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

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log


Revision 720 - (show annotations) (download)
Fri Oct 7 15:18:21 2011 UTC (3 years ago) by ph10
File size: 216524 byte(s)
Fix RunGrepTest colour bug.

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

Properties

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

webmaster@exim.org
ViewVC Help
Powered by ViewVC 1.1.12