as
2997 usual.
2998
2999 (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
3000 and 16-bit fields in a compiled pattern, to simulate a pattern that
3001 was compiled on a host of opposite endianness.
3002
3003 21. The pcre-exec() function can now cope with patterns that were compiled on
3004 hosts of opposite endianness, with this restriction:
3005
3006 As for any compiled expression that is saved and used later, the tables
3007 pointer field cannot be preserved; the extra_data field in the arguments
3008 to pcre_exec() should be used to pass in a tables address if a value
3009 other than the default internal tables were used at compile time.
3010
3011 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
3012 now diagnosed as an error. Previously, most of the time, a negative number
3013 would have been treated as zero, but if in addition "ovector" was passed as
3014 NULL, a crash could occur.
3015
3016 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
3017 new versions from the libtool 1.5 distribution (the last one is a copy of
3018 a file called libtool.m4). This seems to have fixed the need to patch
3019 "configure" to support Darwin 1.3 (which I used to do). However, I still
3020 had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
3021 workstation).
3022
3023 24. Changed the PCRE licence to be the more standard "BSD" licence.
3024
3025
3026 Version 4.5 01-Dec-03
3027 ---------------------
3028
3029 1. There has been some re-arrangement of the code for the match() function so
3030 that it can be compiled in a version that does not call itself recursively.
3031 Instead, it keeps those local variables that need separate instances for
3032 each "recursion" in a frame on the heap, and gets/frees frames whenever it
3033 needs to "recurse". Keeping track of where control must go is done by means
3034 of setjmp/longjmp. The whole thing is implemented by a set of macros that
3035 hide most of the details from the main code, and operates only if
3036 NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
3037 "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
3038 operating.
3039
3040 To make it easier for callers to provide specially tailored get/free
3041 functions for this usage, two new functions, pcre_stack_malloc, and
3042 pcre_stack_free, are used. They are always called in strict stacking order,
3043 and the size of block requested is always the same.
3044
3045 The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
3046 PCRE has been compiled to use the stack or the heap for recursion. The
3047 -C option of pcretest uses this to show which version is compiled.
3048
3049 A new data escape \S, is added to pcretest; it causes the amounts of store
3050 obtained and freed by both kinds of malloc/free at match time to be added
3051 to the output.
3052
3053 2. Changed the locale test to use "fr_FR" instead of "fr" because that's
3054 what's available on my current Linux desktop machine.
3055
3056 3. When matching a UTF-8 string, the test for a valid string at the start has
3057 been extended. If start_offset is not zero, PCRE now checks that it points
3058 to a byte that is the start of a UTF-8 character. If not, it returns
3059 PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
3060 this is necessary because there may be backward assertions in the pattern.
3061 When matching the same subject several times, it may save resources to use
3062 PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
3063
3064 4. The code for checking the validity of UTF-8 strings has been tightened so
3065 that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
3066 containing "overlong sequences".
3067
3068 5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
3069 I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
3070 should have been "&", but it just so happened that all the cases this let
3071 through by mistake were picked up later in the function.
3072
3073 6. I had used a variable called "isblank" - this is a C99 function, causing
3074 some compilers to warn. To avoid this, I renamed it (as "blankclass").
3075
3076 7. Cosmetic: (a) only output another newline at the end of pcretest if it is
3077 prompting; (b) run "./pcretest /dev/null" at the start of the test script
3078 so the version is shown; (c) stop "make test" echoing "./RunTest".
3079
3080 8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
3081
3082 9. The prototype for memmove() for systems that don't have it was using
3083 size_t, but the inclusion of the header that defines size_t was later. I've
3084 moved the #includes for the C headers earlier to avoid this.
3085
3086 10. Added some adjustments to the code to make it easier to compiler on certain
3087 special systems:
3088
3089 (a) Some "const" qualifiers were missing.
3090 (b) Added the macro EXPORT before all exported functions; by default this
3091 is defined to be empty.
3092 (c) Changed the dftables auxiliary program (that builds chartables.c) so
3093 that it reads its output file name as an argument instead of writing
3094 to the standard output and assuming this can be redirected.
3095
3096 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
3097 class containing characters with values greater than 255, PCRE compilation
3098 went into a loop.
3099
3100 12. A recursive reference to a subpattern that was within another subpattern
3101 that had a minimum quantifier of zero caused PCRE to crash. For example,
3102 (x(y(?2))z)? provoked this bug with a subject that got as far as the
3103 recursion. If the recursively-called subpattern itself had a zero repeat,
3104 that was OK.
3105
3106 13. In pcretest, the buffer for reading a data line was set at 30K, but the
3107 buffer into which it was copied (for escape processing) was still set at
3108 1024, so long lines caused crashes.
3109
3110 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
3111 "internal error: code overflow...". This applied to any character class
3112 that was followed by a possessive quantifier.
3113
3114 15. Modified the Makefile to add libpcre.la as a prerequisite for
3115 libpcreposix.la because I was told this is needed for a parallel build to
3116 work.
3117
3118 16. If a pattern that contained .* following optional items at the start was
3119 studied, the wrong optimizing data was generated, leading to matching
3120 errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
3121 matching string must start with a or b or c. The correct conclusion for
3122 this pattern is that a match can start with any character.
3123
3124
3125 Version 4.4 13-Aug-03
3126 ---------------------
3127
3128 1. In UTF-8 mode, a character class containing characters with values between
3129 127 and 255 was not handled correctly if the compiled pattern was studied.
3130 In fixing this, I have also improved the studying algorithm for such
3131 classes (slightly).
3132
3133 2. Three internal functions had redundant arguments passed to them. Removal
3134 might give a very teeny performance improvement.
3135
3136 3. Documentation bug: the value of the capture_top field in a callout is *one
3137 more than* the number of the hightest numbered captured substring.
3138
3139 4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
3140 in incorrectly linking with a previously installed version. They now link
3141 explicitly with libpcre.la.
3142
3143 5. configure.in no longer needs to recognize Cygwin specially.
3144
3145 6. A problem in pcre.in for Windows platforms is fixed.
3146
3147 7. If a pattern was successfully studied, and the -d (or /D) flag was given to
3148 pcretest, it used to include the size of the study block as part of its
3149 output. Unfortunately, the structure contains a field that has a different
3150 size on different hardware architectures. This meant that the tests that
3151 showed this size failed. As the block is currently always of a fixed size,
3152 this information isn't actually particularly useful in pcretest output, so
3153 I have just removed it.
3154
3155 8. Three pre-processor statements accidentally did not start in column 1.
3156 Sadly, there are *still* compilers around that complain, even though
3157 standard C has not required this for well over a decade. Sigh.
3158
3159 9. In pcretest, the code for checking callouts passed small integers in the
3160 callout_data field, which is a void * field. However, some picky compilers
3161 complained about the casts involved for this on 64-bit systems. Now
3162 pcretest passes the address of the small integer instead, which should get
3163 rid of the warnings.
3164
3165 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
3166 both compile and run time, and gives an error if an invalid UTF-8 sequence
3167 is found. There is a option for disabling this check in cases where the
3168 string is known to be correct and/or the maximum performance is wanted.
3169
3170 11. In response to a bug report, I changed one line in Makefile.in from
3171
3172 -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
3173 to
3174 -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
3175
3176 to look similar to other lines, but I have no way of telling whether this
3177 is the right thing to do, as I do not use Windows. No doubt I'll get told
3178 if it's wrong...
3179
3180
3181 Version 4.3 21-May-03
3182 ---------------------
3183
3184 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
3185 Makefile.
3186
3187 2. Some refactoring to improve the quality of the code:
3188
3189 (i) The utf8_table... variables are now declared "const".
3190
3191 (ii) The code for \cx, which used the "case flipping" table to upper case
3192 lower case letters, now just substracts 32. This is ASCII-specific,
3193 but the whole concept of \cx is ASCII-specific, so it seems
3194 reasonable.
3195
3196 (iii) PCRE was using its character types table to recognize decimal and
3197 hexadecimal digits in the pattern. This is silly, because it handles
3198 only 0-9, a-f, and A-F, but the character types table is locale-
3199 specific, which means strange things might happen. A private
3200 table is now used for this - though it costs 256 bytes, a table is
3201 much faster than multiple explicit tests. Of course, the standard
3202 character types table is still used for matching digits in subject
3203 strings against \d.
3204
3205 (iv) Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
3206 ending in _t are). So I've renamed it as ESC_tee.
3207
3208 3. The first argument for regexec() in the POSIX wrapper should have been
3209 defined as "const".
3210
3211 4. Changed pcretest to use malloc() for its buffers so that they can be
3212 Electric Fenced for debugging.
3213
3214 5. There were several places in the code where, in UTF-8 mode, PCRE would try
3215 to read one or more bytes before the start of the subject string. Often this
3216 had no effect on PCRE's behaviour, but in some circumstances it could
3217 provoke a segmentation fault.
3218
3219 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
3220 to try to read one or more bytes before the start of the subject string.
3221
3222 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
3223 UTF-8 support could misbehave in various ways if the subject string
3224 contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
3225 area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
3226 back over UTF-8 characters.)
3227
3228
3229 Version 4.2 14-Apr-03
3230 ---------------------
3231
3232 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
3233
3234 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
3235 [ON_WINDOWS]: new variable, "#" on non-Windows platforms
3236 [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
3237 [WIN_PREFIX]: new variable, "cyg" for Cygwin
3238 * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
3239 and BUILD_EXEEXT
3240 Note: automatic setting of the BUILD variables is not yet working
3241 set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
3242 compile-time but not at link-time
3243 [LINK]: use for linking executables only
3244 make different versions for Windows and non-Windows
3245 [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
3246 libraries
3247 [LINK_FOR_BUILD]: new variable
3248 [OBJEXT]: use throughout
3249 [EXEEXT]: use throughout
3250 <winshared>: new target
3251 <wininstall>: new target
3252 <dftables.o>: use native compiler
3253 <dftables>: use native linker
3254 <install>: handle Windows platform correctly
3255 <clean>: ditto
3256 <check>: ditto
3257 copy DLL to top builddir before testing
3258
3259 As part of these changes, -no-undefined was removed again. This was reported
3260 to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
3261 in any case.
3262
3263 3. Some tidies to get rid of compiler warnings:
3264
3265 . In the match_data structure, match_limit was an unsigned long int, whereas
3266 match_call_count was an int. I've made them both unsigned long ints.
3267
3268 . In pcretest the fact that a const uschar * doesn't automatically cast to
3269 a void * provoked a warning.
3270
3271 . Turning on some more compiler warnings threw up some "shadow" variables
3272 and a few more missing casts.
3273
3274 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
3275 option, a class that contained a single character with a value between 128
3276 and 255 (e.g. /[\xFF]/) caused PCRE to crash.
3277
3278 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
3279 option, a class that contained several characters, but with at least one
3280 whose value was between 128 and 255 caused PCRE to crash.
3281
3282
3283 Version 4.1 12-Mar-03
3284 ---------------------
3285
3286 1. Compiling with gcc -pedantic found a couple of places where casts were
3287 needed, and a string in dftables.c that was longer than standard compilers are
3288 required to support.
3289
3290 2. Compiling with Sun's compiler found a few more places where the code could
3291 be tidied up in order to avoid warnings.
3292
3293 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
3294 first of these names is deprecated in the latest Autoconf in favour of the name
3295 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
3296 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
3297 analogy I have changed it to CFLAGS_FOR_BUILD.
3298
3299 4. Added -no-undefined to the linking command in the Makefile, because this is
3300 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
3301 linking step for the pcreposix library.
3302
3303 5. PCRE was failing to diagnose the case of two named groups with the same
3304 name.
3305
3306 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
3307 literal character that is needed in the subject for a match, and scans along to
3308 ensure that it is present before embarking on the full matching process. This
3309 saves time in cases of nested unlimited repeats that are never going to match.
3310 Problem: the scan can take a lot of time if the subject is very long (e.g.
3311 megabytes), thus penalizing straightforward matches. It is now done only if the
3312 amount of subject to be scanned is less than 1000 bytes.
3313
3314 7. A lesser problem with the same optimization is that it was recording the
3315 first character of an anchored pattern as "needed", thus provoking a search
3316 right along the subject, even when the first match of the pattern was going to
3317 fail. The "needed" character is now not set for anchored patterns, unless it
3318 follows something in the pattern that is of non-fixed length. Thus, it still
3319 fulfils its original purpose of finding quick non-matches in cases of nested
3320 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
3321
3322
3323 Version 4.0 17-Feb-03
3324 ---------------------
3325
3326 1. If a comment in an extended regex that started immediately after a meta-item
3327 extended to the end of string, PCRE compiled incorrect data. This could lead to
3328 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
3329
3330 2. Moved to autoconf 2.53 and libtool 1.4.2.
3331
3332 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
3333 the special perltest8 script is no longer needed - all the tests can be run
3334 from a single perltest script.
3335
3336 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
3337 by \s. It has now been removed in PCRE. This means it isn't recognized as
3338 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
3339 class [:space:] *does* include VT, thereby creating a mess.
3340
3341 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
3342 space and tab.
3343
3344 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
3345 its new features into the main test script, reducing the number of scripts.
3346
3347 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
3348 were backward compatible, and made the (?i) apply to the whole pattern, as if
3349 /i were given. Now it behaves more logically, and applies the option setting
3350 only to what follows. PCRE has been changed to follow suit. However, if it
3351 finds options settings right at the start of the pattern, it extracts them into
3352 the global options, as before. Thus, they show up in the info data.
3353
3354 8. Added support for the \Q...\E escape sequence. Characters in between are
3355 treated as literals. This is slightly different from Perl in that $ and @ are
3356 also handled as literals inside the quotes. In Perl, they will cause variable
3357 interpolation. Note the following examples:
3358
3359 Pattern PCRE matches Perl matches
3360
3361 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
3362 \Qabc\$xyz\E abc\$xyz abc\$xyz
3363 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
3364
3365 For compatibility with Perl, \Q...\E sequences are recognized inside character
3366 classes as well as outside them.
3367
3368 9. Re-organized 3 code statements in pcretest to avoid "overflow in
3369 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
3370 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
3371 signed/unsigned warnings.
3372
3373 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
3374 option for pcretest, so I've replaced it by a simple function that does just
3375 that job.
3376
3377 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
3378 "pcregrep -".
3379
3380 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
3381 Java package. This provides some syntactic sugar for simple cases of what my
3382 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
3383 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
3384 item, you can use this simplified notation. Note that only makes sense with
3385 greedy quantifiers. Consequently, the use of the possessive quantifier forces
3386 greediness, whatever the setting of the PCRE_UNGREEDY option.
3387
3388 13. A change of greediness default within a pattern was not taking effect at
3389 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
3390 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
3391 was abstracted outside.
3392
3393 14. PCRE now supports the \G assertion. It is true when the current matching
3394 position is at the start point of the match. This differs from \A when the
3395 starting offset is non-zero. Used with the /g option of pcretest (or similar
3396 code), it works in the same way as it does for Perl's /g option. If all
3397 alternatives of a regex begin with \G, the expression is anchored to the start
3398 match position, and the "anchored" flag is set in the compiled expression.
3399
3400 15. Some bugs concerning the handling of certain option changes within patterns
3401 have been fixed. These applied to options other than (?ims). For example,
3402 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
3403 been the other way round. Some of this was related to change 7 above.
3404
3405 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
3406 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
3407 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
3408 POSIX classes only within a class (e.g. /[[:alpha:]]/).
3409
3410 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
3411 mode. Unlike ".", it always matches newline, whatever the setting of
3412 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
3413 assertions. Perl allows it, but it doesn't (in general) work because it can't
3414 calculate the length of the lookbehind. At least, that's the case for Perl
3415 5.8.0 - I've been told they are going to document that it doesn't work in
3416 future.
3417
3418 18. Added an error diagnosis for escapes that PCRE does not support: these are
3419 \L, \l, \N, \P, \p, \U, \u, and \X.
3420
3421 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
3422 reading past the end of the pattern in cases such as /[abcd/.
3423
3424 20. PCRE was getting more memory than necessary for patterns with classes that
3425 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
3426
3427 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
3428 compiling PCRE for use with Virtual Pascal.
3429
3430 22. Small fix to the Makefile to make it work properly if the build is done
3431 outside the source tree.
3432
3433 23. Added a new extension: a condition to go with recursion. If a conditional
3434 subpattern starts with (?(R) the "true" branch is used if recursion has
3435 happened, whereas the "false" branch is used only at the top level.
3436
3437 24. When there was a very long string of literal characters (over 255 bytes
3438 without UTF support, over 250 bytes with UTF support), the computation of how
3439 much memory was required could be incorrect, leading to segfaults or other
3440 strange effects.
3441
3442 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
3443 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
3444 there was a subsequent back reference to those brackets. This meant that, for
3445 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
3446 possible to check for precisely this case. All we can do is abandon the
3447 optimization if .* occurs inside capturing brackets when there are any back
3448 references whatsoever. (See below for a better fix that came later.)
3449
3450 26. The handling of the optimization for finding the first character of a
3451 non-anchored pattern, and for finding a character that is required later in the
3452 match were failing in some cases. This didn't break the matching; it just
3453 failed to optimize when it could. The way this is done has been re-implemented.
3454
3455 27. Fixed typo in error message for invalid (?R item (it said "(?p").
3456
3457 28. Added a new feature that provides some of the functionality that Perl
3458 provides with (?{...}). The facility is termed a "callout". The way it is done
3459 in PCRE is for the caller to provide an optional function, by setting
3460 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
3461 global variable. By default it is unset, which disables all calling out. To get
3462 the function called, the regex must include (?C) at appropriate points. This
3463 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
3464 This provides a means of identifying different callout points. When PCRE
3465 reaches such a point in the regex, if pcre_callout has been set, the external
3466 function is called. It is provided with data in a structure called
3467 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
3468 matching continues; if it returns a non-zero value, the match at the current
3469 point fails. However, backtracking will occur if possible. [This was changed
3470 later and other features added - see item 49 below.]
3471
3472 29. pcretest is upgraded to test the callout functionality. It provides a
3473 callout function that displays information. By default, it shows the start of
3474 the match and the current position in the text. There are some new data escapes
3475 to vary what happens:
3476
3477 \C+ in addition, show current contents of captured substrings
3478 \C- do not supply a callout function
3479 \C!n return 1 when callout number n is reached
3480 \C!n!m return 1 when callout number n is reached for the mth time
3481
3482 30. If pcregrep was called with the -l option and just a single file name, it
3483 output "<stdin>" if a match was found, instead of the file name.
3484
3485 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
3486 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
3487 pcre_exec(). This saves a malloc/free per call. The default value of
3488 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
3489 when configuring.
3490
3491 32. The default maximum size of a compiled pattern is 64K. There have been a
3492 few cases of people hitting this limit. The code now uses macros to handle the
3493 storing of links as offsets within the compiled pattern. It defaults to 2-byte
3494 links, but this can be changed to 3 or 4 bytes by --with-link-size when
3495 configuring. Tests 2 and 5 work only with 2-byte links because they output
3496 debugging information about compiled patterns.
3497
3498 33. Internal code re-arrangements:
3499
3500 (a) Moved the debugging function for printing out a compiled regex into
3501 its own source file (printint.c) and used #include to pull it into
3502 pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
3503 separate copies.
3504
3505 (b) Defined the list of op-code names for debugging as a macro in
3506 internal.h so that it is next to the definition of the opcodes.
3507
3508 (c) Defined a table of op-code lengths for simpler skipping along compiled
3509 code. This is again a macro in internal.h so that it is next to the
3510 definition of the opcodes.
3511
3512 34. Added support for recursive calls to individual subpatterns, along the
3513 lines of Robin Houston's patch (but implemented somewhat differently).
3514
3515 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
3516 allow it to read and process whole directories in Win32. This code was
3517 contributed by Lionel Fourquaux; it has not been tested by me.
3518
3519 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
3520 used to name a group. Names consist of alphanumerics and underscores, and must
3521 be unique. Back references use the syntax (?P=name) and recursive calls use
3522 (?P>name) which is a PCRE extension to the Python extension. Groups still have
3523 numbers. The function pcre_fullinfo() can be used after compilation to extract
3524 a name/number map. There are three relevant calls:
3525
3526 PCRE_INFO_NAMEENTRYSIZE yields the size of each entry in the map
3527 PCRE_INFO_NAMECOUNT yields the number of entries
3528 PCRE_INFO_NAMETABLE yields a pointer to the map.
3529
3530 The map is a vector of fixed-size entries. The size of each entry depends on
3531 the length of the longest name used. The first two bytes of each entry are the
3532 group number, most significant byte first. There follows the corresponding
3533 name, zero terminated. The names are in alphabetical order.
3534
3535 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
3536 case instead of 255. Making it the same both with and without UTF-8 support
3537 means that the same test output works with both.
3538
3539 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
3540 calling malloc() with a zero argument.
3541
3542 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
3543 optimization. I've improved things by keeping a bitmap of backreferences with
3544 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
3545 fact referenced, the optimization can be applied. It is unlikely that a
3546 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
3547 the match to follow \n) will appear inside brackets with a number greater than
3548 31, but if it does, any back reference > 31 suppresses the optimization.
3549
3550 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
3551 of disabling numbered capturing parentheses. Any opening parenthesis that is
3552 not followed by ? behaves as if it were followed by ?: but named parentheses
3553 can still be used for capturing (and they will acquire numbers in the usual
3554 way).
3555
3556 41. Redesigned the return codes from the match() function into yes/no/error so
3557 that errors can be passed back from deep inside the nested calls. A malloc
3558 failure while inside a recursive subpattern call now causes the
3559 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
3560
3561 42. It is now possible to set a limit on the number of times the match()
3562 function is called in a call to pcre_exec(). This facility makes it possible to
3563 limit the amount of recursion and backtracking, though not in a directly
3564 obvious way, because the match() function is used in a number of different
3565 circumstances. The count starts from zero for each position in the subject
3566 string (for non-anchored patterns). The default limit is, for compatibility, a
3567 large number, namely 10 000 000. You can change this in two ways:
3568
3569 (a) When configuring PCRE before making, you can use --with-match-limit=n
3570 to set a default value for the compiled library.
3571
3572 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
3573 a different value is set. See 45 below.
3574
3575 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
3576
3577 43. Added a new function pcre_config(int, void *) to enable run-time extraction
3578 of things that can be changed at compile time. The first argument specifies
3579 what is wanted and the second points to where the information is to be placed.
3580 The current list of available information is:
3581
3582 PCRE_CONFIG_UTF8
3583
3584 The output is an integer that is set to one if UTF-8 support is available;
3585 otherwise it is set to zero.
3586
3587 PCRE_CONFIG_NEWLINE
3588
3589 The output is an integer that it set to the value of the code that is used for
3590 newline. It is either LF (10) or CR (13).
3591
3592 PCRE_CONFIG_LINK_SIZE
3593
3594 The output is an integer that contains the number of bytes used for internal
3595 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
3596
3597 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
3598
3599 The output is an integer that contains the threshold above which the POSIX
3600 interface uses malloc() for output vectors. See item 31 above.
3601
3602 PCRE_CONFIG_MATCH_LIMIT
3603
3604 The output is an unsigned integer that contains the default limit of the number
3605 of match() calls in a pcre_exec() execution. See 42 above.
3606
3607 44. pcretest has been upgraded by the addition of the -C option. This causes it
3608 to extract all the available output from the new pcre_config() function, and to
3609 output it. The program then exits immediately.
3610
3611 45. A need has arisen to pass over additional data with calls to pcre_exec() in
3612 order to support additional features. One way would have been to define
3613 pcre_exec2() (for example) with extra arguments, but this would not have been
3614 extensible, and would also have required all calls to the original function to
3615 be mapped to the new one. Instead, I have chosen to extend the mechanism that
3616 is used for passing in "extra" data from pcre_study().
3617
3618 The pcre_extra structure is now exposed and defined in pcre.h. It currently
3619 contains the following fields:
3620
3621 flags a bitmap indicating which of the following fields are set
3622 study_data opaque data from pcre_study()
3623 match_limit a way of specifying a limit on match() calls for a specific
3624 call to pcre_exec()
3625 callout_data data for callouts (see 49 below)
3626
3627 The flag bits are also defined in pcre.h, and are
3628
3629 PCRE_EXTRA_STUDY_DATA
3630 PCRE_EXTRA_MATCH_LIMIT
3631 PCRE_EXTRA_CALLOUT_DATA
3632
3633 The pcre_study() function now returns one of these new pcre_extra blocks, with
3634 the actual study data pointed to by the study_data field, and the
3635 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
3636 before. That is, this change is entirely upwards-compatible and requires no
3637 change to existing code.
3638
3639 If you want to pass in additional data to pcre_exec(), you can either place it
3640 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
3641 block.
3642
3643 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
3644 data string contains the escape sequence \M, pcretest calls pcre_exec() several
3645 times with different match limits, until it finds the minimum value needed for
3646 pcre_exec() to complete. The value is then output. This can be instructive; for
3647 most simple matches the number is quite small, but for pathological cases it
3648 gets very large very quickly.
3649
3650 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
3651 returns the size of the data block pointed to by the study_data field in a
3652 pcre_extra block, that is, the value that was passed as the argument to
3653 pcre_malloc() when PCRE was getting memory in which to place the information
3654 created by pcre_study(). The fourth argument should point to a size_t variable.
3655 pcretest has been extended so that this information is shown after a successful
3656 pcre_study() call when information about the compiled regex is being displayed.
3657
3658 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
3659 because what follows is always an absolute path. (Later: it turns out that this
3660 is more than cosmetic for MinGW, because it doesn't like empty path
3661 components.)
3662
3663 49. Some changes have been made to the callout feature (see 28 above):
3664
3665 (i) A callout function now has three choices for what it returns:
3666
3667 0 => success, carry on matching
3668 > 0 => failure at this point, but backtrack if possible
3669 < 0 => serious error, return this value from pcre_exec()
3670
3671 Negative values should normally be chosen from the set of PCRE_ERROR_xxx
3672 values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
3673 "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
3674 use by callout functions. It will never be used by PCRE itself.
3675
3676 (ii) The pcre_extra structure (see 45 above) has a void * field called
3677 callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
3678 pcre_callout_block structure has a field of the same name. The contents of
3679 the field passed in the pcre_extra structure are passed to the callout
3680 function in the corresponding field in the callout block. This makes it
3681 easier to use the same callout-containing regex from multiple threads. For
3682 testing, the pcretest program has a new data escape
3683
3684 \C*n pass the number n (may be negative) as callout_data
3685
3686 If the callout function in pcretest receives a non-zero value as
3687 callout_data, it returns that value.
3688
3689 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
3690 there were some redundant $(CFLAGS) in commands that are now specified as
3691 $(LINK), which already includes $(CFLAGS).
3692
3693 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
3694 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
3695 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
3696 one-byte characters throughout. Note that case-insensitive matching applies
3697 only to characters whose values are less than 256. PCRE doesn't support the
3698 notion of cases for higher-valued characters.
3699
3700 (i) A character class whose characters are all within 0-255 is handled as
3701 a bit map, and the map is inverted for negative classes. Previously, a
3702 character > 255 always failed to match such a class; however it should
3703 match if the class was a negative one (e.g. [^ab]). This has been fixed.
3704
3705 (ii) A negated character class with a single character < 255 is coded as
3706 "not this character" (OP_NOT). This wasn't working properly when the test
3707 character was multibyte, either singly or repeated.
3708
3709 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
3710 mode, for example: \x{100}{2,3}.
3711
3712 (iv) The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
3713 singly or repeated) now correctly test multibyte characters. However,
3714 PCRE doesn't recognize any characters with values greater than 255 as
3715 digits, spaces, or word characters. Such characters always match \D, \S,
3716 and \W, and never match \d, \s, or \w.
3717
3718 (v) Classes may now contain characters and character ranges with values
3719 greater than 255. For example: [ab\x{100}-\x{400}].
3720
3721 (vi) pcregrep now has a --utf-8 option (synonym -u) which makes it call
3722 PCRE in UTF-8 mode.
3723
3724 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
3725 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
3726 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
3727 value.)
3728
3729 53. The single man page has become too large. I have therefore split it up into
3730 a number of separate man pages. These also give rise to individual HTML pages;
3731 these are now put in a separate directory, and there is an index.html page that
3732 lists them all. Some hyperlinking between the pages has been installed.
3733
3734 54. Added convenience functions for handling named capturing parentheses.
3735
3736 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
3737 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
3738 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
3739 are faulted.
3740
3741 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
3742 calling configure. These values are used when compiling the dftables.c program
3743 which is run to generate the source of the default character tables. They
3744 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
3745 you will need to set these values.
3746
3747 57. Updated the building process for Windows DLL, as provided by Fred Cox.
3748
3749
3750 Version 3.9 02-Jan-02
3751 ---------------------
3752
3753 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
3754
3755 2. If --disable-static was given, the building process failed when trying to
3756 build pcretest and pcregrep. (For some reason it was using libtool to compile
3757 them, which is not right, as they aren't part of the library.)
3758
3759
3760 Version 3.8 18-Dec-01
3761 ---------------------
3762
3763 1. The experimental UTF-8 code was completely screwed up. It was packing the
3764 bytes in the wrong order. How dumb can you get?
3765
3766
3767 Version 3.7 29-Oct-01
3768 ---------------------
3769
3770 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
3771 This caused pcretest, when used on the test data, to segfault. Unfortunately,
3772 this didn't happen under Solaris 8, where I normally test things.
3773
3774 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
3775 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
3776 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
3777 this fix an hour or so after the initial 3.7 release.)
3778
3779
3780 Version 3.6 23-Oct-01
3781 ---------------------
3782
3783 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
3784 offsets passed as NULL with zero offset count.
3785
3786 2. The config.guess and config.sub files had not been updated when I moved to
3787 the latest autoconf.
3788
3789
3790 Version 3.5 15-Aug-01
3791 ---------------------
3792
3793 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
3794 had been forgotten.
3795
3796 2. By using declared but undefined structures, we can avoid using "void"
3797 definitions in pcre.h while keeping the internal definitions of the structures
3798 private.
3799
3800 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
3801 user point of view, this means that both static and shared libraries are built
3802 by default, but this can be individually controlled. More of the work of
3803 handling this static/shared cases is now inside libtool instead of PCRE's make
3804 file.
3805
3806 4. The pcretest utility is now installed along with pcregrep because it is
3807 useful for users (to test regexs) and by doing this, it automatically gets
3808 relinked by libtool. The documentation has been turned into a man page, so
3809 there are now .1, .txt, and .html versions in /doc.
3810
3811 5. Upgrades to pcregrep:
3812 (i) Added long-form option names like gnu grep.
3813 (ii) Added --help to list all options with an explanatory phrase.
3814 (iii) Added -r, --recursive to recurse into sub-directories.
3815 (iv) Added -f, --file to read patterns from a file.
3816
3817 6. pcre_exec() was referring to its "code" argument before testing that
3818 argument for NULL (and giving an error if it was NULL).
3819
3820 7. Upgraded Makefile.in to allow for compiling in a different directory from
3821 the source directory.
3822
3823 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
3824 options bits, the pointer it was passed was to an int instead of to an unsigned
3825 long int. This mattered only on 64-bit systems.
3826
3827 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
3828 generated) instead of pcre.in, which it its source. Also made the same change
3829 in several of the .c files.
3830
3831 10. A new release of gcc defines printf() as a macro, which broke pcretest
3832 because it had an ifdef in the middle of a string argument for printf(). Fixed
3833 by using separate calls to printf().
3834
3835 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
3836 script, to force use of CR or LF instead of \n in the source. On non-Unix
3837 systems, the value can be set in config.h.
3838
3839 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
3840 absolute limit. Changed the text of the error message to make this clear, and
3841 likewise updated the man page.
3842
3843 13. The limit of 99 on the number of capturing subpatterns has been removed.
3844 The new limit is 65535, which I hope will not be a "real" limit.
3845
3846
3847 Version 3.4 22-Aug-00
3848 ---------------------
3849
3850 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
3851
3852 2. Diagnose condition (?(0) as an error instead of crashing on matching.
3853
3854
3855 Version 3.3 01-Aug-00
3856 ---------------------
3857
3858 1. If an octal character was given, but the value was greater than \377, it
3859 was not getting masked to the least significant bits, as documented. This could
3860 lead to crashes in some systems.
3861
3862 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
3863 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
3864
3865 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
3866 These just pass their arguments on to (pcre_free)(), but they are provided
3867 because some uses of PCRE bind it to non-C systems that can call its functions,
3868 but cannot call free() or pcre_free() directly.
3869
3870 4. Add "make test" as a synonym for "make check". Corrected some comments in
3871 the Makefile.
3872
3873 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
3874 Makefile.
3875
3876 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
3877 command called pgrep for grepping around the active processes.
3878
3879 7. Added the beginnings of support for UTF-8 character strings.
3880
3881 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
3882 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
3883 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
3884 out for the ar command.)
3885
3886
3887 Version 3.2 12-May-00
3888 ---------------------
3889
3890 This is purely a bug fixing release.
3891
3892 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
3893 of ZA. This was just one example of several cases that could provoke this bug,
3894 which was introduced by change 9 of version 2.00. The code for breaking
3895 infinite loops after an iteration that matches an empty string was't working
3896 correctly.
3897
3898 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
3899 when matched against abbab (for example). After matching an empty string, it
3900 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
3901 caused it to match further down the string than it should.
3902
3903 3. The code contained an inclusion of sys/types.h. It isn't clear why this
3904 was there because it doesn't seem to be needed, and it causes trouble on some
3905 systems, as it is not a Standard C header. It has been removed.
3906
3907 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
3908 were reported on the Macintosh. The changes were from
3909
3910 while ((c = *(++ptr)) != 0 && c != '\n');
3911 to
3912 while ((c = *(++ptr)) != 0 && c != '\n') ;
3913
3914 Totally extraordinary, but if that's what it takes...
3915
3916 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
3917 available. Added HAVE_BCOPY and an autoconf test for it; if neither
3918 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
3919 assumes the way PCRE uses memmove() (always moving upwards).
3920
3921 6. PCRE is being used in one environment where strchr() is not available. There
3922 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
3923 faster code anyway.
3924
3925
3926 Version 3.1 09-Feb-00
3927 ---------------------
3928
3929 The only change in this release is the fixing of some bugs in Makefile.in for
3930 the "install" target:
3931
3932 (1) It was failing to install pcreposix.h.
3933
3934 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
3935
3936
3937 Version 3.0 01-Feb-00
3938 ---------------------
3939
3940 1. Add support for the /+ modifier to perltest (to output $` like it does in
3941 pcretest).
3942
3943 2. Add support for the /g modifier to perltest.
3944
3945 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
3946 matches null strings.
3947
3948 4. Fix perltest so that it doesn't do unwanted things when fed an empty
3949 pattern. Perl treats empty patterns specially - it reuses the most recent
3950 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
3951 effect.
3952
3953 5. The POSIX interface was broken in that it was just handing over the POSIX
3954 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
3955 required a bigger vector, with some working space on the end. This means that
3956 the POSIX wrapper now has to get and free some memory, and copy the results.
3957
3958 6. Added some simple autoconf support, placing the test data and the
3959 documentation in separate directories, re-organizing some of the
3960 information files, and making it build pcre-config (a GNU standard). Also added
3961 libtool support for building PCRE as a shared library, which is now the
3962 default.
3963
3964 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
3965 09 are not valid octal constants. Single digits will be used for minor values
3966 less than 10.
3967
3968 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
3969 existing programs that set these in the POSIX interface can use PCRE without
3970 modification.
3971
3972 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
3973 return all that pcre_info() returns, plus additional data. The pcre_info()
3974 function is retained for compatibility, but is considered to be obsolete.
3975
3976 10. Added experimental recursion feature (?R) to handle one common case that
3977 Perl 5.6 will be able to do with (?p{...}).
3978
3979 11. Added support for POSIX character classes like [:alpha:], which Perl is
3980 adopting.
3981
3982
3983 Version 2.08 31-Aug-99
3984 ----------------------
3985
3986 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
3987 trying to match at the startoffset position, but instead was moving forward to
3988 the next newline as if a previous match had failed.
3989
3990 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
3991 and could get into a loop if a null string was matched other than at the start
3992 of the subject.
3993
3994 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
3995 be distinguished at compile time, and for completeness also added PCRE_DATE.
3996
3997 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
3998 in GnuWin32 environments.
3999
4000
4001 Version 2.07 29-Jul-99
4002 ----------------------
4003
4004 1. The documentation is now supplied in plain text form and HTML as well as in
4005 the form of man page sources.
4006
4007 2. C++ compilers don't like assigning (void *) values to other pointer types.
4008 In particular this affects malloc(). Although there is no problem in Standard
4009 C, I've put in casts to keep C++ compilers happy.
4010
4011 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
4012 should be (const char *).
4013
4014 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
4015 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
4016 However, I haven't made this a standard facility. The documentation doesn't
4017 mention it, and the Makefile doesn't support it.
4018
4019 5. The Makefile now contains an "install" target, with editable destinations at
4020 the top of the file. The pcretest program is not installed.
4021
4022 6. pgrep -V now gives the PCRE version number and date.
4023
4024 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
4025 causing the entire string to be ignored, instead of just the last character.
4026
4027 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
4028 non-matching string, it can take a very, very long time, even for strings of
4029 quite modest length, because of the nested recursion. PCRE now does better in
4030 some of these cases. It does this by remembering the last required literal
4031 character in the pattern, and pre-searching the subject to ensure it is present
4032 before running the real match. In other words, it applies a heuristic to detect
4033 some types of certain failure quickly, and in the above example, if presented
4034 with a string that has no trailing " it gives "no match" very quickly.
4035
4036 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
4037 other alternatives are tried instead.
4038
4039
4040 Version 2.06 09-Jun-99
4041 ----------------------
4042
4043 1. Change pcretest's output for amount of store used to show just the code
4044 space, because the remainder (the data block) varies in size between 32-bit and
4045 64-bit systems.
4046
4047 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
4048 start matching at. This allows lookbehinds to work when searching for multiple
4049 occurrences in a string.
4050
4051 3. Added additional options to pcretest for testing multiple occurrences:
4052
4053 /+ outputs the rest of the string that follows a match
4054 /g loops for multiple occurrences, using the new startoffset argument
4055 /G loops for multiple occurrences by passing an incremented pointer
4056
4057 4. PCRE wasn't doing the "first character" optimization for patterns starting
4058 with \b or \B, though it was doing it for other lookbehind assertions. That is,
4059 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
4060 the letter 'x'. On long subject strings, this gives a significant speed-up.
4061
4062
4063 Version 2.05 21-Apr-99
4064 ----------------------
4065
4066 1. Changed the type of magic_number from int to long int so that it works
4067 properly on 16-bit systems.
4068
4069 2. Fixed a bug which caused patterns starting with .* not to work correctly
4070 when the subject string contained newline characters. PCRE was assuming
4071 anchoring for such patterns in all cases, which is not correct because .* will
4072 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
4073 DOTALL is set at top level; otherwise it knows that patterns starting with .*
4074 must be retried after every newline in the subject.
4075
4076
4077 Version 2.04 18-Feb-99
4078 ----------------------
4079
4080 1. For parenthesized subpatterns with repeats whose minimum was zero, the
4081 computation of the store needed to hold the pattern was incorrect (too large).
4082 If such patterns were nested a few deep, this could multiply and become a real
4083 problem.
4084
4085 2. Added /M option to pcretest to show the memory requirement of a specific
4086 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
4087
4088 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
4089 compiled in such a way that the backtracking after subsequent failure was
4090 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
4091 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
4092
4093
4094 Version 2.03 02-Feb-99
4095 ----------------------
4096
4097 1. Fixed typo and small mistake in man page.
4098
4099 2. Added 4th condition (GPL supersedes if conflict) and created separate
4100 LICENCE file containing the conditions.
4101
4102 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
4103 Perl, that is the internal \ allows the delimiter to be included in the
4104 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
4105 the final delimiter, add \ to the end of the pattern (to test the error).
4106
4107 4. Added the convenience functions for extracting substrings after a successful
4108 match. Updated pcretest to make it able to test these functions.
4109
4110
4111 Version 2.02 14-Jan-99
4112 ----------------------
4113
4114 1. Initialized the working variables associated with each extraction so that
4115 their saving and restoring doesn't refer to uninitialized store.
4116
4117 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
4118 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
4119 fix the problem.
4120
4121 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
4122 calls, and wasn't printing the correct value for compiling calls. Increased the
4123 default value of LOOPREPEAT, and the number of significant figures in the
4124 times.
4125
4126 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
4127
4128 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
4129 a building problem on Windows NT with a FAT file system.
4130
4131
4132 Version 2.01 21-Oct-98
4133 ----------------------
4134
4135 1. Changed the API for pcre_compile() to allow for the provision of a pointer
4136 to character tables built by pcre_maketables() in the current locale. If NULL
4137 is passed, the default tables are used.
4138
4139
4140 Version 2.00 24-Sep-98
4141 ----------------------
4142
4143 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
4144 it any more.
4145
4146 2. Allow quantification of (?>) groups, and make it work correctly.
4147
4148 3. The first character computation wasn't working for (?>) groups.
4149
4150 4. Correct the implementation of \Z (it is permitted to match on the \n at the
4151 end of the subject) and add 5.005's \z, which really does match only at the
4152 very end of the subject.
4153
4154 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
4155
4156 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
4157 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
4158 localized options. All options to pcre_study() were also removed.
4159
4160 7. Add other new features from 5.005:
4161
4162 $(?<= positive lookbehind
4163 $(?<! negative lookbehind
4164 (?imsx-imsx) added the unsetting capability
4165 such a setting is global if at outer level; local otherwise
4166 (?imsx-imsx:) non-capturing groups with option setting
4167 (?(cond)re|re) conditional pattern matching
4168
4169 A backreference to itself in a repeated group matches the previous
4170 captured string.
4171
4172 8. General tidying up of studying (both automatic and via "study")
4173 consequential on the addition of new assertions.
4174
4175 9. As in 5.005, unlimited repeated groups that could match an empty substring
4176 are no longer faulted at compile time. Instead, the loop is forcibly broken at
4177 runtime if any iteration does actually match an empty substring.
4178
4179 10. Include the RunTest script in the distribution.
4180
4181 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
4182 discrepancies, some of which were old and were also with respect to 5.004. They
4183 have now been fixed.
4184
4185
4186 Version 1.09 28-Apr-98
4187 ----------------------
4188
4189 1. A negated single character class followed by a quantifier with a minimum
4190 value of one (e.g. [^x]{1,6} ) was not compiled correctly. This could lead to
4191 program crashes, or just wrong answers. This did not apply to negated classes
4192 containing more than one character, or to minima other than one.
4193
4194
4195 Version 1.08 27-Mar-98
4196 ----------------------
4197
4198 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
4199
4200 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
4201 latter must appear before anything that relies on it in the pattern.
4202
4203
4204 Version 1.07 16-Feb-98
4205 ----------------------
4206
4207 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
4208 repeat of a potentially empty string).
4209
4210
4211 Version 1.06 23-Jan-98
4212 ----------------------
4213
4214 1. Added Markus Oberhumer's little patches for C++.
4215
4216 2. Literal strings longer than 255 characters were broken.
4217
4218
4219 Version 1.05 23-Dec-97
4220 ----------------------
4221
4222 1. Negated character classes containing more than one character were failing if
4223 PCRE_CASELESS was set at run time.
4224
4225
4226 Version 1.04 19-Dec-97
4227 ----------------------
4228
4229 1. Corrected the man page, where some "const" qualifiers had been omitted.
4230
4231 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
4232 input syntax.
4233
4234 3. Fixed memory leak which occurred when a regex with back references was
4235 matched with an offsets vector that wasn't big enough. The temporary memory
4236 that is used in this case wasn't being freed if the match failed.
4237
4238 4. Tidied pcretest to ensure it frees memory that it gets.
4239
4240 5. Temporary memory was being obtained in the case where the passed offsets
4241 vector was exactly big enough.
4242
4243 6. Corrected definition of offsetof() from change 5 below.
4244
4245 7. I had screwed up change 6 below and broken the rules for the use of
4246 setjmp(). Now fixed.
4247
4248
4249 Version 1.03 18-Dec-97
4250 ----------------------
4251
4252 1. A erroneous regex with a missing opening parenthesis was correctly
4253 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
4254 on some systems.
4255
4256 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
4257 it was reported that one broken compiler failed on the former because "code" is
4258 also an independent variable.
4259
4260 3. The erroneous regex a[]b caused an array overrun reference.
4261
4262 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
4263 fail on data ending with that character. (It was going on too far, and checking
4264 the next character, typically a binary zero.) This was specific to the
4265 optimized code for single-character negative classes.
4266
4267 5. Added a contributed patch from the TIN world which does the following:
4268
4269 + Add an undef for memmove, in case the the system defines a macro for it.
4270
4271 + Add a definition of offsetof(), in case there isn't one. (I don't know
4272 the reason behind this - offsetof() is part of the ANSI standard - but
4273 it does no harm).
4274
4275 + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
4276 most of the places where whitespace preceded '#'. I have given up and
4277 allowed the remaining 2 cases to be at the margin.
4278
4279 + Rename some variables in pcre to eliminate shadowing. This seems very
4280 pedantic, but does no harm, of course.
4281
4282 6. Moved the call to setjmp() into its own function, to get rid of warnings
4283 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
4284
4285 7. Constructs such as \d{8,} were compiling into the equivalent of
4286 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
4287 outcome, but in this particular case used more store than had been allocated,
4288 which caused the bug to be discovered because it threw up an internal error.
4289
4290 8. The debugging code in both pcre and pcretest for outputting the compiled
4291 form of a regex was going wrong in the case of back references followed by
4292 curly-bracketed repeats.
4293
4294
4295 Version 1.02 12-Dec-97
4296 ----------------------
4297
4298 1. Typos in pcre.3 and comments in the source fixed.
4299
4300 2. Applied a contributed patch to get rid of places where it used to remove
4301 'const' from variables, and fixed some signed/unsigned and uninitialized
4302 variable warnings.
4303
4304 3. Added the "runtest" target to Makefile.
4305
4306 4. Set default compiler flag to -O2 rather than just -O.
4307
4308
4309 Version 1.01 19-Nov-97
4310 ----------------------
4311
4312 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
4313 like /([ab]*)*/, that is, for classes with more than one character in them.
4314
4315 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
4316 as /((?>a*))*/ (a PCRE_EXTRA facility).
4317
4318
4319 Version 1.00 18-Nov-97
4320 ----------------------
4321
4322 1. Added compile-time macros to support systems such as SunOS4 which don't have
4323 memmove() or strerror() but have other things that can be used instead.
4324
4325 2. Arranged that "make clean" removes the executables.
4326
4327
4328 Version 0.99 27-Oct-97
4329 ----------------------
4330
4331 1. Fixed bug in code for optimizing classes with only one character. It was
4332 initializing a 32-byte map regardless, which could cause it to run off the end
4333 of the memory it had got.
4334
4335 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
4336
4337
4338 Version 0.98 22-Oct-97
4339 ----------------------
4340
4341 1. Fixed bug in code for handling temporary memory usage when there are more
4342 back references than supplied space in the ovector. This could cause segfaults.
4343
4344
4345 Version 0.97 21-Oct-97
4346 ----------------------
4347
4348 1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
4349
4350 2. Optimized negated single characters not to use a bit map.
4351
4352 3. Brought error texts together as macro definitions; clarified some of them;
4353 fixed one that was wrong - it said "range out of order" when it meant "invalid
4354 escape sequence".
4355
4356 4. Changed some char * arguments to const char *.
4357
4358 5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
4359
4360 6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
4361 pcretest.
4362
4363
4364 Version 0.96 16-Oct-97
4365 ----------------------
4366
4367 1. Added a simple "pgrep" utility to the distribution.
4368
4369 2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
4370 unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
4371 where "ddd" means "one or more decimal digits".
4372
4373 3. Fixed serious bug. If a pattern had a back reference, but the call to
4374 pcre_exec() didn't supply a large enough ovector to record the related
4375 identifying subpattern, the match always failed. PCRE now remembers the number
4376 of the largest back reference, and gets some temporary memory in which to save
4377 the offsets during matching if necessary, in order to ensure that
4378 backreferences always work.
4379
4380 4. Increased the compatibility with Perl in a number of ways:
4381
4382 (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
4383 to request this handling. The option can be set at compile or exec time.
4384
4385 (b) $ matches before a terminating newline by default; an option
4386 PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
4387 mode). The option can be set at compile or exec time.
4388
4389 (c) The handling of \ followed by a digit other than 0 is now supposed to be
4390 the same as Perl's. If the decimal number it represents is less than 10
4391 or there aren't that many previous left capturing parentheses, an octal
4392 escape is read. Inside a character class, it's always an octal escape,
4393 even if it is a single digit.
4394
4395 (d) An escaped but undefined alphabetic character is taken as a literal,
4396 unless PCRE_EXTRA is set. Currently this just reserves the remaining
4397 escapes.
4398
4399 (e) {0} is now permitted. (The previous item is removed from the compiled
4400 pattern).
4401
4402 5. Changed all the names of code files so that the basic parts are no longer
4403 than 10 characters, and abolished the teeny "globals.c" file.
4404
4405 6. Changed the handling of character classes; they are now done with a 32-byte
4406 bit map always.
4407
4408 7. Added the -d and /D options to pcretest to make it possible to look at the
4409 internals of compilation without having to recompile pcre.
4410
4411
4412 Version 0.95 23-Sep-97
4413 ----------------------
4414
4415 1. Fixed bug in pre-pass concerning escaped "normal" characters such as \x5c or
4416 \x20 at the start of a run of normal characters. These were being treated as
4417 real characters, instead of the source characters being re-checked.
4418
4419
4420 Version 0.94 18-Sep-97
4421 ----------------------
4422
4423 1. The functions are now thread-safe, with the caveat that the global variables
4424 containing pointers to malloc() and free() or alternative functions are the
4425 same for all threads.
4426
4427 2. Get pcre_study() to generate a bitmap of initial characters for non-
4428 anchored patterns when this is possible, and use it if passed to pcre_exec().
4429
4430
4431 Version 0.93 15-Sep-97
4432 ----------------------
4433
4434 1. /(b)|(:+)/ was computing an incorrect first character.
4435
4436 2. Add pcre_study() to the API and the passing of pcre_extra to pcre_exec(),
4437 but not actually doing anything yet.
4438
4439 3. Treat "-" characters in classes that cannot be part of ranges as literals,
4440 as Perl does (e.g. [-az] or [az-]).
4441
4442 4. Set the anchored flag if a branch starts with .* or .*? because that tests
4443 all possible positions.
4444
4445 5. Split up into different modules to avoid including unneeded functions in a
4446 compiled binary. However, compile and exec are still in one module. The "study"
4447 function is split off.
4448
4449 6. The character tables are now in a separate module whose source is generated
4450 by an auxiliary program - but can then be edited by hand if required. There are
4451 now no calls to isalnum(), isspace(), isdigit(), isxdigit(), tolower() or
4452 toupper() in the code.
4453
4454 7. Turn the malloc/free funtions variables into pcre_malloc and pcre_free and
4455 make them global. Abolish the function for setting them, as the caller can now
4456 set them directly.
4457
4458
4459 Version 0.92 11-Sep-97
4460 ----------------------
4461
4462 1. A repeat with a fixed maximum and a minimum of 1 for an ordinary character
4463 (e.g. /a{1,3}/) was broken (I mis-optimized it).
4464
4465 2. Caseless matching was not working in character classes if the characters in
4466 the pattern were in upper case.
4467
4468 3. Make ranges like [W-c] work in the same way as Perl for caseless matching.
4469
4470 4. Make PCRE_ANCHORED public and accept as a compile option.
4471
4472 5. Add an options word to pcre_exec() and accept PCRE_ANCHORED and
4473 PCRE_CASELESS at run time. Add escapes \A and \I to pcretest to cause it to
4474 pass them.
4475
4476 6. Give an error if bad option bits passed at compile or run time.
4477
4478 7. Add PCRE_MULTILINE at compile and exec time, and (?m) as well. Add \M to
4479 pcretest to cause it to pass that flag.
4480
4481 8. Add pcre_info(), to get the number of identifying subpatterns, the stored
4482 options, and the first character, if set.
4483
4484 9. Recognize C+ or C{n,m} where n >= 1 as providing a fixed starting character.
4485
4486
4487 Version 0.91 10-Sep-97
4488 ----------------------
4489
4490 1. PCRE was failing to diagnose unlimited repeats of subpatterns that could
4491 match the empty string as in /(a*)*/. It was looping and ultimately crashing.
4492
4493 2. PCRE was looping on encountering an indefinitely repeated back reference to
4494 a subpattern that had matched an empty string, e.g. /(a|)\1*/. It now does what
4495 Perl does - treats the match as successful.
4496
4497 ****

Properties

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

  ViewVC Help
Powered by ViewVC 1.1.5