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

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log


Revision 700 - (show annotations)
Tue Sep 20 10:51:26 2011 UTC (3 years, 10 months ago) by ph10
File size: 214321 byte(s)
Error occurred while calculating annotation data.
New RunTest.bat provided by Sheri Pierce.
1 ChangeLog for PCRE
2 ------------------
3
4 Version 8.20 12-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 captured
14 substring, but while checking the failing first alternative, substring 1 is
15 temporarily captured. If the output vector supplied to pcre_exec() was not
16 big enough for this capture, the yield of the function was still zero
17 ("insufficient space for captured substrings"). This cannot be totally fixed
18 without adding another stack variable, which seems a lot of expense for a
19 edge case. However, I have improved the situation in cases such as
20 /(a)(b)x|abc/ matched against "abc", where the return code indicates that
21 fewer than the maximum number of slots in the ovector have been set.
22
23 3. Related to (2) above: when there are more back references in a pattern than
24 slots in the output vector, pcre_exec() uses temporary memory during
25 matching, and copies in the captures as far as possible afterwards. It was
26 using the entire output vector, but this conflicts with the specification
27 that only 2/3 is used for passing back captured substrings. Now it uses only
28 the first 2/3, for compatibility. This is, of course, another edge case.
29
30 4. Zoltan Herczeg's just-in-time compiler support has been integrated into the
31 main code base, and can be used by building with --enable-jit. When this is
32 done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
33 runtime --no-jit option is given.
34
35 5. When the number of matches in a pcre_dfa_exec() run exactly filled the
36 ovector, the return from the function was zero, implying that there were
37 other matches that did not fit. The correct "exactly full" value is now
38 returned.
39
40 6. If a subpattern that was called recursively or as a subroutine contained
41 (*PRUNE) or any other control that caused it to give a non-standard return,
42 invalid errors such as "Error -26 (nested recursion at the same subject
43 position)" or even infinite loops could occur.
44
45 7. If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
46 computing the minimum length on reaching *ACCEPT, and so ended up with the
47 wrong value of 1 rather than 0. Further investigation indicates that
48 computing a minimum subject length in the presence of *ACCEPT is difficult
49 (think back references, subroutine calls), and so I have changed the code so
50 that no minimum is registered for a pattern that contains *ACCEPT.
51
52 8. If (*THEN) was present in the first (true) branch of a conditional group,
53 it was not handled as intended.
54
55 9. Replaced RunTest.bat with the much improved version provided by Sheri
56 Pierce.
57
58
59 Version 8.13 16-Aug-2011
60 ------------------------
61
62 1. The Unicode data tables have been updated to Unicode 6.0.0.
63
64 2. Two minor typos in pcre_internal.h have been fixed.
65
66 3. Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
67 pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
68 in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
69
70 4. There were a number of related bugs in the code for matching backrefences
71 caselessly in UTF-8 mode when codes for the characters concerned were
72 different numbers of bytes. For example, U+023A and U+2C65 are an upper
73 and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
74 (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
75 code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
76 2-byte code at the end of the subject (it thought there wasn't enough data
77 left).
78
79 5. Comprehensive information about what went wrong is now returned by
80 pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
81 as the output vector has at least 2 elements. The offset of the start of
82 the failing character and a reason code are placed in the vector.
83
84 6. When the UTF-8 string check fails for pcre_compile(), the offset that is
85 now returned is for the first byte of the failing character, instead of the
86 last byte inspected. This is an incompatible change, but I hope it is small
87 enough not to be a problem. It makes the returned offset consistent with
88 pcre_exec() and pcre_dfa_exec().
89
90 7. pcretest now gives a text phrase as well as the error number when
91 pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
92 failure, the offset and reason code are output.
93
94 8. When \R was used with a maximizing quantifier it failed to skip backwards
95 over a \r\n pair if the subsequent match failed. Instead, it just skipped
96 back over a single character (\n). This seems wrong (because it treated the
97 two characters as a single entity when going forwards), conflicts with the
98 documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
99 behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
100 has been changed.
101
102 9. Some internal refactoring has changed the processing so that the handling
103 of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
104 time (the PCRE_DOTALL option was changed this way some time ago: version
105 7.7 change 16). This has made it possible to abolish the OP_OPT op code,
106 which was always a bit of a fudge. It also means that there is one less
107 argument for the match() function, which reduces its stack requirements
108 slightly. This change also fixes an incompatibility with Perl: the pattern
109 (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
110
111 10. More internal refactoring has drastically reduced the number of recursive
112 calls to match() for possessively repeated groups such as (abc)++ when
113 using pcre_exec().
114
115 11. While implementing 10, a number of bugs in the handling of groups were
116 discovered and fixed:
117
118 (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
119 (a|)*(?1) gave a compile-time internal error.
120 ((a|)+)+ did not notice that the outer group could match an empty string.
121 (^a|^)+ was not marked as anchored.
122 (.*a|.*)+ was not marked as matching at start or after a newline.
123
124 12. Yet more internal refactoring has removed another argument from the match()
125 function. Special calls to this function are now indicated by setting a
126 value in a variable in the "match data" data block.
127
128 13. Be more explicit in pcre_study() instead of relying on "default" for
129 opcodes that mean there is no starting character; this means that when new
130 ones are added and accidentally left out of pcre_study(), testing should
131 pick them up.
132
133 14. The -s option of pcretest has been documented for ages as being an old
134 synonym of -m (show memory usage). I have changed it to mean "force study
135 for every regex", that is, assume /S for every regex. This is similar to -i
136 and -d etc. It's slightly incompatible, but I'm hoping nobody is still
137 using it. It makes it easier to run collections of tests with and without
138 study enabled, and thereby test pcre_study() more easily. All the standard
139 tests are now run with and without -s (but some patterns can be marked as
140 "never study" - see 20 below).
141
142 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
143 restoration of the capturing data to the outer values was not happening
144 correctly.
145
146 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
147 empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
148 pattern had matched an empty string, and so incorrectly returned a no
149 match.
150
151 17. There was optimizing code for the last branch of non-capturing parentheses,
152 and also for the obeyed branch of a conditional subexpression, which used
153 tail recursion to cut down on stack usage. Unfortunately, now that there is
154 the possibility of (*THEN) occurring in these branches, tail recursion is
155 no longer possible because the return has to be checked for (*THEN). These
156 two optimizations have therefore been removed.
157
158 18. If a pattern containing \R was studied, it was assumed that \R always
159 matched two bytes, thus causing the minimum subject length to be
160 incorrectly computed because \R can also match just one byte.
161
162 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
163 was incorrectly computed.
164
165 20. If /S is present twice on a test pattern in pcretest input, it now
166 *disables* studying, thereby overriding the use of -s on the command line
167 (see 14 above). This is necessary for one or two tests to keep the output
168 identical in both cases.
169
170 21. When (*ACCEPT) was used in an assertion that matched an empty string and
171 PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
172
173 22. When an atomic group that contained a capturing parenthesis was
174 successfully matched, but the branch in which it appeared failed, the
175 capturing was not being forgotten if a higher numbered group was later
176 captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
177 group 1 to "a", when in fact it should be unset. This applied to multi-
178 branched capturing and non-capturing groups, repeated or not, and also to
179 positive assertions (capturing in negative assertions does not happen
180 in PCRE) and also to nested atomic groups.
181
182 23. Add the ++ qualifier feature to pcretest, to show the remainder of the
183 subject after a captured substring, to make it easier to tell which of a
184 number of identical substrings has been captured.
185
186 24. The way atomic groups are processed by pcre_exec() has been changed so that
187 if they are repeated, backtracking one repetition now resets captured
188 values correctly. For example, if ((?>(a+)b)+aabab) is matched against
189 "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
190 "aaa". Previously, it would have been "a". As part of this code
191 refactoring, the way recursive calls are handled has also been changed.
192
193 25. If an assertion condition captured any substrings, they were not passed
194 back unless some other capturing happened later. For example, if
195 (?(?=(a))a) was matched against "a", no capturing was returned.
196
197 26. When studying a pattern that contained subroutine calls or assertions,
198 the code for finding the minimum length of a possible match was handling
199 direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
200 group 1 called group 2 while simultaneously a separate group 2 called group
201 1). A stack overflow occurred in this case. I have fixed this by limiting
202 the recursion depth to 10.
203
204 27. Updated RunTest.bat in the distribution to the version supplied by Tom
205 Fortmann. This supports explicit test numbers on the command line, and has
206 argument validation and error reporting.
207
208 28. An instance of \X with an unlimited repeat could fail if at any point the
209 first character it looked at was a mark character.
210
211 29. Some minor code refactoring concerning Unicode properties and scripts
212 should reduce the stack requirement of match() slightly.
213
214 30. Added the '=' option to pcretest to check the setting of unused capturing
215 slots at the end of the pattern, which are documented as being -1, but are
216 not included in the return count.
217
218 31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
219 compiled something random. Now it gives a compile-time error (as does
220 Perl).
221
222 32. A *MARK encountered during the processing of a positive assertion is now
223 recorded and passed back (compatible with Perl).
224
225 33. If --only-matching or --colour was set on a pcregrep call whose pattern
226 had alternative anchored branches, the search for a second match in a line
227 was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
228 matched the line "0102" twice. The same bug affected patterns that started
229 with a backwards assertion. For example /\b01|\b02/ also matched "0102"
230 twice.
231
232 34. Previously, PCRE did not allow quantification of assertions. However, Perl
233 does, and because of capturing effects, quantifying parenthesized
234 assertions may at times be useful. Quantifiers are now allowed for
235 parenthesized assertions.
236
237 35. A minor code tidy in pcre_compile() when checking options for \R usage.
238
239 36. \g was being checked for fancy things in a character class, when it should
240 just be a literal "g".
241
242 37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
243 appearance of a nested POSIX class supersedes an apparent external class.
244 For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
245 unescaped square brackets may also appear as part of class names. For
246 example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
247 more like Perl. (But see 8.20/1 above.)
248
249 38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
250 was because it thought it was \N{name}, which is not supported).
251
252 39. Add minix to OS list not supporting the -S option in pcretest.
253
254 40. PCRE tries to detect cases of infinite recursion at compile time, but it
255 cannot analyze patterns in sufficient detail to catch mutual recursions
256 such as ((?1))((?2)). There is now a runtime test that gives an error if a
257 subgroup is called recursively as a subpattern for a second time at the
258 same position in the subject string. In previous releases this might have
259 been caught by the recursion limit, or it might have run out of stack.
260
261 41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
262 happen only once. PCRE was, however incorrectly giving a compile time error
263 "recursive call could loop indefinitely" because it cannot analyze the
264 pattern in sufficient detail. The compile time test no longer happens when
265 PCRE is compiling a conditional subpattern, but actual runaway loops are
266 now caught at runtime (see 40 above).
267
268 42. It seems that Perl allows any characters other than a closing parenthesis
269 to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
270 has been changed to be the same.
271
272 43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
273 as not to get warnings when autogen.sh is called. Also changed
274 AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
275
276 44. To help people who use pcregrep to scan files containing exceedingly long
277 lines, the following changes have been made:
278
279 (a) The default value of the buffer size parameter has been increased from
280 8K to 20K. (The actual buffer used is three times this size.)
281
282 (b) The default can be changed by ./configure --with-pcregrep-bufsize when
283 PCRE is built.
284
285 (c) A --buffer-size=n option has been added to pcregrep, to allow the size
286 to be set at run time.
287
288 (d) Numerical values in pcregrep options can be followed by K or M, for
289 example --buffer-size=50K.
290
291 (e) If a line being scanned overflows pcregrep's buffer, an error is now
292 given and the return code is set to 2.
293
294 45. Add a pointer to the latest mark to the callout data block.
295
296 46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
297 partial match of an empty string instead of no match. This was specific to
298 the use of ".".
299
300 47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
301 complete match instead of a partial match. This bug was dependent on both
302 the PCRE_UTF8 and PCRE_DOTALL options being set.
303
304 48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
305 starting byte set, because \b was not being ignored.
306
307
308 Version 8.12 15-Jan-2011
309 ------------------------
310
311 1. Fixed some typos in the markup of the man pages, and wrote a script that
312 checks for such things as part of the documentation building process.
313
314 2. On a big-endian 64-bit system, pcregrep did not correctly process the
315 --match-limit and --recursion-limit options (added for 8.11). In
316 particular, this made one of the standard tests fail. (The integer value
317 went into the wrong half of a long int.)
318
319 3. If the --colour option was given to pcregrep with -v (invert match), it
320 did strange things, either producing crazy output, or crashing. It should,
321 of course, ignore a request for colour when reporting lines that do not
322 match.
323
324 4. Another pcregrep bug caused similar problems if --colour was specified with
325 -M (multiline) and the pattern match finished with a line ending.
326
327 5. In pcregrep, when a pattern that ended with a literal newline sequence was
328 matched in multiline mode, the following line was shown as part of the
329 match. This seems wrong, so I have changed it.
330
331 6. Another pcregrep bug in multiline mode, when --colour was specified, caused
332 the check for further matches in the same line (so they could be coloured)
333 to overrun the end of the current line. If another match was found, it was
334 incorrectly shown (and then shown again when found in the next line).
335
336 7. If pcregrep was compiled under Windows, there was a reference to the
337 function pcregrep_exit() before it was defined. I am assuming this was
338 the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
339 reported by a user. I've moved the definition above the reference.
340
341
342 Version 8.11 10-Dec-2010
343 ------------------------
344
345 1. (*THEN) was not working properly if there were untried alternatives prior
346 to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
347 backtracked to try for "b" instead of moving to the next alternative branch
348 at the same level (in this case, to look for "c"). The Perl documentation
349 is clear that when (*THEN) is backtracked onto, it goes to the "next
350 alternative in the innermost enclosing group".
351
352 2. (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
353 such as (A(*COMMIT)B(*THEN)C|D) any failure after matching A should
354 result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
355 (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
356 (*THEN).
357
358 3. If \s appeared in a character class, it removed the VT character from
359 the class, even if it had been included by some previous item, for example
360 in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
361 of \s, but is part of the POSIX "space" class.)
362
363 4. A partial match never returns an empty string (because you can always
364 match an empty string at the end of the subject); however the checking for
365 an empty string was starting at the "start of match" point. This has been
366 changed to the "earliest inspected character" point, because the returned
367 data for a partial match starts at this character. This means that, for
368 example, /(?<=abc)def/ gives a partial match for the subject "abc"
369 (previously it gave "no match").
370
371 5. Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
372 of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
373 previously a full match would be given. However, setting PCRE_PARTIAL_HARD
374 has an implication that the given string is incomplete (because a partial
375 match is preferred over a full match). For this reason, these items now
376 give a partial match in this situation. [Aside: previously, the one case
377 /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
378 match rather than a full match, which was wrong by the old rules, but is
379 now correct.]
380
381 6. There was a bug in the handling of #-introduced comments, recognized when
382 PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
383 If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
384 UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
385 scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
386 but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
387 places in pcre_compile().
388
389 7. Related to (6) above, when pcre_compile() was skipping #-introduced
390 comments when looking ahead for named forward references to subpatterns,
391 the only newline sequence it recognized was NL. It now handles newlines
392 according to the set newline convention.
393
394 8. SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
395 former, but used strtoul(), whereas pcretest avoided strtoul() but did not
396 cater for a lack of strerror(). These oversights have been fixed.
397
398 9. Added --match-limit and --recursion-limit to pcregrep.
399
400 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
401
402 11. When the -o option was used, pcregrep was setting a return code of 1, even
403 when matches were found, and --line-buffered was not being honoured.
404
405 12. Added an optional parentheses number to the -o and --only-matching options
406 of pcregrep.
407
408 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
409 can match an empty string. The code to do it in pcretest and pcredemo
410 needed fixing:
411
412 (a) When the newline convention was "crlf", pcretest got it wrong, skipping
413 only one byte after an empty string match just before CRLF (this case
414 just got forgotten; "any" and "anycrlf" were OK).
415
416 (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
417 mode when an empty string match preceded an ASCII character followed by
418 a non-ASCII character. (The code for advancing by one character rather
419 than one byte was nonsense.)
420
421 (c) The pcredemo.c sample program did not have any code at all to handle
422 the cases when CRLF is a valid newline sequence.
423
424 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
425 as a starting offset was within the subject string. There is now a new
426 error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
427 negative or greater than the length of the string. In order to test this,
428 pcretest is extended to allow the setting of negative starting offsets.
429
430 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
431 starting offset points to the beginning of a UTF-8 character was
432 unnecessarily clumsy. I tidied it up.
433
434 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
435 bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
436
437 17. Nobody had reported that the --include_dir option, which was added in
438 release 7.7 should have been called --include-dir (hyphen, not underscore)
439 for compatibility with GNU grep. I have changed it to --include-dir, but
440 left --include_dir as an undocumented synonym, and the same for
441 --exclude-dir, though that is not available in GNU grep, at least as of
442 release 2.5.4.
443
444 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
445 characters from a string of bytes) have been redefined so as not to use
446 loops, in order to improve performance in some environments. At the same
447 time, I abstracted some of the common code into auxiliary macros to save
448 repetition (this should not affect the compiled code).
449
450 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
451 compile-time error is now given if \c is not followed by an ASCII
452 character, that is, a byte less than 128. (In EBCDIC mode, the code is
453 different, and any byte value is allowed.)
454
455 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
456 START_OPTIMIZE option, which is now allowed at compile time - but just
457 passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
458 to pcregrep and other applications that have no direct access to PCRE
459 options. The new /Y option in pcretest sets this option when calling
460 pcre_compile().
461
462 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
463 back references. Groups containing recursive back references were forced to
464 be atomic by that change, but in the case of named groups, the amount of
465 memory required was incorrectly computed, leading to "Failed: internal
466 error: code overflow". This has been fixed.
467
468 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
469 pcretest.c, to avoid build problems in some Borland environments.
470
471
472 Version 8.10 25-Jun-2010
473 ------------------------
474
475 1. Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
476 THEN.
477
478 2. (*ACCEPT) was not working when inside an atomic group.
479
480 3. Inside a character class, \B is treated as a literal by default, but
481 faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
482 causes the error). The code is unchanged, but I tidied the documentation.
483
484 4. Inside a character class, PCRE always treated \R and \X as literals,
485 whereas Perl faults them if its -w option is set. I have changed PCRE so
486 that it faults them when PCRE_EXTRA is set.
487
488 5. Added support for \N, which always matches any character other than
489 newline. (It is the same as "." when PCRE_DOTALL is not set.)
490
491 6. When compiling pcregrep with newer versions of gcc which may have
492 FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
493 declared with attribute warn_unused_result" were given. Just casting the
494 result to (void) does not stop the warnings; a more elaborate fudge is
495 needed. I've used a macro to implement this.
496
497 7. Minor change to pcretest.c to avoid a compiler warning.
498
499 8. Added four artifical Unicode properties to help with an option to make
500 \s etc use properties (see next item). The new properties are: Xan
501 (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
502
503 9. Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
504 use Unicode properties. (*UCP) at the start of a pattern can be used to set
505 this option. Modified pcretest to add /W to test this facility. Added
506 REG_UCP to make it available via the POSIX interface.
507
508 10. Added --line-buffered to pcregrep.
509
510 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
511 studied, and the match started with a letter with a code point greater than
512 127 whose first byte was different to the first byte of the other case of
513 the letter, the other case of this starting letter was not recognized
514 (#976).
515
516 12. If a pattern that was studied started with a repeated Unicode property
517 test, for example, \p{Nd}+, there was the theoretical possibility of
518 setting up an incorrect bitmap of starting bytes, but fortunately it could
519 not have actually happened in practice until change 8 above was made (it
520 added property types that matched character-matching opcodes).
521
522 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
523 possible starting bytes for non-anchored patterns.
524
525 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
526 \R, and also a number of cases that involve Unicode properties, both
527 explicit and implicit when PCRE_UCP is set.
528
529 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
530 input, it could crash or give wrong results if characters with values
531 greater than 0xc0 were present in the subject string. (Detail: it assumed
532 UTF-8 input when processing these items.)
533
534 16. Added a lot of (int) casts to avoid compiler warnings in systems where
535 size_t is 64-bit (#991).
536
537 17. Added a check for running out of memory when PCRE is compiled with
538 --disable-stack-for-recursion (#990).
539
540 18. If the last data line in a file for pcretest does not have a newline on
541 the end, a newline was missing in the output.
542
543 19. The default pcre_chartables.c file recognizes only ASCII characters (values
544 less than 128) in its various bitmaps. However, there is a facility for
545 generating tables according to the current locale when PCRE is compiled. It
546 turns out that in some environments, 0x85 and 0xa0, which are Unicode space
547 characters, are recognized by isspace() and therefore were getting set in
548 these tables, and indeed these tables seem to approximate to ISO 8859. This
549 caused a problem in UTF-8 mode when pcre_study() was used to create a list
550 of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
551 which of course cannot start UTF-8 characters. I have changed the code so
552 that only real ASCII characters (less than 128) and the correct starting
553 bytes for UTF-8 encodings are set for characters greater than 127 when in
554 UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
555 altogether.)
556
557 20. Added the /T option to pcretest so as to be able to run tests with non-
558 standard character tables, thus making it possible to include the tests
559 used for 19 above in the standard set of tests.
560
561 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
562 reference to a subpattern the other side of a comment that contains an
563 opening parenthesis caused either an internal compiling error, or a
564 reference to the wrong subpattern.
565
566
567 Version 8.02 19-Mar-2010
568 ------------------------
569
570 1. The Unicode data tables have been updated to Unicode 5.2.0.
571
572 2. Added the option --libs-cpp to pcre-config, but only when C++ support is
573 configured.
574
575 3. Updated the licensing terms in the pcregexp.pas file, as agreed with the
576 original author of that file, following a query about its status.
577
578 4. On systems that do not have stdint.h (e.g. Solaris), check for and include
579 inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
580
581 5. A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
582 quantifier applied to a forward-referencing subroutine call, could compile
583 incorrect code or give the error "internal error: previously-checked
584 referenced subpattern not found".
585
586 6. Both MS Visual Studio and Symbian OS have problems with initializing
587 variables to point to external functions. For these systems, therefore,
588 pcre_malloc etc. are now initialized to local functions that call the
589 relevant global functions.
590
591 7. There were two entries missing in the vectors called coptable and poptable
592 in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
593 I've fixed the data, and added a kludgy way of testing at compile time that
594 the lengths are correct (equal to the number of opcodes).
595
596 8. Following on from 7, I added a similar kludge to check the length of the
597 eint vector in pcreposix.c.
598
599 9. Error texts for pcre_compile() are held as one long string to avoid too
600 much relocation at load time. To find a text, the string is searched,
601 counting zeros. There was no check for running off the end of the string,
602 which could happen if a new error number was added without updating the
603 string.
604
605 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
606
607 11. \K was not working if it appeared in an atomic group or in a group that
608 was called as a "subroutine", or in an assertion. Perl 5.11 documents that
609 \K is "not well defined" if used in an assertion. PCRE now accepts it if
610 the assertion is positive, but not if it is negative.
611
612 12. Change 11 fortuitously reduced the size of the stack frame used in the
613 "match()" function of pcre_exec.c by one pointer. Forthcoming
614 implementation of support for (*MARK) will need an extra pointer on the
615 stack; I have reserved it now, so that the stack frame size does not
616 decrease.
617
618 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
619 item in branch that calls a recursion is a subroutine call - as in the
620 second branch in the above example - was incorrectly given the compile-
621 time error "recursive call could loop indefinitely" because pcre_compile()
622 was not correctly checking the subroutine for matching a non-empty string.
623
624 14. The checks for overrunning compiling workspace could trigger after an
625 overrun had occurred. This is a "should never occur" error, but it can be
626 triggered by pathological patterns such as hundreds of nested parentheses.
627 The checks now trigger 100 bytes before the end of the workspace.
628
629 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
630
631
632 Version 8.01 19-Jan-2010
633 ------------------------
634
635 1. If a pattern contained a conditional subpattern with only one branch (in
636 particular, this includes all (*DEFINE) patterns), a call to pcre_study()
637 computed the wrong minimum data length (which is of course zero for such
638 subpatterns). This could cause incorrect "no match" results.
639
640 2. For patterns such as (?i)a(?-i)b|c where an option setting at the start of
641 the pattern is reset in the first branch, pcre_compile() failed with
642 "internal error: code overflow at offset...". This happened only when
643 the reset was to the original external option setting. (An optimization
644 abstracts leading options settings into an external setting, which was the
645 cause of this.)
646
647 3. A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
648 of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
649 assertion pattern did not match (meaning that the assertion was true), it
650 was incorrectly treated as false if the SKIP had been reached during the
651 matching. This also applied to assertions used as conditions.
652
653 4. If an item that is not supported by pcre_dfa_exec() was encountered in an
654 assertion subpattern, including such a pattern used as a condition,
655 unpredictable results occurred, instead of the error return
656 PCRE_ERROR_DFA_UITEM.
657
658 5. The C++ GlobalReplace function was not working like Perl for the special
659 situation when an empty string is matched. It now does the fancy magic
660 stuff that is necessary.
661
662 6. In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
663 removed. (These were left over from very, very early versions of PCRE.)
664
665 7. Some cosmetic changes to the code to make life easier when compiling it
666 as part of something else:
667
668 (a) Change DEBUG to PCRE_DEBUG.
669
670 (b) In pcre_compile(), rename the member of the "branch_chain" structure
671 called "current" as "current_branch", to prevent a collision with the
672 Linux macro when compiled as a kernel module.
673
674 (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
675 prevent a collision with the Linux macro when compiled as a kernel
676 module.
677
678 8. In pcre_compile() there are some checks for integer overflows that used to
679 cast potentially large values to (double). This has been changed to that
680 when building, a check for int64_t is made, and if it is found, it is used
681 instead, thus avoiding the use of floating point arithmetic. (There is no
682 other use of FP in PCRE.) If int64_t is not found, the fallback is to
683 double.
684
685 9. Added two casts to avoid signed/unsigned warnings from VS Studio Express
686 2005 (difference between two addresses compared to an unsigned value).
687
688 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
689 custom one, because of the following reported problem in Windows:
690
691 - libbz2 uses the Pascal calling convention (WINAPI) for the functions
692 under Win32.
693 - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
694 therefore missing the function definition.
695 - The compiler thus generates a "C" signature for the test function.
696 - The linker fails to find the "C" function.
697 - PCRE fails to configure if asked to do so against libbz2.
698
699 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
700 messages were output:
701
702 Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
703 rerunning libtoolize, to keep the correct libtool macros in-tree.
704 Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
705
706 I have done both of these things.
707
708 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
709 most of the time, it *can* run out if it is given a pattern that contains a
710 runaway infinite recursion. I updated the discussion in the pcrestack man
711 page.
712
713 13. Now that we have gone to the x.xx style of version numbers, the minor
714 version may start with zero. Using 08 or 09 is a bad idea because users
715 might check the value of PCRE_MINOR in their code, and 08 or 09 may be
716 interpreted as invalid octal numbers. I've updated the previous comment in
717 configure.ac, and also added a check that gives an error if 08 or 09 are
718 used.
719
720 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
721 causing partial matching to fail when the end of the subject matched \W
722 in a UTF-8 pattern where \W was quantified with a minimum of 3.
723
724 15. There were some discrepancies between the declarations in pcre_internal.h
725 of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
726 their definitions. The declarations used "const uschar *" and the
727 definitions used USPTR. Even though USPTR is normally defined as "const
728 unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
729 reported that: "This difference in casting confuses some C++ compilers, for
730 example, SunCC recognizes above declarations as different functions and
731 generates broken code for hbpcre." I have changed the declarations to use
732 USPTR.
733
734 16. GNU libtool is named differently on some systems. The autogen.sh script now
735 tries several variants such as glibtoolize (MacOSX) and libtoolize1x
736 (FreeBSD).
737
738 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
739 (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
740 comment: "Figure out how to create a longlong from a string: strtoll and
741 equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
742 instance, but it only takes 2 args instead of 3!"
743
744 18. A subtle bug concerned with back references has been fixed by a change of
745 specification, with a corresponding code fix. A pattern such as
746 ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
747 refers, was giving matches when it shouldn't. For example, xa=xaaa would
748 match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
749 same bug. Such groups have to be quantified to be useful, or contained
750 inside another quantified group. (If there's no repetition, the reference
751 can never match.) The problem arises because, having left the group and
752 moved on to the rest of the pattern, a later failure that backtracks into
753 the group uses the captured value from the final iteration of the group
754 rather than the correct earlier one. I have fixed this in PCRE by forcing
755 any group that contains a reference to itself to be an atomic group; that
756 is, there cannot be any backtracking into it once it has completed. This is
757 similar to recursive and subroutine calls.
758
759
760 Version 8.00 19-Oct-09
761 ----------------------
762
763 1. The table for translating pcre_compile() error codes into POSIX error codes
764 was out-of-date, and there was no check on the pcre_compile() error code
765 being within the table. This could lead to an OK return being given in
766 error.
767
768 2. Changed the call to open a subject file in pcregrep from fopen(pathname,
769 "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
770 in a Windows environment.
771
772 3. The pcregrep --count option prints the count for each file even when it is
773 zero, as does GNU grep. However, pcregrep was also printing all files when
774 --files-with-matches was added. Now, when both options are given, it prints
775 counts only for those files that have at least one match. (GNU grep just
776 prints the file name in this circumstance, but including the count seems
777 more useful - otherwise, why use --count?) Also ensured that the
778 combination -clh just lists non-zero counts, with no names.
779
780 4. The long form of the pcregrep -F option was incorrectly implemented as
781 --fixed_strings instead of --fixed-strings. This is an incompatible change,
782 but it seems right to fix it, and I didn't think it was worth preserving
783 the old behaviour.
784
785 5. The command line items --regex=pattern and --regexp=pattern were not
786 recognized by pcregrep, which required --regex pattern or --regexp pattern
787 (with a space rather than an '='). The man page documented the '=' forms,
788 which are compatible with GNU grep; these now work.
789
790 6. No libpcreposix.pc file was created for pkg-config; there was just
791 libpcre.pc and libpcrecpp.pc. The omission has been rectified.
792
793 7. Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
794 when UCP support is not needed, by modifying the Python script that
795 generates it from Unicode data files. This should not matter if the module
796 is correctly used as a library, but I received one complaint about 50K of
797 unwanted data. My guess is that the person linked everything into his
798 program rather than using a library. Anyway, it does no harm.
799
800 8. A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
801 was a minimum greater than 1 for a wide character in a possessive
802 repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
803 which had an unlimited repeat of a nested, fixed maximum repeat of a wide
804 character. Chaos in the form of incorrect output or a compiling loop could
805 result.
806
807 9. The restrictions on what a pattern can contain when partial matching is
808 requested for pcre_exec() have been removed. All patterns can now be
809 partially matched by this function. In addition, if there are at least two
810 slots in the offset vector, the offset of the earliest inspected character
811 for the match and the offset of the end of the subject are set in them when
812 PCRE_ERROR_PARTIAL is returned.
813
814 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
815 synonymous with PCRE_PARTIAL, for backwards compatibility, and
816 PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
817 and may be more useful for multi-segment matching.
818
819 11. Partial matching with pcre_exec() is now more intuitive. A partial match
820 used to be given if ever the end of the subject was reached; now it is
821 given only if matching could not proceed because another character was
822 needed. This makes a difference in some odd cases such as Z(*FAIL) with the
823 string "Z", which now yields "no match" instead of "partial match". In the
824 case of pcre_dfa_exec(), "no match" is given if every matching path for the
825 final character ended with (*FAIL).
826
827 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
828 if the pattern had a "must contain" character that was already found in the
829 earlier partial match, unless partial matching was again requested. For
830 example, with the pattern /dog.(body)?/, the "must contain" character is
831 "g". If the first part-match was for the string "dog", restarting with
832 "sbody" failed. This bug has been fixed.
833
834 13. The string returned by pcre_dfa_exec() after a partial match has been
835 changed so that it starts at the first inspected character rather than the
836 first character of the match. This makes a difference only if the pattern
837 starts with a lookbehind assertion or \b or \B (\K is not supported by
838 pcre_dfa_exec()). It's an incompatible change, but it makes the two
839 matching functions compatible, and I think it's the right thing to do.
840
841 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
842 so that the demonstration program is easily available in environments where
843 PCRE has not been installed from source.
844
845 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
846 libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
847 library.
848
849 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
850 It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
851 is not the first non-POSIX option to be added. Clearly some people find
852 these options useful.
853
854 17. If a caller to the POSIX matching function regexec() passes a non-zero
855 value for nmatch with a NULL value for pmatch, the value of
856 nmatch is forced to zero.
857
858 18. RunGrepTest did not have a test for the availability of the -u option of
859 the diff command, as RunTest does. It now checks in the same way as
860 RunTest, and also checks for the -b option.
861
862 19. If an odd number of negated classes containing just a single character
863 interposed, within parentheses, between a forward reference to a named
864 subpattern and the definition of the subpattern, compilation crashed with
865 an internal error, complaining that it could not find the referenced
866 subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
867 [The bug was that it was starting one character too far in when skipping
868 over the character class, thus treating the ] as data rather than
869 terminating the class. This meant it could skip too much.]
870
871 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
872 /g option in pcretest when the pattern contains \K, which makes it possible
873 to have an empty string match not at the start, even when the pattern is
874 anchored. Updated pcretest and pcredemo to use this option.
875
876 21. If the maximum number of capturing subpatterns in a recursion was greater
877 than the maximum at the outer level, the higher number was returned, but
878 with unset values at the outer level. The correct (outer level) value is
879 now given.
880
881 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
882 PCRE did not set those parentheses (unlike Perl). I have now found a way to
883 make it do so. The string so far is captured, making this feature
884 compatible with Perl.
885
886 23. The tests have been re-organized, adding tests 11 and 12, to make it
887 possible to check the Perl 5.10 features against Perl 5.10.
888
889 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
890 pattern matches a fixed length string. PCRE did not allow this; now it
891 does. Neither allows recursion.
892
893 25. I finally figured out how to implement a request to provide the minimum
894 length of subject string that was needed in order to match a given pattern.
895 (It was back references and recursion that I had previously got hung up
896 on.) This code has now been added to pcre_study(); it finds a lower bound
897 to the length of subject needed. It is not necessarily the greatest lower
898 bound, but using it to avoid searching strings that are too short does give
899 some useful speed-ups. The value is available to calling programs via
900 pcre_fullinfo().
901
902 26. While implementing 25, I discovered to my embarrassment that pcretest had
903 not been passing the result of pcre_study() to pcre_dfa_exec(), so the
904 study optimizations had never been tested with that matching function.
905 Oops. What is worse, even when it was passed study data, there was a bug in
906 pcre_dfa_exec() that meant it never actually used it. Double oops. There
907 were also very few tests of studied patterns with pcre_dfa_exec().
908
909 27. If (?| is used to create subpatterns with duplicate numbers, they are now
910 allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
911 on the other side of the coin, they are no longer allowed to have different
912 names, because these cannot be distinguished in PCRE, and this has caused
913 confusion. (This is a difference from Perl.)
914
915 28. When duplicate subpattern names are present (necessarily with different
916 numbers, as required by 27 above), and a test is made by name in a
917 conditional pattern, either for a subpattern having been matched, or for
918 recursion in such a pattern, all the associated numbered subpatterns are
919 tested, and the overall condition is true if the condition is true for any
920 one of them. This is the way Perl works, and is also more like the way
921 testing by number works.
922
923
924 Version 7.9 11-Apr-09
925 ---------------------
926
927 1. When building with support for bzlib/zlib (pcregrep) and/or readline
928 (pcretest), all targets were linked against these libraries. This included
929 libpcre, libpcreposix, and libpcrecpp, even though they do not use these
930 libraries. This caused unwanted dependencies to be created. This problem
931 has been fixed, and now only pcregrep is linked with bzlib/zlib and only
932 pcretest is linked with readline.
933
934 2. The "typedef int BOOL" in pcre_internal.h that was included inside the
935 "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
936 moved outside it again, because FALSE and TRUE are already defined in AIX,
937 but BOOL is not.
938
939 3. The pcre_config() function was treating the PCRE_MATCH_LIMIT and
940 PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
941
942 4. The pcregrep documentation said spaces were inserted as well as colons (or
943 hyphens) following file names and line numbers when outputting matching
944 lines. This is not true; no spaces are inserted. I have also clarified the
945 wording for the --colour (or --color) option.
946
947 5. In pcregrep, when --colour was used with -o, the list of matching strings
948 was not coloured; this is different to GNU grep, so I have changed it to be
949 the same.
950
951 6. When --colo(u)r was used in pcregrep, only the first matching substring in
952 each matching line was coloured. Now it goes on to look for further matches
953 of any of the test patterns, which is the same behaviour as GNU grep.
954
955 7. A pattern that could match an empty string could cause pcregrep to loop; it
956 doesn't make sense to accept an empty string match in pcregrep, so I have
957 locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
958 seems to be how GNU grep behaves.
959
960 8. The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
961 start or after a newline", because the conditional assertion was not being
962 correctly handled. The rule now is that both the assertion and what follows
963 in the first alternative must satisfy the test.
964
965 9. If auto-callout was enabled in a pattern with a conditional group whose
966 condition was an assertion, PCRE could crash during matching, both with
967 pcre_exec() and pcre_dfa_exec().
968
969 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
970 used for matching.
971
972 11. Unicode property support in character classes was not working for
973 characters (bytes) greater than 127 when not in UTF-8 mode.
974
975 12. Added the -M command line option to pcretest.
976
977 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
978
979 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
980
981 16. Added comments and documentation about mis-use of no_arg in the C++
982 wrapper.
983
984 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
985 from Martin Jerabek that uses macro names for all relevant character and
986 string constants.
987
988 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
989 SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
990 SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
991 these, but not everybody uses configure.
992
993 19. A conditional group that had only one branch was not being correctly
994 recognized as an item that could match an empty string. This meant that an
995 enclosing group might also not be so recognized, causing infinite looping
996 (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
997 with the subject "ab", where knowledge that the repeated group can match
998 nothing is needed in order to break the loop.
999
1000 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
1001 exec(), but without supplying a callout function, matching went wrong.
1002
1003 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
1004 leak if the size of the offset vector was greater than 30. When the vector
1005 is smaller, the saved offsets during recursion go onto a local stack
1006 vector, but for larger vectors malloc() is used. It was failing to free
1007 when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
1008 error, in fact).
1009
1010 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
1011 heapframe that is used only when UTF-8 support is enabled. This caused no
1012 problem, but was untidy.
1013
1014 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
1015 CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
1016 included within another project.
1017
1018 24. Steven Van Ingelgem's patches to add more options to the CMake support,
1019 slightly modified by me:
1020
1021 (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
1022 not building pcregrep.
1023
1024 (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
1025 if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
1026
1027 25. Forward references, both numeric and by name, in patterns that made use of
1028 duplicate group numbers, could behave incorrectly or give incorrect errors,
1029 because when scanning forward to find the reference group, PCRE was not
1030 taking into account the duplicate group numbers. A pattern such as
1031 ^X(?3)(a)(?|(b)|(q))(Y) is an example.
1032
1033 26. Changed a few more instances of "const unsigned char *" to USPTR, making
1034 the feature of a custom pointer more persuasive (as requested by a user).
1035
1036 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
1037 pcretest.c, inside #ifndefs, because it seems they are sometimes already
1038 pre-defined.
1039
1040 28. Added support for (*UTF8) at the start of a pattern.
1041
1042 29. Arrange for flags added by the "release type" setting in CMake to be shown
1043 in the configuration summary.
1044
1045
1046 Version 7.8 05-Sep-08
1047 ---------------------
1048
1049 1. Replaced UCP searching code with optimized version as implemented for Ad
1050 Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
1051 stage table and inline lookup instead of a function, giving speed ups of 2
1052 to 5 times on some simple patterns that I tested. Permission was given to
1053 distribute the MultiStage2.py script that generates the tables (it's not in
1054 the tarball, but is in the Subversion repository).
1055
1056 2. Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
1057 scripts.
1058
1059 3. Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
1060 a group with a zero qualifier. The result of the study could be incorrect,
1061 or the function might crash, depending on the pattern.
1062
1063 4. Caseless matching was not working for non-ASCII characters in back
1064 references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
1065 It now works when Unicode Property Support is available.
1066
1067 5. In pcretest, an escape such as \x{de} in the data was always generating
1068 a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
1069 non-UTF-8 mode. If the value is greater than 255, it gives a warning about
1070 truncation.
1071
1072 6. Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
1073
1074 7. Added two (int) casts to pcregrep when printing the difference of two
1075 pointers, in case they are 64-bit values.
1076
1077 8. Added comments about Mac OS X stack usage to the pcrestack man page and to
1078 test 2 if it fails.
1079
1080 9. Added PCRE_CALL_CONVENTION just before the names of all exported functions,
1081 and a #define of that name to empty if it is not externally set. This is to
1082 allow users of MSVC to set it if necessary.
1083
1084 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
1085 the convenience functions in the pcre_get.c source file.
1086
1087 11. An option change at the start of a pattern that had top-level alternatives
1088 could cause overwriting and/or a crash. This command provoked a crash in
1089 some environments:
1090
1091 printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
1092
1093 This potential security problem was recorded as CVE-2008-2371.
1094
1095 12. For a pattern where the match had to start at the beginning or immediately
1096 after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
1097 pcre_dfa_exec() could read past the end of the passed subject if there was
1098 no match. To help with detecting such bugs (e.g. with valgrind), I modified
1099 pcretest so that it places the subject at the end of its malloc-ed buffer.
1100
1101 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
1102 exec() might read past the end of the data buffer in UTF-8 mode.
1103
1104 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
1105 the data contained the byte 0x85 as part of a UTF-8 character within its
1106 first line. This applied both to normal and DFA matching.
1107
1108 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
1109 /^[^d]*?$/8 failed to match "abc".
1110
1111 16. Added a missing copyright notice to pcrecpp_internal.h.
1112
1113 17. Make it more clear in the documentation that values returned from
1114 pcre_exec() in ovector are byte offsets, not character counts.
1115
1116 18. Tidied a few places to stop certain compilers from issuing warnings.
1117
1118 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
1119 supplied by Stefan Weber. I made a further small update for 7.8 because
1120 there is a change of source arrangements: the pcre_searchfuncs.c module is
1121 replaced by pcre_ucd.c.
1122
1123
1124 Version 7.7 07-May-08
1125 ---------------------
1126
1127 1. Applied Craig's patch to sort out a long long problem: "If we can't convert
1128 a string to a long long, pretend we don't even have a long long." This is
1129 done by checking for the strtoq, strtoll, and _strtoi64 functions.
1130
1131 2. Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
1132 pre-7.6 versions, which defined a global no_arg variable instead of putting
1133 it in the RE class. (See also #8 below.)
1134
1135 3. Remove a line of dead code, identified by coverity and reported by Nuno
1136 Lopes.
1137
1138 4. Fixed two related pcregrep bugs involving -r with --include or --exclude:
1139
1140 (1) The include/exclude patterns were being applied to the whole pathnames
1141 of files, instead of just to the final components.
1142
1143 (2) If there was more than one level of directory, the subdirectories were
1144 skipped unless they satisfied the include/exclude conditions. This is
1145 inconsistent with GNU grep (and could even be seen as contrary to the
1146 pcregrep specification - which I improved to make it absolutely clear).
1147 The action now is always to scan all levels of directory, and just
1148 apply the include/exclude patterns to regular files.
1149
1150 5. Added the --include_dir and --exclude_dir patterns to pcregrep, and used
1151 --exclude_dir in the tests to avoid scanning .svn directories.
1152
1153 6. Applied Craig's patch to the QuoteMeta function so that it escapes the
1154 NUL character as backslash + 0 rather than backslash + NUL, because PCRE
1155 doesn't support NULs in patterns.
1156
1157 7. Added some missing "const"s to declarations of static tables in
1158 pcre_compile.c and pcre_dfa_exec.c.
1159
1160 8. Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
1161 caused by fix #2 above. (Subsequently also a second patch to fix the
1162 first patch. And a third patch - this was a messy problem.)
1163
1164 9. Applied Craig's patch to remove the use of push_back().
1165
1166 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
1167 matching function regexec().
1168
1169 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
1170 which, however, unlike Perl's \g{...}, are subroutine calls, not back
1171 references. PCRE supports relative numbers with this syntax (I don't think
1172 Oniguruma does).
1173
1174 12. Previously, a group with a zero repeat such as (...){0} was completely
1175 omitted from the compiled regex. However, this means that if the group
1176 was called as a subroutine from elsewhere in the pattern, things went wrong
1177 (an internal error was given). Such groups are now left in the compiled
1178 pattern, with a new opcode that causes them to be skipped at execution
1179 time.
1180
1181 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
1182 to the way PCRE behaves:
1183
1184 (a) A lone ] character is dis-allowed (Perl treats it as data).
1185
1186 (b) A back reference to an unmatched subpattern matches an empty string
1187 (Perl fails the current match path).
1188
1189 (c) A data ] in a character class must be notated as \] because if the
1190 first data character in a class is ], it defines an empty class. (In
1191 Perl it is not possible to have an empty class.) The empty class []
1192 never matches; it forces failure and is equivalent to (*FAIL) or (?!).
1193 The negative empty class [^] matches any one character, independently
1194 of the DOTALL setting.
1195
1196 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
1197 non-existent subpattern following a character class starting with ']' and
1198 containing () gave an internal compiling error instead of "reference to
1199 non-existent subpattern". Fortunately, when the pattern did exist, the
1200 compiled code was correct. (When scanning forwards to check for the
1201 existencd of the subpattern, it was treating the data ']' as terminating
1202 the class, so got the count wrong. When actually compiling, the reference
1203 was subsequently set up correctly.)
1204
1205 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
1206 it was being rejected as not supported by pcre_dfa_exec(), even though
1207 other assertions are supported. I have made pcre_dfa_exec() support
1208 (*FAIL).
1209
1210 16. The implementation of 13c above involved the invention of a new opcode,
1211 OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
1212 cannot be changed at match time, I realized I could make a small
1213 improvement to matching performance by compiling OP_ALLANY instead of
1214 OP_ANY for "." when DOTALL was set, and then removing the runtime tests
1215 on the OP_ANY path.
1216
1217 17. Compiling pcretest on Windows with readline support failed without the
1218 following two fixes: (1) Make the unistd.h include conditional on
1219 HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
1220
1221 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
1222 ncurses library to be included for pcretest when ReadLine support is
1223 requested, but also to allow for it to be overridden. This patch came from
1224 Daniel Bergström.
1225
1226 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
1227 as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
1228 any errors with the current Unicode tables. Thanks to Peter Kankowski for
1229 spotting this.
1230
1231
1232 Version 7.6 28-Jan-08
1233 ---------------------
1234
1235 1. A character class containing a very large number of characters with
1236 codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
1237 overflow.
1238
1239 2. Patch to cut out the "long long" test in pcrecpp_unittest when
1240 HAVE_LONG_LONG is not defined.
1241
1242 3. Applied Christian Ehrlicher's patch to update the CMake build files to
1243 bring them up to date and include new features. This patch includes:
1244
1245 - Fixed PH's badly added libz and libbz2 support.
1246 - Fixed a problem with static linking.
1247 - Added pcredemo. [But later removed - see 7 below.]
1248 - Fixed dftables problem and added an option.
1249 - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
1250 HAVE_LONG_LONG.
1251 - Added readline support for pcretest.
1252 - Added an listing of the option settings after cmake has run.
1253
1254 4. A user submitted a patch to Makefile that makes it easy to create
1255 "pcre.dll" under mingw when using Configure/Make. I added stuff to
1256 Makefile.am that cause it to include this special target, without
1257 affecting anything else. Note that the same mingw target plus all
1258 the other distribution libraries and programs are now supported
1259 when configuring with CMake (see 6 below) instead of with
1260 Configure/Make.
1261
1262 5. Applied Craig's patch that moves no_arg into the RE class in the C++ code.
1263 This is an attempt to solve the reported problem "pcrecpp::no_arg is not
1264 exported in the Windows port". It has not yet been confirmed that the patch
1265 solves the problem, but it does no harm.
1266
1267 6. Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
1268 NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
1269 with CMake, and also correct the comment about stack recursion.
1270
1271 7. Remove the automatic building of pcredemo from the ./configure system and
1272 from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
1273 of a program that users should build themselves after PCRE is installed, so
1274 building it automatically is not really right. What is more, it gave
1275 trouble in some build environments.
1276
1277 8. Further tidies to CMakeLists.txt from Sheri and Christian.
1278
1279
1280 Version 7.5 10-Jan-08
1281 ---------------------
1282
1283 1. Applied a patch from Craig: "This patch makes it possible to 'ignore'
1284 values in parens when parsing an RE using the C++ wrapper."
1285
1286 2. Negative specials like \S did not work in character classes in UTF-8 mode.
1287 Characters greater than 255 were excluded from the class instead of being
1288 included.
1289
1290 3. The same bug as (2) above applied to negated POSIX classes such as
1291 [:^space:].
1292
1293 4. PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
1294 defined or documented. It seems to have been a typo for PCRE_STATIC, so
1295 I have changed it.
1296
1297 5. The construct (?&) was not diagnosed as a syntax error (it referenced the
1298 first named subpattern) and a construct such as (?&a) would reference the
1299 first named subpattern whose name started with "a" (in other words, the
1300 length check was missing). Both these problems are fixed. "Subpattern name
1301 expected" is now given for (?&) (a zero-length name), and this patch also
1302 makes it give the same error for \k'' (previously it complained that that
1303 was a reference to a non-existent subpattern).
1304
1305 6. The erroneous patterns (?+-a) and (?-+a) give different error messages;
1306 this is right because (?- can be followed by option settings as well as by
1307 digits. I have, however, made the messages clearer.
1308
1309 7. Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
1310 than the number used in the conditional) now cause a compile-time error.
1311 This is actually not compatible with Perl, which accepts such patterns, but
1312 treats the conditional as always being FALSE (as PCRE used to), but it
1313 seems to me that giving a diagnostic is better.
1314
1315 8. Change "alphameric" to the more common word "alphanumeric" in comments
1316 and messages.
1317
1318 9. Fix two occurrences of "backslash" in comments that should have been
1319 "backspace".
1320
1321 10. Remove two redundant lines of code that can never be obeyed (their function
1322 was moved elsewhere).
1323
1324 11. The program that makes PCRE's Unicode character property table had a bug
1325 which caused it to generate incorrect table entries for sequences of
1326 characters that have the same character type, but are in different scripts.
1327 It amalgamated them into a single range, with the script of the first of
1328 them. In other words, some characters were in the wrong script. There were
1329 thirteen such cases, affecting characters in the following ranges:
1330
1331 U+002b0 - U+002c1
1332 U+0060c - U+0060d
1333 U+0061e - U+00612
1334 U+0064b - U+0065e
1335 U+0074d - U+0076d
1336 U+01800 - U+01805
1337 U+01d00 - U+01d77
1338 U+01d9b - U+01dbf
1339 U+0200b - U+0200f
1340 U+030fc - U+030fe
1341 U+03260 - U+0327f
1342 U+0fb46 - U+0fbb1
1343 U+10450 - U+1049d
1344
1345 12. The -o option (show only the matching part of a line) for pcregrep was not
1346 compatible with GNU grep in that, if there was more than one match in a
1347 line, it showed only the first of them. It now behaves in the same way as
1348 GNU grep.
1349
1350 13. If the -o and -v options were combined for pcregrep, it printed a blank
1351 line for every non-matching line. GNU grep prints nothing, and pcregrep now
1352 does the same. The return code can be used to tell if there were any
1353 non-matching lines.
1354
1355 14. Added --file-offsets and --line-offsets to pcregrep.
1356
1357 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
1358 infinitely looping recursion. The bug was that positive lookaheads were not
1359 being skipped when checking for a possible empty match (negative lookaheads
1360 and both kinds of lookbehind were skipped).
1361
1362 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
1363 inclusion of <windows.h> to before rather than after the definition of
1364 INVALID_FILE_ATTRIBUTES (patch from David Byron).
1365
1366 17. Specifying a possessive quantifier with a specific limit for a Unicode
1367 character property caused pcre_compile() to compile bad code, which led at
1368 runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
1369 are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
1370 caused the error; without that there was no problem.
1371
1372 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
1373
1374 19. Added --enable-pcretest-libreadline.
1375
1376 20. In pcrecpp.cc, the variable 'count' was incremented twice in
1377 RE::GlobalReplace(). As a result, the number of replacements returned was
1378 double what it should be. I removed one of the increments, but Craig sent a
1379 later patch that removed the other one (the right fix) and added unit tests
1380 that check the return values (which was not done before).
1381
1382 21. Several CMake things:
1383
1384 (1) Arranged that, when cmake is used on Unix, the libraries end up with
1385 the names libpcre and libpcreposix, not just pcre and pcreposix.
1386
1387 (2) The above change means that pcretest and pcregrep are now correctly
1388 linked with the newly-built libraries, not previously installed ones.
1389
1390 (3) Added PCRE_SUPPORT_LIBREADLINE, PCRE_SUPPORT_LIBZ, PCRE_SUPPORT_LIBBZ2.
1391
1392 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
1393 crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
1394 UTF-8 newline character). The key issue is that the pattern starts .*;
1395 this means that the match must be either at the beginning, or after a
1396 newline. The bug was in the code for advancing after a failed match and
1397 checking that the new position followed a newline. It was not taking
1398 account of UTF-8 characters correctly.
1399
1400 23. PCRE was behaving differently from Perl in the way it recognized POSIX
1401 character classes. PCRE was not treating the sequence [:...:] as a
1402 character class unless the ... were all letters. Perl, however, seems to
1403 allow any characters between [: and :], though of course it rejects as
1404 unknown any "names" that contain non-letters, because all the known class
1405 names consist only of letters. Thus, Perl gives an error for [[:1234:]],
1406 for example, whereas PCRE did not - it did not recognize a POSIX character
1407 class. This seemed a bit dangerous, so the code has been changed to be
1408 closer to Perl. The behaviour is not identical to Perl, because PCRE will
1409 diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
1410 treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
1411 Perl does, and where it didn't before.
1412
1413 24. Rewrite so as to remove the single use of %n from pcregrep because in some
1414 Windows environments %n is disabled by default.
1415
1416
1417 Version 7.4 21-Sep-07
1418 ---------------------
1419
1420 1. Change 7.3/28 was implemented for classes by looking at the bitmap. This
1421 means that a class such as [\s] counted as "explicit reference to CR or
1422 LF". That isn't really right - the whole point of the change was to try to
1423 help when there was an actual mention of one of the two characters. So now
1424 the change happens only if \r or \n (or a literal CR or LF) character is
1425 encountered.
1426
1427 2. The 32-bit options word was also used for 6 internal flags, but the numbers
1428 of both had grown to the point where there were only 3 bits left.
1429 Fortunately, there was spare space in the data structure, and so I have
1430 moved the internal flags into a new 16-bit field to free up more option
1431 bits.
1432
1433 3. The appearance of (?J) at the start of a pattern set the DUPNAMES option,
1434 but did not set the internal JCHANGED flag - either of these is enough to
1435 control the way the "get" function works - but the PCRE_INFO_JCHANGED
1436 facility is supposed to tell if (?J) was ever used, so now (?J) at the
1437 start sets both bits.
1438
1439 4. Added options (at build time, compile time, exec time) to change \R from
1440 matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
1441
1442 5. doc/pcresyntax.html was missing from the distribution.
1443
1444 6. Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
1445 compatibility, even though it is no longer used.
1446
1447 7. Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
1448 strtoull to pcrecpp.cc to select the available functions in WIN32 when the
1449 windows.h file is present (where different names are used). [This was
1450 reversed later after testing - see 16 below.]
1451
1452 8. Changed all #include <config.h> to #include "config.h". There were also
1453 some further <pcre.h> cases that I changed to "pcre.h".
1454
1455 9. When pcregrep was used with the --colour option, it missed the line ending
1456 sequence off the lines that it output.
1457
1458 10. It was pointed out to me that arrays of string pointers cause lots of
1459 relocations when a shared library is dynamically loaded. A technique of
1460 using a single long string with a table of offsets can drastically reduce
1461 these. I have refactored PCRE in four places to do this. The result is
1462 dramatic:
1463
1464 Originally: 290
1465 After changing UCP table: 187
1466 After changing error message table: 43
1467 After changing table of "verbs" 36
1468 After changing table of Posix names 22
1469
1470 Thanks to the folks working on Gregex for glib for this insight.
1471
1472 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
1473 unicode-properties was also set.
1474
1475 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
1476
1477 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
1478 checked only for CRLF.
1479
1480 14. Added casts to pcretest.c to avoid compiler warnings.
1481
1482 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
1483
1484 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
1485 and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
1486 entirely. This removes changes made in 7 above.
1487
1488 17. The CMake files have been updated, and there is now more information about
1489 building with CMake in the NON-UNIX-USE document.
1490
1491
1492 Version 7.3 28-Aug-07
1493 ---------------------
1494
1495 1. In the rejigging of the build system that eventually resulted in 7.1, the
1496 line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
1497 brackets there is not right, since it causes compilers to look for an
1498 installed pcre.h, not the version that is in the source that is being
1499 compiled (which of course may be different). I have changed it back to:
1500
1501 #include "pcre.h"
1502
1503 I have a vague recollection that the change was concerned with compiling in
1504 different directories, but in the new build system, that is taken care of
1505 by the VPATH setting the Makefile.
1506
1507 2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
1508 when the subject happened to end in the byte 0x85 (e.g. if the last
1509 character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
1510 characters but of course it shouldn't be taken as a newline when it is part
1511 of another character. The bug was that, for an unlimited repeat of . in
1512 not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
1513 characters when looking for a newline.
1514
1515 3. A small performance improvement in the DOTALL UTF-8 mode .* case.
1516
1517 4. Debugging: adjusted the names of opcodes for different kinds of parentheses
1518 in debug output.
1519
1520 5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
1521 long printing in the pcrecpp unittest when running under MinGW.
1522
1523 6. ESC_K was left out of the EBCDIC table.
1524
1525 7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
1526 parentheses; I made it 1000, which seemed large enough. Unfortunately, the
1527 limit also applies to "virtual nesting" when a pattern is recursive, and in
1528 this case 1000 isn't so big. I have been able to remove this limit at the
1529 expense of backing off one optimization in certain circumstances. Normally,
1530 when pcre_exec() would call its internal match() function recursively and
1531 immediately return the result unconditionally, it uses a "tail recursion"
1532 feature to save stack. However, when a subpattern that can match an empty
1533 string has an unlimited repetition quantifier, it no longer makes this
1534 optimization. That gives it a stack frame in which to save the data for
1535 checking that an empty string has been matched. Previously this was taken
1536 from the 1000-entry workspace that had been reserved. So now there is no
1537 explicit limit, but more stack is used.
1538
1539 8. Applied Daniel's patches to solve problems with the import/export magic
1540 syntax that is required for Windows, and which was going wrong for the
1541 pcreposix and pcrecpp parts of the library. These were overlooked when this
1542 problem was solved for the main library.
1543
1544 9. There were some crude static tests to avoid integer overflow when computing
1545 the size of patterns that contain repeated groups with explicit upper
1546 limits. As the maximum quantifier is 65535, the maximum group length was
1547 set at 30,000 so that the product of these two numbers did not overflow a
1548 32-bit integer. However, it turns out that people want to use groups that
1549 are longer than 30,000 bytes (though not repeat them that many times).
1550 Change 7.0/17 (the refactoring of the way the pattern size is computed) has
1551 made it possible to implement the integer overflow checks in a much more
1552 dynamic way, which I have now done. The artificial limitation on group
1553 length has been removed - we now have only the limit on the total length of
1554 the compiled pattern, which depends on the LINK_SIZE setting.
1555
1556 10. Fixed a bug in the documentation for get/copy named substring when
1557 duplicate names are permitted. If none of the named substrings are set, the
1558 functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
1559 empty string.
1560
1561 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
1562 instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
1563 because the ] is interpreted as the first data character and the
1564 terminating ] is not found. PCRE has been made compatible with Perl in this
1565 regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
1566 cause memory overwriting.
1567
1568 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
1569 string has been matched (to stop an infinite loop). It was not recognizing
1570 a conditional subpattern that could match an empty string if that
1571 subpattern was within another subpattern. For example, it looped when
1572 trying to match (((?(1)X|))*) but it was OK with ((?(1)X|)*) where the
1573 condition was not nested. This bug has been fixed.
1574
1575 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
1576 past the start of the subject in the presence of bytes with the top bit
1577 set, for example "\x8aBCD".
1578
1579 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
1580 (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
1581
1582 14. Optimized (?!) to (*FAIL).
1583
1584 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
1585 This restricts code points to be within the range 0 to 0x10FFFF, excluding
1586 the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
1587 full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
1588 does: it's just the validity check that is more restrictive.
1589
1590 16. Inserted checks for integer overflows during escape sequence (backslash)
1591 processing, and also fixed erroneous offset values for syntax errors during
1592 backslash processing.
1593
1594 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
1595 for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
1596
1597 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
1598 caused an overrun.
1599
1600 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
1601 something other than just ASCII characters) inside a group that had an
1602 unlimited repeat caused a loop at compile time (while checking to see
1603 whether the group could match an empty string).
1604
1605 20. Debugging a pattern containing \p or \P could cause a crash. For example,
1606 [\P{Any}] did so. (Error in the code for printing property names.)
1607
1608 21. An orphan \E inside a character class could cause a crash.
1609
1610 22. A repeated capturing bracket such as (A)? could cause a wild memory
1611 reference during compilation.
1612
1613 23. There are several functions in pcre_compile() that scan along a compiled
1614 expression for various reasons (e.g. to see if it's fixed length for look
1615 behind). There were bugs in these functions when a repeated \p or \P was
1616 present in the pattern. These operators have additional parameters compared
1617 with \d, etc, and these were not being taken into account when moving along
1618 the compiled data. Specifically:
1619
1620 (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
1621 length.
1622
1623 (b) An item such as \pL+ within a repeated group could cause crashes or
1624 loops.
1625
1626 (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
1627 "reference to non-existent subpattern" error.
1628
1629 (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
1630
1631 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
1632 characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
1633
1634 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
1635
1636 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
1637 character were causing crashes (broken optimization).
1638
1639 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
1640 \p or \P) caused a compile-time loop.
1641
1642 28. More problems have arisen in unanchored patterns when CRLF is a valid line
1643 break. For example, the unstudied pattern [\r\n]A does not match the string
1644 "\r\nA" because change 7.0/46 below moves the current point on by two
1645 characters after failing to match at the start. However, the pattern \nA
1646 *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
1647 the same is true. There doesn't seem any very clean way out of this, but
1648 what I have chosen to do makes the common cases work: PCRE now takes note
1649 of whether there can be an explicit match for \r or \n anywhere in the
1650 pattern, and if so, 7.0/46 no longer applies. As part of this change,
1651 there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
1652 pattern has explicit CR or LF references.
1653
1654 29. Added (*CR) etc for changing newline setting at start of pattern.
1655
1656
1657 Version 7.2 19-Jun-07
1658 ---------------------
1659
1660 1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
1661 which is apparently normally available under Windows.
1662
1663 2. Re-jig the pcregrep tests with different newline settings in an attempt
1664 to make them independent of the local environment's newline setting.
1665
1666 3. Add code to configure.ac to remove -g from the CFLAGS default settings.
1667
1668 4. Some of the "internals" tests were previously cut out when the link size
1669 was not 2, because the output contained actual offsets. The recent new
1670 "Z" feature of pcretest means that these can be cut out, making the tests
1671 usable with all link sizes.
1672
1673 5. Implemented Stan Switzer's goto replacement for longjmp() when not using
1674 stack recursion. This gives a massive performance boost under BSD, but just
1675 a small improvement under Linux. However, it saves one field in the frame
1676 in all cases.
1677
1678 6. Added more features from the forthcoming Perl 5.10:
1679
1680 (a) (?-n) (where n is a string of digits) is a relative subroutine or
1681 recursion call. It refers to the nth most recently opened parentheses.
1682
1683 (b) (?+n) is also a relative subroutine call; it refers to the nth next
1684 to be opened parentheses.
1685
1686 (c) Conditions that refer to capturing parentheses can be specified
1687 relatively, for example, (?(-2)... or (?(+3)...
1688
1689 (d) \K resets the start of the current match so that everything before
1690 is not part of it.
1691
1692 (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
1693
1694 (f) \g{name} is another synonym - part of Perl 5.10's unification of
1695 reference syntax.
1696
1697 (g) (?| introduces a group in which the numbering of parentheses in each
1698 alternative starts with the same number.
1699
1700 (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
1701
1702 7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
1703 PCRE_INFO_JCHANGED.
1704
1705 8. A pattern such as (.*(.)?)* caused pcre_exec() to fail by either not
1706 terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
1707 for detecting groups that can match an empty string.
1708
1709 9. A pattern with a very large number of alternatives (more than several
1710 hundred) was running out of internal workspace during the pre-compile
1711 phase, where pcre_compile() figures out how much memory will be needed. A
1712 bit of new cunning has reduced the workspace needed for groups with
1713 alternatives. The 1000-alternative test pattern now uses 12 bytes of
1714 workspace instead of running out of the 4096 that are available.
1715
1716 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
1717
1718 11. Applied patch from Google to remove an optimization that didn't quite work.
1719 The report of the bug said:
1720
1721 pcrecpp::RE("a*").FullMatch("aaa") matches, while
1722 pcrecpp::RE("a*?").FullMatch("aaa") does not, and
1723 pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
1724
1725 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
1726 it matched the wrong number of bytes.
1727
1728
1729 Version 7.1 24-Apr-07
1730 ---------------------
1731
1732 1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
1733 that is more "standard", making use of automake and other Autotools. There
1734 is some re-arrangement of the files and adjustment of comments consequent
1735 on this.
1736
1737 2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
1738 for recursive directory scanning broke on some systems because the files
1739 are not scanned in any specific order and on different systems the order
1740 was different. A call to "sort" has been inserted into RunGrepTest for the
1741 approprate test as a short-term fix. In the longer term there may be an
1742 alternative.
1743
1744 3. I had an email from Eric Raymond about problems translating some of PCRE's
1745 man pages to HTML (despite the fact that I distribute HTML pages, some
1746 people do their own conversions for various reasons). The problems
1747 concerned the use of low-level troff macros .br and .in. I have therefore
1748 removed all such uses from the man pages (some were redundant, some could
1749 be replaced by .nf/.fi pairs). The 132html script that I use to generate
1750 HTML has been updated to handle .nf/.fi and to complain if it encounters
1751 .br or .in.
1752
1753 4. Updated comments in configure.ac that get placed in config.h.in and also
1754 arranged for config.h to be included in the distribution, with the name
1755 config.h.generic, for the benefit of those who have to compile without
1756 Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
1757
1758 5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
1759 Weber: (1) pcre_internal.h was missing some function renames; (2) updated
1760 makevp.bat for the current PCRE, using the additional files
1761 makevp_c.txt, makevp_l.txt, and pcregexp.pas.
1762
1763 6. A Windows user reported a minor discrepancy with test 2, which turned out
1764 to be caused by a trailing space on an input line that had got lost in his
1765 copy. The trailing space was an accident, so I've just removed it.
1766
1767 7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
1768 that is needed.
1769
1770 8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
1771 as "const" (a) because they are and (b) because it helps the PHP
1772 maintainers who have recently made a script to detect big data structures
1773 in the php code that should be moved to the .rodata section. I remembered
1774 to update Builducptable as well, so it won't revert if ucptable.h is ever
1775 re-created.
1776
1777 9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
1778 pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
1779 order to be able to cut out the UTF-8 tables in the latter when UTF-8
1780 support is not required. This saves 1.5-2K of code, which is important in
1781 some applications.
1782
1783 Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
1784 so as not to refer to the tables, even though these functions will never be
1785 called when UTF-8 support is disabled. Otherwise there are problems with a
1786 shared library.
1787
1788 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
1789
1790 (a) It was defining its arguments as char * instead of void *.
1791
1792 (b) It was assuming that all moves were upwards in memory; this was true
1793 a long time ago when I wrote it, but is no longer the case.
1794
1795 The emulated memove() is provided for those environments that have neither
1796 memmove() nor bcopy(). I didn't think anyone used it these days, but that
1797 is clearly not the case, as these two bugs were recently reported.
1798
1799 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
1800 and Detrail to create the HTML documentation, the .txt form of the man
1801 pages, and it removes trailing spaces from listed files. It also creates
1802 pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
1803 case, it wraps all the #defines with #ifndefs. This script should be run
1804 before "make dist".
1805
1806 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
1807 with Unicode property support.
1808
1809 (a) For a maximizing quantifier, if the two different cases of the
1810 character were of different lengths in their UTF-8 codings (there are
1811 some cases like this - I found 11), and the matching function had to
1812 back up over a mixture of the two cases, it incorrectly assumed they
1813 were both the same length.
1814
1815 (b) When PCRE was configured to use the heap rather than the stack for
1816 recursion during matching, it was not correctly preserving the data for
1817 the other case of a UTF-8 character when checking ahead for a match
1818 while processing a minimizing repeat. If the check also involved
1819 matching a wide character, but failed, corruption could cause an
1820 erroneous result when trying to check for a repeat of the original
1821 character.
1822
1823 13. Some tidying changes to the testing mechanism:
1824
1825 (a) The RunTest script now detects the internal link size and whether there
1826 is UTF-8 and UCP support by running ./pcretest -C instead of relying on
1827 values substituted by "configure". (The RunGrepTest script already did
1828 this for UTF-8.) The configure.ac script no longer substitutes the
1829 relevant variables.
1830
1831 (b) The debugging options /B and /D in pcretest show the compiled bytecode
1832 with length and offset values. This means that the output is different
1833 for different internal link sizes. Test 2 is skipped for link sizes
1834 other than 2 because of this, bypassing the problem. Unfortunately,
1835 there was also a test in test 3 (the locale tests) that used /B and
1836 failed for link sizes other than 2. Rather than cut the whole test out,
1837 I have added a new /Z option to pcretest that replaces the length and
1838 offset values with spaces. This is now used to make test 3 independent
1839 of link size. (Test 2 will be tidied up later.)
1840
1841 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
1842 segmentation fault instead of returning the appropriate error message.
1843
1844 15. In multiline mode when the newline sequence was set to "any", the pattern
1845 ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
1846 This doesn't seem right; it now treats the CRLF combination as the line
1847 ending, and so does not match in that case. It's only a pattern such as ^$
1848 that would hit this one: something like ^ABC$ would have failed after \r
1849 and then tried again after \r\n.
1850
1851 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
1852 in an attempt to make files that differ only in their line terminators
1853 compare equal. This works on Linux.
1854
1855 17. Under certain error circumstances pcregrep might try to free random memory
1856 as it exited. This is now fixed, thanks to valgrind.
1857
1858 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
1859 "abc\r\n\r\n", it found an unwanted second match after the second \r. This
1860 was because its rules for how to advance for /g after matching an empty
1861 string at the end of a line did not allow for this case. They now check for
1862 it specially.
1863
1864 20. pcretest is supposed to handle patterns and data of any length, by
1865 extending its buffers when necessary. It was getting this wrong when the
1866 buffer for a data line had to be extended.
1867
1868 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
1869 CRLF as a newline sequence.
1870
1871 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
1872 out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
1873 I have nevertheless tidied it up.
1874
1875 23. Added some casts to kill warnings from HP-UX ia64 compiler.
1876
1877 24. Added a man page for pcre-config.
1878
1879
1880 Version 7.0 19-Dec-06
1881 ---------------------
1882
1883 1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
1884 moving to gcc 4.1.1.
1885
1886 2. The -S option for pcretest uses setrlimit(); I had omitted to #include
1887 sys/time.h, which is documented as needed for this function. It doesn't
1888 seem to matter on Linux, but it showed up on some releases of OS X.
1889
1890 3. It seems that there are systems where bytes whose values are greater than
1891 127 match isprint() in the "C" locale. The "C" locale should be the
1892 default when a C program starts up. In most systems, only ASCII printing
1893 characters match isprint(). This difference caused the output from pcretest
1894 to vary, making some of the tests fail. I have changed pcretest so that:
1895
1896 (a) When it is outputting text in the compiled version of a pattern, bytes
1897 other than 32-126 are always shown as hex escapes.
1898
1899 (b) When it is outputting text that is a matched part of a subject string,
1900 it does the same, unless a different locale has been set for the match
1901 (using the /L modifier). In this case, it uses isprint() to decide.
1902
1903 4. Fixed a major bug that caused incorrect computation of the amount of memory
1904 required for a compiled pattern when options that changed within the
1905 pattern affected the logic of the preliminary scan that determines the
1906 length. The relevant options are -x, and -i in UTF-8 mode. The result was
1907 that the computed length was too small. The symptoms of this bug were
1908 either the PCRE error "internal error: code overflow" from pcre_compile(),
1909 or a glibc crash with a message such as "pcretest: free(): invalid next
1910 size (fast)". Examples of patterns that provoked this bug (shown in
1911 pcretest format) are:
1912
1913 /(?-x: )/x
1914 /(?x)(?-x: \s*#\s*)/
1915 /((?i)[\x{c0}])/8
1916 /(?i:[\x{c0}])/8
1917
1918 HOWEVER: Change 17 below makes this fix obsolete as the memory computation
1919 is now done differently.
1920
1921 5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
1922 wrapper classes; (b) implement a new function in the C++ scanner that is
1923 more efficient than the old way of doing things because it avoids levels of
1924 recursion in the regex matching; (c) add a paragraph to the documentation
1925 for the FullMatch() function.
1926
1927 6. The escape sequence \n was being treated as whatever was defined as
1928 "newline". Not only was this contrary to the documentation, which states
1929 that \n is character 10 (hex 0A), but it also went horribly wrong when
1930 "newline" was defined as CRLF. This has been fixed.
1931
1932 7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
1933 was being set to -1 for the "end of line" case (supposedly a value that no
1934 character can have). Though this value is never used (the check for end of
1935 line is "zero bytes in current character"), it caused compiler complaints.
1936 I've changed it to 0xffffffff.
1937
1938 8. In pcre_version.c, the version string was being built by a sequence of
1939 C macros that, in the event of PCRE_PRERELEASE being defined as an empty
1940 string (as it is for production releases) called a macro with an empty
1941 argument. The C standard says the result of this is undefined. The gcc
1942 compiler treats it as an empty string (which was what was wanted) but it is
1943 reported that Visual C gives an error. The source has been hacked around to
1944 avoid this problem.
1945
1946 9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
1947 builds of pcretest, and changed the call to _setmode() to use _O_BINARY
1948 instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
1949 of them did).
1950
1951 10. Originally, pcretest opened its input and output without "b"; then I was
1952 told that "b" was needed in some environments, so it was added for release
1953 5.0 to both the input and output. (It makes no difference on Unix-like
1954 systems.) Later I was told that it is wrong for the input on Windows. I've
1955 now abstracted the modes into two macros, to make it easier to fiddle with
1956 them, and removed "b" from the input mode under Windows.
1957
1958 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
1959
1960 12. Added -help and --help to pcretest as an official way of being reminded
1961 of the options.
1962
1963 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
1964 and pcrecpp.cc because they annoy compilers at high warning levels.
1965
1966 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
1967
1968 15. Fixed an occurrence of == in configure.ac that should have been = (shell
1969 scripts are not C programs :-) and which was not noticed because it works
1970 on Linux.
1971
1972 16. pcretest is supposed to handle any length of pattern and data line (as one
1973 line or as a continued sequence of lines) by extending its input buffer if
1974 necessary. This feature was broken for very long pattern lines, leading to
1975 a string of junk being passed to pcre_compile() if the pattern was longer
1976 than about 50K.
1977
1978 17. I have done a major re-factoring of the way pcre_compile() computes the
1979 amount of memory needed for a compiled pattern. Previously, there was code
1980 that made a preliminary scan of the pattern in order to do this. That was
1981 OK when PCRE was new, but as the facilities have expanded, it has become
1982 harder and harder to keep it in step with the real compile phase, and there
1983 have been a number of bugs (see for example, 4 above). I have now found a
1984 cunning way of running the real compile function in a "fake" mode that
1985 enables it to compute how much memory it would need, while actually only
1986 ever using a few hundred bytes of working memory and without too many
1987 tests of the mode. This should make future maintenance and development
1988 easier. A side effect of this work is that the limit of 200 on the nesting
1989 depth of parentheses has been removed (though this was never a serious
1990 limitation, I suspect). However, there is a downside: pcre_compile() now
1991 runs more slowly than before (30% or more, depending on the pattern). I
1992 hope this isn't a big issue. There is no effect on runtime performance.
1993
1994 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
1995 newline (only possible for the last line of a file) and it was a
1996 pattern that set a locale (followed by /Lsomething), pcretest crashed.
1997
1998 19. Added additional timing features to pcretest. (1) The -tm option now times
1999 matching only, not compiling. (2) Both -t and -tm can be followed, as a
2000 separate command line item, by a number that specifies the number of
2001 repeats to use when timing. The default is 50000; this gives better
2002 precision, but takes uncomfortably long for very large patterns.
2003
2004 20. Extended pcre_study() to be more clever in cases where a branch of a
2005 subpattern has no definite first character. For example, (a*|b*)[cd] would
2006 previously give no result from pcre_study(). Now it recognizes that the
2007 first character must be a, b, c, or d.
2008
2009 21. There was an incorrect error "recursive call could loop indefinitely" if
2010 a subpattern (or the entire pattern) that was being tested for matching an
2011 empty string contained only one non-empty item after a nested subpattern.
2012 For example, the pattern (?>\x{100}*)\d(?R) provoked this error
2013 incorrectly, because the \d was being skipped in the check.
2014
2015 22. The pcretest program now has a new pattern option /B and a command line
2016 option -b, which is equivalent to adding /B to every pattern. This causes
2017 it to show the compiled bytecode, without the additional information that
2018 -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
2019 is the same as /B/I).
2020
2021 23. A new optimization is now able automatically to treat some sequences such
2022 as a*b as a*+b. More specifically, if something simple (such as a character
2023 or a simple class like \d) has an unlimited quantifier, and is followed by
2024 something that cannot possibly match the quantified thing, the quantifier
2025 is automatically "possessified".
2026
2027 24. A recursive reference to a subpattern whose number was greater than 39
2028 went wrong under certain circumstances in UTF-8 mode. This bug could also
2029 have affected the operation of pcre_study().
2030
2031 25. Realized that a little bit of performance could be had by replacing
2032 (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
2033
2034 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
2035
2036 27. Possessive quantifiers such as a++ were previously implemented by turning
2037 them into atomic groups such as ($>a+). Now they have their own opcodes,
2038 which improves performance. This includes the automatically created ones
2039 from 23 above.
2040
2041 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
2042 lookahead was broken if it was not anchored. PCRE was mistakenly expecting
2043 the first matched character to be a colon. This applied both to named and
2044 numbered groups.
2045
2046 29. The ucpinternal.h header file was missing its idempotency #ifdef.
2047
2048 30. I was sent a "project" file called libpcre.a.dev which I understand makes
2049 building PCRE on Windows easier, so I have included it in the distribution.
2050
2051 31. There is now a check in pcretest against a ridiculously large number being
2052 returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
2053 loop, the loop is abandoned.
2054
2055 32. Forward references to subpatterns in conditions such as (?(2)...) where
2056 subpattern 2 is defined later cause pcre_compile() to search forwards in
2057 the pattern for the relevant set of parentheses. This search went wrong
2058 when there were unescaped parentheses in a character class, parentheses
2059 escaped with \Q...\E, or parentheses in a #-comment in /x mode.
2060
2061 33. "Subroutine" calls and backreferences were previously restricted to
2062 referencing subpatterns earlier in the regex. This restriction has now
2063 been removed.
2064
2065 34. Added a number of extra features that are going to be in Perl 5.10. On the
2066 whole, these are just syntactic alternatives for features that PCRE had
2067 previously implemented using the Python syntax or my own invention. The
2068 other formats are all retained for compatibility.
2069
2070 (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
2071 as (?P<name>...). The new forms, as well as being in Perl 5.10, are
2072 also .NET compatible.
2073
2074 (b) A recursion or subroutine call to a named group can now be defined as
2075 (?&name) as well as (?P>name).
2076
2077 (c) A backreference to a named group can now be defined as \k<name> or
2078 \k'name' as well as (?P=name). The new forms, as well as being in Perl
2079 5.10, are also .NET compatible.
2080
2081 (d) A conditional reference to a named group can now use the syntax
2082 (?(<name>) or (?('name') as well as (?(name).
2083
2084 (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
2085 groups (named and numbered) that are never evaluated inline, but can be
2086 called as "subroutines" from elsewhere. In effect, the DEFINE condition
2087 is always false. There may be only one alternative in such a group.
2088
2089 (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
2090 as the simple (?(R). The condition is true only if the most recent
2091 recursion is that of the given number or name. It does not search out
2092 through the entire recursion stack.
2093
2094 (g) The escape \gN or \g{N} has been added, where N is a positive or
2095 negative number, specifying an absolute or relative reference.
2096
2097 35. Tidied to get rid of some further signed/unsigned compiler warnings and
2098 some "unreachable code" warnings.
2099
2100 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
2101 things, this adds five new scripts.
2102
2103 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
2104 There were also incompatibilities regarding the handling of \Q..\E inside
2105 character classes, for example with patterns like [\Qa\E-\Qz\E] where the
2106 hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
2107
2108 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
2109 matches an empty string, and forcibly breaks the loop. There were bugs in
2110 this code in non-simple cases. For a pattern such as ^(a()*)* matched
2111 against aaaa the result was just "a" rather than "aaaa", for example. Two
2112 separate and independent bugs (that affected different cases) have been
2113 fixed.
2114
2115 39. Refactored the code to abolish the use of different opcodes for small
2116 capturing bracket numbers. This is a tidy that I avoided doing when I
2117 removed the limit on the number of capturing brackets for 3.5 back in 2001.
2118 The new approach is not only tidier, it makes it possible to reduce the
2119 memory needed to fix the previous bug (38).
2120
2121 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
2122 sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
2123 processing dot, circumflex, or dollar metacharacters, or #-comments in /x
2124 mode.
2125
2126 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
2127 report.
2128
2129 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
2130 copy construction and assignment in the C++ wrapper.
2131
2132 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
2133 couple of bugs that could have given wrong results in the "--newline=crlf"
2134 case.
2135
2136 44. Added a number of casts and did some reorganization of signed/unsigned int
2137 variables following suggestions from Dair Grant. Also renamed the variable
2138 "this" as "item" because it is a C++ keyword.
2139
2140 45. Arranged for dftables to add
2141
2142 #include "pcre_internal.h"
2143
2144 to pcre_chartables.c because without it, gcc 4.x may remove the array
2145 definition from the final binary if PCRE is built into a static library and
2146 dead code stripping is activated.
2147
2148 46. For an unanchored pattern, if a match attempt fails at the start of a
2149 newline sequence, and the newline setting is CRLF or ANY, and the next two
2150 characters are CRLF, advance by two characters instead of one.
2151
2152
2153 Version 6.7 04-Jul-06
2154 ---------------------
2155
2156 1. In order to handle tests when input lines are enormously long, pcretest has
2157 been re-factored so that it automatically extends its buffers when
2158 necessary. The code is crude, but this _is_ just a test program. The
2159 default size has been increased from 32K to 50K.
2160
2161 2. The code in pcre_study() was using the value of the re argument before
2162 testing it for NULL. (Of course, in any sensible call of the function, it
2163 won't be NULL.)
2164
2165 3. The memmove() emulation function in pcre_internal.h, which is used on
2166 systems that lack both memmove() and bcopy() - that is, hardly ever -
2167 was missing a "static" storage class specifier.
2168
2169 4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
2170 containing an extended class (one that cannot be represented by a bitmap
2171 because it contains high-valued characters or Unicode property items, e.g.
2172 [\pZ]). Almost always one would set UTF-8 mode when processing such a
2173 pattern, but PCRE should not loop if you do not (it no longer does).
2174 [Detail: two cases were found: (a) a repeated subpattern containing an
2175 extended class; (b) a recursive reference to a subpattern that followed a
2176 previous extended class. It wasn't skipping over the extended class
2177 correctly when UTF-8 mode was not set.]
2178
2179 5. A negated single-character class was not being recognized as fixed-length
2180 in lookbehind assertions such as (?<=[^f]), leading to an incorrect
2181 compile error "lookbehind assertion is not fixed length".
2182
2183 6. The RunPerlTest auxiliary script was showing an unexpected difference
2184 between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
2185 write a Perl script that can interpret lines of an input file either as
2186 byte characters or as UTF-8, which is what "perltest" was being required to
2187 do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
2188 can't do is switch easily at run time between having the "use utf8;" pragma
2189 or not. In the end, I fudged it by using the RunPerlTest script to insert
2190 "use utf8;" explicitly for the UTF-8 tests.
2191
2192 7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
2193 the end of the subject string, contrary to the documentation and to what
2194 Perl does. This was true of both matching functions. Now it matches only at
2195 the start of the subject and immediately after *internal* newlines.
2196
2197 8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
2198 a pointer to an int instead of a pointer to an unsigned long int. This
2199 caused problems on 64-bit systems.
2200
2201 9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
2202 instance of the 'standard' template library not being so standard".
2203
2204 10. There was no check on the number of named subpatterns nor the maximum
2205 length of a subpattern name. The product of these values is used to compute
2206 the size of the memory block for a compiled pattern. By supplying a very
2207 long subpattern name and a large number of named subpatterns, the size
2208 computation could be caused to overflow. This is now prevented by limiting
2209 the length of names to 32 characters, and the number of named subpatterns
2210 to 10,000.
2211
2212 11. Subpatterns that are repeated with specific counts have to be replicated in
2213 the compiled pattern. The size of memory for this was computed from the
2214 length of the subpattern and the repeat count. The latter is limited to
2215 65535, but there was no limit on the former, meaning that integer overflow
2216 could in principle occur. The compiled length of a repeated subpattern is
2217 now limited to 30,000 bytes in order to prevent this.
2218
2219 12. Added the optional facility to have named substrings with the same name.
2220
2221 13. Added the ability to use a named substring as a condition, using the
2222 Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
2223 are numbers (not recommended). Forward references are permitted.
2224
2225 14. Added forward references in named backreferences (if you see what I mean).
2226
2227 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
2228 pattern could run off the end of the subject. For example, the pattern
2229 "(?s)(.{1,5})"8 did this with the subject "ab".
2230
2231 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
2232 PCRE_CASELESS was set when matching characters that were quantified with ?
2233 or *.
2234
2235 17. A character class other than a single negated character that had a minimum
2236 but no maximum quantifier - for example [ab]{6,} - was not handled
2237 correctly by pce_dfa_exec(). It would match only one character.
2238
2239 18. A valid (though odd) pattern that looked like a POSIX character
2240 class but used an invalid character after [ (for example [[,abc,]]) caused
2241 pcre_compile() to give the error "Failed: internal error: code overflow" or
2242 in some cases to crash with a glibc free() error. This could even happen if
2243 the pattern terminated after [[ but there just happened to be a sequence of
2244 letters, a binary zero, and a closing ] in the memory that followed.
2245
2246 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
2247 over the years. Originally (before any Unicode support), just the bottom 8
2248 bits were taken. Thus, for example, \500 really meant \100. Nowadays the
2249 output from "man perlunicode" includes this:
2250
2251 The regular expression compiler produces polymorphic opcodes. That
2252 is, the pattern adapts to the data and automatically switches to
2253 the Unicode character scheme when presented with Unicode data--or
2254 instead uses a traditional byte scheme when presented with byte
2255 data.
2256
2257 Sadly, a wide octal escape does not cause a switch, and in a string with
2258 no other multibyte characters, these octal escapes are treated as before.
2259 Thus, in Perl, the pattern /\500/ actually matches \100 but the pattern
2260 /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
2261 Unicode string.
2262
2263 I have not perpetrated such confusion in PCRE. Up till now, it took just
2264 the bottom 8 bits, as in old Perl. I have now made octal escapes with
2265 values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
2266 translate to the appropriate multibyte character.
2267
2268 29. Applied some refactoring to reduce the number of warnings from Microsoft
2269 and Borland compilers. This has included removing the fudge introduced
2270 seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
2271 a warning about an unused variable.
2272
2273 21. PCRE has not included VT (character 0x0b) in the set of whitespace
2274 characters since release 4.0, because Perl (from release 5.004) does not.
2275 [Or at least, is documented not to: some releases seem to be in conflict
2276 with the documentation.] However, when a pattern was studied with
2277 pcre_study() and all its branches started with \s, PCRE still included VT
2278 as a possible starting character. Of course, this did no harm; it just
2279 caused an unnecessary match attempt.
2280
2281 22. Removed a now-redundant internal flag bit that recorded the fact that case
2282 dependency changed within the pattern. This was once needed for "required
2283 byte" processing, but is no longer used. This recovers a now-scarce options
2284 bit. Also moved the least significant internal flag bit to the most-
2285 significant bit of the word, which was not previously used (hangover from
2286 the days when it was an int rather than a uint) to free up another bit for
2287 the future.
2288
2289 23. Added support for CRLF line endings as well as CR and LF. As well as the
2290 default being selectable at build time, it can now be changed at runtime
2291 via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
2292 specify that it is scanning data with non-default line endings.
2293
2294 24. Changed the definition of CXXLINK to make it agree with the definition of
2295 LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
2296
2297 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
2298 recursions. This makes a big different to stack usage for some patterns.
2299
2300 26. If a subpattern containing a named recursion or subroutine reference such
2301 as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
2302 the space required for the compiled pattern went wrong and gave too small a
2303 value. Depending on the environment, this could lead to "Failed: internal
2304 error: code overflow at offset 49" or "glibc detected double free or
2305 corruption" errors.
2306
2307 27. Applied patches from Google (a) to support the new newline modes and (b) to
2308 advance over multibyte UTF-8 characters in GlobalReplace.
2309
2310 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
2311 difference for some implementation of PCRE in some Windows version.
2312
2313 29. Added some extra testing facilities to pcretest:
2314
2315 \q<number> in a data line sets the "match limit" value
2316 \Q<number> in a data line sets the "match recursion limt" value
2317 -S <number> sets the stack size, where <number> is in megabytes
2318
2319 The -S option isn't available for Windows.
2320
2321
2322 Version 6.6 06-Feb-06
2323 ---------------------
2324
2325 1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
2326 in pcreposix.h. I have copied the definition from pcre.h.
2327
2328 2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
2329 because pcre.h is no longer a built file.
2330
2331 3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
2332 not normally included in the compiled code.
2333
2334
2335 Version 6.5 01-Feb-06
2336 ---------------------
2337
2338 1. When using the partial match feature with pcre_dfa_exec(), it was not
2339 anchoring the second and subsequent partial matches at the new starting
2340 point. This could lead to incorrect results. For example, with the pattern
2341 /1234/, partially matching against "123" and then "a4" gave a match.
2342
2343 2. Changes to pcregrep:
2344
2345 (a) All non-match returns from pcre_exec() were being treated as failures
2346 to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
2347 error message is output. Some extra information is given for the
2348 PCRE_ERROR_MATCHLIMIT and PCRE_ERROR_RECURSIONLIMIT errors, which are
2349 probably the only errors that are likely to be caused by users (by
2350 specifying a regex that has nested indefinite repeats, for instance).
2351 If there are more than 20 of these errors, pcregrep is abandoned.
2352
2353 (b) A binary zero was treated as data while matching, but terminated the
2354 output line if it was written out. This has been fixed: binary zeroes
2355 are now no different to any other data bytes.
2356
2357 (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
2358 used to set a locale for matching. The --locale=xxxx long option has
2359 been added (no short equivalent) to specify a locale explicitly on the
2360 pcregrep command, overriding the environment variables.
2361
2362 (d) When -B was used with -n, some line numbers in the output were one less
2363 than they should have been.
2364
2365 (e) Added the -o (--only-matching) option.
2366
2367 (f) If -A or -C was used with -c (count only), some lines of context were
2368 accidentally printed for the final match.
2369
2370 (g) Added the -H (--with-filename) option.
2371
2372 (h) The combination of options -rh failed to suppress file names for files
2373 that were found from directory arguments.
2374
2375 (i) Added the -D (--devices) and -d (--directories) options.
2376
2377 (j) Added the -F (--fixed-strings) option.
2378
2379 (k) Allow "-" to be used as a file name for -f as well as for a data file.
2380
2381 (l) Added the --colo(u)r option.
2382
2383 (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
2384 is not present by default.
2385
2386 3. A nasty bug was discovered in the handling of recursive patterns, that is,
2387 items such as (?R) or (?1), when the recursion could match a number of
2388 alternatives. If it matched one of the alternatives, but subsequently,
2389 outside the recursion, there was a failure, the code tried to back up into
2390 the recursion. However, because of the way PCRE is implemented, this is not
2391 possible, and the result was an incorrect result from the match.
2392
2393 In order to prevent this happening, the specification of recursion has
2394 been changed so that all such subpatterns are automatically treated as
2395 atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
2396
2397 4. I had overlooked the fact that, in some locales, there are characters for
2398 which isalpha() is true but neither isupper() nor islower() are true. In
2399 the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
2400 and ordfeminine) are like this. This affected the treatment of \w and \W
2401 when they appeared in character classes, but not when they appeared outside
2402 a character class. The bit map for "word" characters is now created
2403 separately from the results of isalnum() instead of just taking it from the
2404 upper, lower, and digit maps. (Plus the underscore character, of course.)
2405
2406 5. The above bug also affected the handling of POSIX character classes such as
2407 [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
2408 permanent tables. Instead, the bit maps for such a class were previously
2409 created as the appropriate unions of the upper, lower, and digit bitmaps.
2410 Now they are created by subtraction from the [[:word:]] class, which has
2411 its own bitmap.
2412
2413 6. The [[:blank:]] character class matches horizontal, but not vertical space.
2414 It is created by subtracting the vertical space characters (\x09, \x0a,
2415 \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
2416 subtraction was done in the overall bitmap for a character class, meaning
2417 that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
2418 be recognized. This bug has been fixed.
2419
2420 7. Patches from the folks at Google:
2421
2422 (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
2423 real life, but is still worth protecting against".
2424
2425 (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
2426 regular expressions".
2427
2428 (c) pcre_scanner.cc: avoid use of std::count() because not all systems
2429 have it.
2430
2431 (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
2432 "configure" and the latter not, in order to fix a problem somebody had
2433 with compiling the Arg class on HP-UX.
2434
2435 (e) Improve the error-handling of the C++ wrapper a little bit.
2436
2437 (f) New tests for checking recursion limiting.
2438
2439 8. The pcre_memmove() function, which is used only if the environment does not
2440 have a standard memmove() function (and is therefore rarely compiled),
2441 contained two bugs: (a) use of int instead of size_t, and (b) it was not
2442 returning a result (though PCRE never actually uses the result).
2443
2444 9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
2445 large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
2446 returned instead of calling malloc() with an overflowing number that would
2447 most likely cause subsequent chaos.
2448
2449 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
2450
2451 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
2452 with this option is matched, the nmatch and pmatch options of regexec() are
2453 ignored.
2454
2455 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
2456 provided in case anyone wants to the the POSIX interface with UTF-8
2457 strings.
2458
2459 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
2460 C++ linking (needed for some HP-UX environments).
2461
2462 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
2463 (unused parameter) and in the pcre_printint() function (omitted "default"
2464 switch label when the default is to do nothing).
2465
2466 15. Added some code to make it possible, when PCRE is compiled as a C++
2467 library, to replace subject pointers for pcre_exec() with a smart pointer
2468 class, thus making it possible to process discontinuous strings.
2469
2470 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
2471 much the same function. They were added by different people who were trying
2472 to make PCRE easy to compile on non-Unix systems. It has been suggested
2473 that PCRE_EXPORT be abolished now that there is more automatic apparatus
2474 for compiling on Windows systems. I have therefore replaced it with
2475 PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
2476 defaults to "extern" for C or "extern C" for C++, which works fine on
2477 Unix-like systems. It is now possible to override the value of PCRE_DATA_
2478 SCOPE with something explicit in config.h. In addition:
2479
2480 (a) pcreposix.h still had just "extern" instead of either of these macros;
2481 I have replaced it with PCRE_DATA_SCOPE.
2482
2483 (b) Functions such as _pcre_xclass(), which are internal to the library,
2484 but external in the C sense, all had PCRE_EXPORT in their definitions.
2485 This is apparently wrong for the Windows case, so I have removed it.
2486 (It makes no difference on Unix-like systems.)
2487
2488 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
2489 of recursive calls to match(). This is different to MATCH_LIMIT because
2490 that limits the total number of calls to match(), not all of which increase
2491 the depth of recursion. Limiting the recursion depth limits the amount of
2492 stack (or heap if NO_RECURSE is set) that is used. The default can be set
2493 when PCRE is compiled, and changed at run time. A patch from Google adds
2494 this functionality to the C++ interface.
2495
2496 18. Changes to the handling of Unicode character properties:
2497
2498 (a) Updated the table to Unicode 4.1.0.
2499
2500 (b) Recognize characters that are not in the table as "Cn" (undefined).
2501
2502 (c) I revised the way the table is implemented to a much improved format
2503 which includes recognition of ranges. It now supports the ranges that
2504 are defined in UnicodeData.txt, and it also amalgamates other
2505 characters into ranges. This has reduced the number of entries in the
2506 table from around 16,000 to around 3,000, thus reducing its size
2507 considerably. I realized I did not need to use a tree structure after
2508 all - a binary chop search is just as efficient. Having reduced the
2509 number of entries, I extended their size from 6 bytes to 8 bytes to
2510 allow for more data.
2511
2512 (d) Added support for Unicode script names via properties such as \p{Han}.
2513
2514 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
2515 matching that character.
2516
2517 20. When matching a repeated Unicode property with a minimum greater than zero,
2518 (for example \pL{2,}), PCRE could look past the end of the subject if it
2519 reached it while seeking the minimum number of characters. This could
2520 happen only if some of the characters were more than one byte long, because
2521 there is a check for at least the minimum number of bytes.
2522
2523 21. Refactored the implementation of \p and \P so as to be more general, to
2524 allow for more different types of property in future. This has changed the
2525 compiled form incompatibly. Anybody with saved compiled patterns that use
2526 \p or \P will have to recompile them.
2527
2528 22. Added "Any" and "L&" to the supported property types.
2529
2530 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
2531 but give a compile time error if the value is greater than 0xff.
2532
2533 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
2534 accidentally not being installed or uninstalled.
2535
2536 25. The pcre.h file was built from pcre.h.in, but the only changes that were
2537 made were to insert the current release number. This seemed silly, because
2538 it made things harder for people building PCRE on systems that don't run
2539 "configure". I have turned pcre.h into a distributed file, no longer built
2540 by "configure", with the version identification directly included. There is
2541 no longer a pcre.h.in file.
2542
2543 However, this change necessitated a change to the pcre-config script as
2544 well. It is built from pcre-config.in, and one of the substitutions was the
2545 release number. I have updated configure.ac so that ./configure now finds
2546 the release number by grepping pcre.h.
2547
2548 26. Added the ability to run the tests under valgrind.
2549
2550
2551 Version 6.4 05-Sep-05
2552 ---------------------
2553
2554 1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
2555 "--" to be printed when multiple files were scanned, even when none of the
2556 -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
2557 consider it to be a bug, and have restored the previous behaviour.
2558
2559 2. A couple of code tidies to get rid of compiler warnings.
2560
2561 3. The pcretest program used to cheat by referring to symbols in the library
2562 whose names begin with _pcre_. These are internal symbols that are not
2563 really supposed to be visible externally, and in some environments it is
2564 possible to suppress them. The cheating is now confined to including
2565 certain files from the library's source, which is a bit cleaner.
2566
2567 4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
2568 file's purpose clearer.
2569
2570 5. Reorganized pcre_ucp_findchar().
2571
2572
2573 Version 6.3 15-Aug-05
2574 ---------------------
2575
2576 1. The file libpcre.pc.in did not have general read permission in the tarball.
2577
2578 2. There were some problems when building without C++ support:
2579
2580 (a) If C++ support was not built, "make install" and "make test" still
2581 tried to test it.
2582
2583 (b) There were problems when the value of CXX was explicitly set. Some
2584 changes have been made to try to fix these, and ...
2585
2586 (c) --disable-cpp can now be used to explicitly disable C++ support.
2587
2588 (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
2589 backslash in a target when C++ was disabled. This confuses some
2590 versions of "make", apparently. Using an intermediate variable solves
2591 this. (Same for CPP_LOBJ.)
2592
2593 3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
2594 (non-Windows) now includes $(CFLAGS) because these flags are sometimes
2595 necessary on certain architectures.
2596
2597 4. Added a setting of -export-symbols-regex to the link command to remove
2598 those symbols that are exported in the C sense, but actually are local
2599 within the library, and not documented. Their names all begin with
2600 "_pcre_". This is not a perfect job, because (a) we have to except some
2601 symbols that pcretest ("illegally") uses, and (b) the facility isn't always
2602 available (and never for static libraries). I have made a note to try to
2603 find a way round (a) in the future.
2604
2605
2606 Version 6.2 01-Aug-05
2607 ---------------------
2608
2609 1. There was no test for integer overflow of quantifier values. A construction
2610 such as {1111111111111111} would give undefined results. What is worse, if
2611 a minimum quantifier for a parenthesized subpattern overflowed and became
2612 negative, the calculation of the memory size went wrong. This could have
2613 led to memory overwriting.
2614
2615 2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
2616
2617 3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
2618 operating environments where this matters.
2619
2620 4. Applied Giuseppe Maxia's patch to add additional features for controlling
2621 PCRE options from within the C++ wrapper.
2622
2623 5. Named capturing subpatterns were not being correctly counted when a pattern
2624 was compiled. This caused two problems: (a) If there were more than 100
2625 such subpatterns, the calculation of the memory needed for the whole
2626 compiled pattern went wrong, leading to an overflow error. (b) Numerical
2627 back references of the form \12, where the number was greater than 9, were
2628 not recognized as back references, even though there were sufficient
2629 previous subpatterns.
2630
2631 6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
2632 versions of gcc, e.g. 2.95.4.
2633
2634
2635 Version 6.1 21-Jun-05
2636 ---------------------
2637
2638 1. There was one reference to the variable "posix" in pcretest.c that was not
2639 surrounded by "#if !defined NOPOSIX".
2640
2641 2. Make it possible to compile pcretest without DFA support, UTF8 support, or
2642 the cross-check on the old pcre_info() function, for the benefit of the
2643 cut-down version of PCRE that is currently imported into Exim.
2644
2645 3. A (silly) pattern starting with (?i)(?-i) caused an internal space
2646 allocation error. I've done the easy fix, which wastes 2 bytes for sensible
2647 patterns that start (?i) but I don't think that matters. The use of (?i) is
2648 just an example; this all applies to the other options as well.
2649
2650 4. Since libtool seems to echo the compile commands it is issuing, the output
2651 from "make" can be reduced a bit by putting "@" in front of each libtool
2652 compile command.
2653
2654 5. Patch from the folks at Google for configure.in to be a bit more thorough
2655 in checking for a suitable C++ installation before trying to compile the
2656 C++ stuff. This should fix a reported problem when a compiler was present,
2657 but no suitable headers.
2658
2659 6. The man pages all had just "PCRE" as their title. I have changed them to
2660 be the relevant file name. I have also arranged that these names are
2661 retained in the file doc/pcre.txt, which is a concatenation in text format
2662 of all the man pages except the little individual ones for each function.
2663
2664 7. The NON-UNIX-USE file had not been updated for the different set of source
2665 files that come with release 6. I also added a few comments about the C++
2666 wrapper.
2667
2668
2669 Version 6.0 07-Jun-05
2670 ---------------------
2671
2672 1. Some minor internal re-organization to help with my DFA experiments.
2673
2674 2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
2675 didn't matter for the library itself when fully configured, but did matter
2676 when compiling without UCP support, or within Exim, where the ucp files are
2677 not imported.
2678
2679 3. Refactoring of the library code to split up the various functions into
2680 different source modules. The addition of the new DFA matching code (see
2681 below) to a single monolithic source would have made it really too
2682 unwieldy, quite apart from causing all the code to be include in a
2683 statically linked application, when only some functions are used. This is
2684 relevant even without the DFA addition now that patterns can be compiled in
2685 one application and matched in another.
2686
2687 The downside of splitting up is that there have to be some external
2688 functions and data tables that are used internally in different modules of
2689 the library but which are not part of the API. These have all had their
2690 names changed to start with "_pcre_" so that they are unlikely to clash
2691 with other external names.
2692
2693 4. Added an alternate matching function, pcre_dfa_exec(), which matches using
2694 a different (DFA) algorithm. Although it is slower than the original
2695 function, it does have some advantages for certain types of matching
2696 problem.
2697
2698 5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
2699 including restarting after a partial match.
2700
2701 6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
2702 defined when compiling for Windows was sent to me. I have put it into the
2703 code, though I have no means of testing or verifying it.
2704
2705 7. Added the pcre_refcount() auxiliary function.
2706
2707 8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
2708 match before or at the first newline in the subject string. In pcretest,
2709 the /f option on a pattern can be used to set this.
2710
2711 9. A repeated \w when used in UTF-8 mode with characters greater than 256
2712 would behave wrongly. This has been present in PCRE since release 4.0.
2713
2714 10. A number of changes to the pcregrep command:
2715
2716 (a) Refactored how -x works; insert ^(...)$ instead of setting
2717 PCRE_ANCHORED and checking the length, in preparation for adding
2718 something similar for -w.
2719
2720 (b) Added the -w (match as a word) option.
2721
2722 (c) Refactored the way lines are read and buffered so as to have more
2723 than one at a time available.
2724
2725 (d) Implemented a pcregrep test script.
2726
2727 (e) Added the -M (multiline match) option. This allows patterns to match
2728 over several lines of the subject. The buffering ensures that at least
2729 8K, or the rest of the document (whichever is the shorter) is available
2730 for matching (and similarly the previous 8K for lookbehind assertions).
2731
2732 (f) Changed the --help output so that it now says
2733
2734 -w, --word-regex(p)
2735
2736 instead of two lines, one with "regex" and the other with "regexp"
2737 because that confused at least one person since the short forms are the
2738 same. (This required a bit of code, as the output is generated
2739 automatically from a table. It wasn't just a text change.)
2740
2741 (g) -- can be used to terminate pcregrep options if the next thing isn't an
2742 option but starts with a hyphen. Could be a pattern or a path name
2743 starting with a hyphen, for instance.
2744
2745 (h) "-" can be given as a file name to represent stdin.
2746
2747 (i) When file names are being printed, "(standard input)" is used for
2748 the standard input, for compatibility with GNU grep. Previously
2749 "<stdin>" was used.
2750
2751 (j) The option --label=xxx can be used to supply a name to be used for
2752 stdin when file names are being printed. There is no short form.
2753
2754 (k) Re-factored the options decoding logic because we are going to add
2755 two more options that take data. Such options can now be given in four
2756 different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
2757
2758 (l) Added the -A, -B, and -C options for requesting that lines of context
2759 around matches be printed.
2760
2761 (m) Added the -L option to print the names of files that do not contain
2762 any matching lines, that is, the complement of -l.
2763
2764 (n) The return code is 2 if any file cannot be opened, but pcregrep does
2765 continue to scan other files.
2766
2767 (o) The -s option was incorrectly implemented. For compatibility with other
2768 greps, it now suppresses the error message for a non-existent or non-
2769 accessible file (but not the return code). There is a new option called
2770 -q that suppresses the output of matching lines, which was what -s was
2771 previously doing.
2772
2773 (p) Added --include and --exclude options to specify files for inclusion
2774 and exclusion when recursing.
2775
2776 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
2777 Hopefully, it now does.
2778
2779 12. Missing cast in pcre_study().
2780
2781 13. Added an "uninstall" target to the makefile.
2782
2783 14. Replaced "extern" in the function prototypes in Makefile.in with
2784 "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
2785 world, but is set differently for Windows.
2786
2787 15. Added a second compiling function called pcre_compile2(). The only
2788 difference is that it has an extra argument, which is a pointer to an
2789 integer error code. When there is a compile-time failure, this is set
2790 non-zero, in addition to the error test pointer being set to point to an
2791 error message. The new argument may be NULL if no error number is required
2792 (but then you may as well call pcre_compile(), which is now just a
2793 wrapper). This facility is provided because some applications need a
2794 numeric error indication, but it has also enabled me to tidy up the way
2795 compile-time errors are handled in the POSIX wrapper.
2796
2797 16. Added VPATH=.libs to the makefile; this should help when building with one
2798 prefix path and installing with another. (Or so I'm told by someone who
2799 knows more about this stuff than I do.)
2800
2801 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
2802 passes PCRE_DOTALL to the pcre_compile() function, making the "." character
2803 match everything, including newlines. This is not POSIX-compatible, but
2804 somebody wanted the feature. From pcretest it can be activated by using
2805 both the P and the s flags.
2806
2807 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
2808
2809 19. libpcre.pc was being incorrectly installed as executable.
2810
2811 20. A couple of places in pcretest check for end-of-line by looking for '\n';
2812 it now also looks for '\r' so that it will work unmodified on Windows.
2813
2814 21. Added Google's contributed C++ wrapper to the distribution.
2815
2816 22. Added some untidy missing memory free() calls in pcretest, to keep
2817 Electric Fence happy when testing.
2818
2819
2820
2821 Version 5.0 13-Sep-04
2822 ---------------------
2823
2824 1. Internal change: literal characters are no longer packed up into items
2825 containing multiple characters in a single byte-string. Each character
2826 is now matched using a separate opcode. However, there may be more than one
2827 byte in the character in UTF-8 mode.
2828
2829 2. The pcre_callout_block structure has two new fields: pattern_position and
2830 next_item_length. These contain the offset in the pattern to the next match
2831 item, and its length, respectively.
2832
2833 3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
2834 insertion of callouts before each pattern item. Added the /C option to
2835 pcretest to make use of this.
2836
2837 4. On the advice of a Windows user, the lines
2838
2839 #if defined(_WIN32) || defined(WIN32)
2840 _setmode( _fileno( stdout ), 0x8000 );
2841 #endif /* defined(_WIN32) || defined(WIN32) */
2842
2843 have been added to the source of pcretest. This apparently does useful
2844 magic in relation to line terminators.
2845
2846 5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
2847 for the benefit of those environments where the "b" makes a difference.
2848
2849 6. The icc compiler has the same options as gcc, but "configure" doesn't seem
2850 to know about it. I have put a hack into configure.in that adds in code
2851 to set GCC=yes if CC=icc. This seems to end up at a point in the
2852 generated configure script that is early enough to affect the setting of
2853 compiler options, which is what is needed, but I have no means of testing
2854 whether it really works. (The user who reported this had patched the
2855 generated configure script, which of course I cannot do.)
2856
2857 LATER: After change 22 below (new libtool files), the configure script
2858 seems to know about icc (and also ecc). Therefore, I have commented out
2859 this hack in configure.in.
2860
2861 7. Added support for pkg-config (2 patches were sent in).
2862
2863 8. Negated POSIX character classes that used a combination of internal tables
2864 were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
2865 [[:^ascii]]. Typically, they would match almost any characters. The other
2866 POSIX classes were not broken in this way.
2867
2868 9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
2869 to find the match, as PCRE was deluded into thinking that the match had to
2870 start at the start point or following a newline. The same bug applied to
2871 patterns with negative forward assertions or any backward assertions
2872 preceding ".*" at the start, unless the pattern required a fixed first
2873 character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
2874
2875 10. In UTF-8 mode, when moving forwards in the subject after a failed match
2876 starting at the last subject character, bytes beyond the end of the subject
2877 string were read.
2878
2879 11. Renamed the variable "class" as "classbits" to make life easier for C++
2880 users. (Previously there was a macro definition, but it apparently wasn't
2881 enough.)
2882
2883 12. Added the new field "tables" to the extra data so that tables can be passed
2884 in at exec time, or the internal tables can be re-selected. This allows
2885 a compiled regex to be saved and re-used at a later time by a different
2886 program that might have everything at different addresses.
2887
2888 13. Modified the pcre-config script so that, when run on Solaris, it shows a
2889 -R library as well as a -L library.
2890
2891 14. The debugging options of pcretest (-d on the command line or D on a
2892 pattern) showed incorrect output for anything following an extended class
2893 that contained multibyte characters and which was followed by a quantifier.
2894
2895 15. Added optional support for general category Unicode character properties
2896 via the \p, \P, and \X escapes. Unicode property support implies UTF-8
2897 support. It adds about 90K to the size of the library. The meanings of the
2898 inbuilt class escapes such as \d and \s have NOT been changed.
2899
2900 16. Updated pcredemo.c to include calls to free() to release the memory for the
2901 compiled pattern.
2902
2903 17. The generated file chartables.c was being created in the source directory
2904 instead of in the building directory. This caused the build to fail if the
2905 source directory was different from the building directory, and was
2906 read-only.
2907
2908 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
2909 file. No doubt somebody will tell me if they don't make sense... Also added
2910 Dan Mooney's comments about building on OpenVMS.
2911
2912 19. Added support for partial matching via the PCRE_PARTIAL option for
2913 pcre_exec() and the \P data escape in pcretest.
2914
2915 20. Extended pcretest with 3 new pattern features:
2916
2917 (i) A pattern option of the form ">rest-of-line" causes pcretest to
2918 write the compiled pattern to the file whose name is "rest-of-line".
2919 This is a straight binary dump of the data, with the saved pointer to
2920 the character tables forced to be NULL. The study data, if any, is
2921 written too. After writing, pcretest reads a new pattern.
2922
2923 (ii) If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
2924 compiled pattern from the given file. There must not be any
2925 occurrences of "<" in the file name (pretty unlikely); if there are,
2926 pcretest will instead treat the initial "<" as a pattern delimiter.
2927 After reading in the pattern, pcretest goes on to read data lines as
2928 usual.
2929
2930 (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
2931 and 16-bit fields in a compiled pattern, to simulate a pattern that
2932 was compiled on a host of opposite endianness.
2933
2934 21. The pcre-exec() function can now cope with patterns that were compiled on
2935 hosts of opposite endianness, with this restriction:
2936
2937 As for any compiled expression that is saved and used later, the tables
2938 pointer field cannot be preserved; the extra_data field in the arguments
2939 to pcre_exec() should be used to pass in a tables address if a value
2940 other than the default internal tables were used at compile time.
2941
2942 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
2943 now diagnosed as an error. Previously, most of the time, a negative number
2944 would have been treated as zero, but if in addition "ovector" was passed as
2945 NULL, a crash could occur.
2946
2947 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
2948 new versions from the libtool 1.5 distribution (the last one is a copy of
2949 a file called libtool.m4). This seems to have fixed the need to patch
2950 "configure" to support Darwin 1.3 (which I used to do). However, I still
2951 had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
2952 workstation).
2953
2954 24. Changed the PCRE licence to be the more standard "BSD" licence.
2955
2956
2957 Version 4.5 01-Dec-03
2958 ---------------------
2959
2960 1. There has been some re-arrangement of the code for the match() function so
2961 that it can be compiled in a version that does not call itself recursively.
2962 Instead, it keeps those local variables that need separate instances for
2963 each "recursion" in a frame on the heap, and gets/frees frames whenever it
2964 needs to "recurse". Keeping track of where control must go is done by means
2965 of setjmp/longjmp. The whole thing is implemented by a set of macros that
2966 hide most of the details from the main code, and operates only if
2967 NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
2968 "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
2969 operating.
2970
2971 To make it easier for callers to provide specially tailored get/free
2972 functions for this usage, two new functions, pcre_stack_malloc, and
2973 pcre_stack_free, are used. They are always called in strict stacking order,
2974 and the size of block requested is always the same.
2975
2976 The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
2977 PCRE has been compiled to use the stack or the heap for recursion. The
2978 -C option of pcretest uses this to show which version is compiled.
2979
2980 A new data escape \S, is added to pcretest; it causes the amounts of store
2981 obtained and freed by both kinds of malloc/free at match time to be added
2982 to the output.
2983
2984 2. Changed the locale test to use "fr_FR" instead of "fr" because that's
2985 what's available on my current Linux desktop machine.
2986
2987 3. When matching a UTF-8 string, the test for a valid string at the start has
2988 been extended. If start_offset is not zero, PCRE now checks that it points
2989 to a byte that is the start of a UTF-8 character. If not, it returns
2990 PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
2991 this is necessary because there may be backward assertions in the pattern.
2992 When matching the same subject several times, it may save resources to use
2993 PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
2994
2995 4. The code for checking the validity of UTF-8 strings has been tightened so
2996 that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
2997 containing "overlong sequences".
2998
2999 5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
3000 I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
3001 should have been "&", but it just so happened that all the cases this let
3002 through by mistake were picked up later in the function.
3003
3004 6. I had used a variable called "isblank" - this is a C99 function, causing
3005 some compilers to warn. To avoid this, I renamed it (as "blankclass").
3006
3007 7. Cosmetic: (a) only output another newline at the end of pcretest if it is
3008 prompting; (b) run "./pcretest /dev/null" at the start of the test script
3009 so the version is shown; (c) stop "make test" echoing "./RunTest".
3010
3011 8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
3012
3013 9. The prototype for memmove() for systems that don't have it was using
3014 size_t, but the inclusion of the header that defines size_t was later. I've
3015 moved the #includes for the C headers earlier to avoid this.
3016
3017 10. Added some adjustments to the code to make it easier to compiler on certain
3018 special systems:
3019
3020 (a) Some "const" qualifiers were missing.
3021 (b) Added the macro EXPORT before all exported functions; by default this
3022 is defined to be empty.
3023 (c) Changed the dftables auxiliary program (that builds chartables.c) so
3024 that it reads its output file name as an argument instead of writing
3025 to the standard output and assuming this can be redirected.
3026
3027 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
3028 class containing characters with values greater than 255, PCRE compilation
3029 went into a loop.
3030
3031 12. A recursive reference to a subpattern that was within another subpattern
3032 that had a minimum quantifier of zero caused PCRE to crash. For example,
3033 (x(y(?2))z)? provoked this bug with a subject that got as far as the
3034 recursion. If the recursively-called subpattern itself had a zero repeat,
3035 that was OK.
3036
3037 13. In pcretest, the buffer for reading a data line was set at 30K, but the
3038 buffer into which it was copied (for escape processing) was still set at
3039 1024, so long lines caused crashes.
3040
3041 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
3042 "internal error: code overflow...". This applied to any character class
3043 that was followed by a possessive quantifier.
3044
3045 15. Modified the Makefile to add libpcre.la as a prerequisite for
3046 libpcreposix.la because I was told this is needed for a parallel build to
3047 work.
3048
3049 16. If a pattern that contained .* following optional items at the start was
3050 studied, the wrong optimizing data was generated, leading to matching
3051 errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
3052 matching string must start with a or b or c. The correct conclusion for
3053 this pattern is that a match can start with any character.
3054
3055
3056 Version 4.4 13-Aug-03
3057 ---------------------
3058
3059 1. In UTF-8 mode, a character class containing characters with values between
3060 127 and 255 was not handled correctly if the compiled pattern was studied.
3061 In fixing this, I have also improved the studying algorithm for such
3062 classes (slightly).
3063
3064 2. Three internal functions had redundant arguments passed to them. Removal
3065 might give a very teeny performance improvement.
3066
3067 3. Documentation bug: the value of the capture_top field in a callout is *one
3068 more than* the number of the hightest numbered captured substring.
3069
3070 4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
3071 in incorrectly linking with a previously installed version. They now link
3072 explicitly with libpcre.la.
3073
3074 5. configure.in no longer needs to recognize Cygwin specially.
3075
3076 6. A problem in pcre.in for Windows platforms is fixed.
3077
3078 7. If a pattern was successfully studied, and the -d (or /D) flag was given to
3079 pcretest, it used to include the size of the study block as part of its
3080 output. Unfortunately, the structure contains a field that has a different
3081 size on different hardware architectures. This meant that the tests that
3082 showed this size failed. As the block is currently always of a fixed size,
3083 this information isn't actually particularly useful in pcretest output, so
3084 I have just removed it.
3085
3086 8. Three pre-processor statements accidentally did not start in column 1.
3087 Sadly, there are *still* compilers around that complain, even though
3088 standard C has not required this for well over a decade. Sigh.
3089
3090 9. In pcretest, the code for checking callouts passed small integers in the
3091 callout_data field, which is a void * field. However, some picky compilers
3092 complained about the casts involved for this on 64-bit systems. Now
3093 pcretest passes the address of the small integer instead, which should get
3094 rid of the warnings.
3095
3096 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
3097 both compile and run time, and gives an error if an invalid UTF-8 sequence
3098 is found. There is a option for disabling this check in cases where the
3099 string is known to be correct and/or the maximum performance is wanted.
3100
3101 11. In response to a bug report, I changed one line in Makefile.in from
3102
3103 -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
3104 to
3105 -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
3106
3107 to look similar to other lines, but I have no way of telling whether this
3108 is the right thing to do, as I do not use Windows. No doubt I'll get told
3109 if it's wrong...
3110
3111
3112 Version 4.3 21-May-03
3113 ---------------------
3114
3115 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
3116 Makefile.
3117
3118 2. Some refactoring to improve the quality of the code:
3119
3120 (i) The utf8_table... variables are now declared "const".
3121
3122 (ii) The code for \cx, which used the "case flipping" table to upper case
3123 lower case letters, now just substracts 32. This is ASCII-specific,
3124 but the whole concept of \cx is ASCII-specific, so it seems
3125 reasonable.
3126
3127 (iii) PCRE was using its character types table to recognize decimal and
3128 hexadecimal digits in the pattern. This is silly, because it handles
3129 only 0-9, a-f, and A-F, but the character types table is locale-
3130 specific, which means strange things might happen. A private
3131 table is now used for this - though it costs 256 bytes, a table is
3132 much faster than multiple explicit tests. Of course, the standard
3133 character types table is still used for matching digits in subject
3134 strings against \d.
3135
3136 (iv) Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
3137 ending in _t are). So I've renamed it as ESC_tee.
3138
3139 3. The first argument for regexec() in the POSIX wrapper should have been
3140 defined as "const".
3141
3142 4. Changed pcretest to use malloc() for its buffers so that they can be
3143 Electric Fenced for debugging.
3144
3145 5. There were several places in the code where, in UTF-8 mode, PCRE would try
3146 to read one or more bytes before the start of the subject string. Often this
3147 had no effect on PCRE's behaviour, but in some circumstances it could
3148 provoke a segmentation fault.
3149
3150 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
3151 to try to read one or more bytes before the start of the subject string.
3152
3153 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
3154 UTF-8 support could misbehave in various ways if the subject string
3155 contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
3156 area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
3157 back over UTF-8 characters.)
3158
3159
3160 Version 4.2 14-Apr-03
3161 ---------------------
3162
3163 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
3164
3165 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
3166 [ON_WINDOWS]: new variable, "#" on non-Windows platforms
3167 [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
3168 [WIN_PREFIX]: new variable, "cyg" for Cygwin
3169 * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
3170 and BUILD_EXEEXT
3171 Note: automatic setting of the BUILD variables is not yet working
3172 set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
3173 compile-time but not at link-time
3174 [LINK]: use for linking executables only
3175 make different versions for Windows and non-Windows
3176 [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
3177 libraries
3178 [LINK_FOR_BUILD]: new variable
3179 [OBJEXT]: use throughout
3180 [EXEEXT]: use throughout
3181 <winshared>: new target
3182 <wininstall>: new target
3183 <dftables.o>: use native compiler
3184 <dftables>: use native linker
3185 <install>: handle Windows platform correctly
3186 <clean>: ditto
3187 <check>: ditto
3188 copy DLL to top builddir before testing
3189
3190 As part of these changes, -no-undefined was removed again. This was reported
3191 to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
3192 in any case.
3193
3194 3. Some tidies to get rid of compiler warnings:
3195
3196 . In the match_data structure, match_limit was an unsigned long int, whereas
3197 match_call_count was an int. I've made them both unsigned long ints.
3198
3199 . In pcretest the fact that a const uschar * doesn't automatically cast to
3200 a void * provoked a warning.
3201
3202 . Turning on some more compiler warnings threw up some "shadow" variables
3203 and a few more missing casts.
3204
3205 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
3206 option, a class that contained a single character with a value between 128
3207 and 255 (e.g. /[\xFF]/) caused PCRE to crash.
3208
3209 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
3210 option, a class that contained several characters, but with at least one
3211 whose value was between 128 and 255 caused PCRE to crash.
3212
3213
3214 Version 4.1 12-Mar-03
3215 ---------------------
3216
3217 1. Compiling with gcc -pedantic found a couple of places where casts were
3218 needed, and a string in dftables.c that was longer than standard compilers are
3219 required to support.
3220
3221 2. Compiling with Sun's compiler found a few more places where the code could
3222 be tidied up in order to avoid warnings.
3223
3224 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
3225 first of these names is deprecated in the latest Autoconf in favour of the name
3226 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
3227 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
3228 analogy I have changed it to CFLAGS_FOR_BUILD.
3229
3230 4. Added -no-undefined to the linking command in the Makefile, because this is
3231 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
3232 linking step for the pcreposix library.
3233
3234 5. PCRE was failing to diagnose the case of two named groups with the same
3235 name.
3236
3237 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
3238 literal character that is needed in the subject for a match, and scans along to
3239 ensure that it is present before embarking on the full matching process. This
3240 saves time in cases of nested unlimited repeats that are never going to match.
3241 Problem: the scan can take a lot of time if the subject is very long (e.g.
3242 megabytes), thus penalizing straightforward matches. It is now done only if the
3243 amount of subject to be scanned is less than 1000 bytes.
3244
3245 7. A lesser problem with the same optimization is that it was recording the
3246 first character of an anchored pattern as "needed", thus provoking a search
3247 right along the subject, even when the first match of the pattern was going to
3248 fail. The "needed" character is now not set for anchored patterns, unless it
3249 follows something in the pattern that is of non-fixed length. Thus, it still
3250 fulfils its original purpose of finding quick non-matches in cases of nested
3251 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
3252
3253
3254 Version 4.0 17-Feb-03
3255 ---------------------
3256
3257 1. If a comment in an extended regex that started immediately after a meta-item
3258 extended to the end of string, PCRE compiled incorrect data. This could lead to
3259 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
3260
3261 2. Moved to autoconf 2.53 and libtool 1.4.2.
3262
3263 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
3264 the special perltest8 script is no longer needed - all the tests can be run
3265 from a single perltest script.
3266
3267 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
3268 by \s. It has now been removed in PCRE. This means it isn't recognized as
3269 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
3270 class [:space:] *does* include VT, thereby creating a mess.
3271
3272 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
3273 space and tab.
3274
3275 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
3276 its new features into the main test script, reducing the number of scripts.
3277
3278 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
3279 were backward compatible, and made the (?i) apply to the whole pattern, as if
3280 /i were given. Now it behaves more logically, and applies the option setting
3281 only to what follows. PCRE has been changed to follow suit. However, if it
3282 finds options settings right at the start of the pattern, it extracts them into
3283 the global options, as before. Thus, they show up in the info data.
3284
3285 8. Added support for the \Q...\E escape sequence. Characters in between are
3286 treated as literals. This is slightly different from Perl in that $ and @ are
3287 also handled as literals inside the quotes. In Perl, they will cause variable
3288 interpolation. Note the following examples:
3289
3290 Pattern PCRE matches Perl matches
3291
3292 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
3293 \Qabc\$xyz\E abc\$xyz abc\$xyz
3294 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
3295
3296 For compatibility with Perl, \Q...\E sequences are recognized inside character
3297 classes as well as outside them.
3298
3299 9. Re-organized 3 code statements in pcretest to avoid "overflow in
3300 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
3301 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
3302 signed/unsigned warnings.
3303
3304 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
3305 option for pcretest, so I've replaced it by a simple function that does just
3306 that job.
3307
3308 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
3309 "pcregrep -".
3310
3311 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
3312 Java package. This provides some syntactic sugar for simple cases of what my
3313 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
3314 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
3315 item, you can use this simplified notation. Note that only makes sense with
3316 greedy quantifiers. Consequently, the use of the possessive quantifier forces
3317 greediness, whatever the setting of the PCRE_UNGREEDY option.
3318
3319 13. A change of greediness default within a pattern was not taking effect at
3320 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
3321 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
3322 was abstracted outside.
3323
3324 14. PCRE now supports the \G assertion. It is true when the current matching
3325 position is at the start point of the match. This differs from \A when the
3326 starting offset is non-zero. Used with the /g option of pcretest (or similar
3327 code), it works in the same way as it does for Perl's /g option. If all
3328 alternatives of a regex begin with \G, the expression is anchored to the start
3329 match position, and the "anchored" flag is set in the compiled expression.
3330
3331 15. Some bugs concerning the handling of certain option changes within patterns
3332 have been fixed. These applied to options other than (?ims). For example,
3333 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
3334 been the other way round. Some of this was related to change 7 above.
3335
3336 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
3337 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
3338 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
3339 POSIX classes only within a class (e.g. /[[:alpha:]]/).
3340
3341 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
3342 mode. Unlike ".", it always matches newline, whatever the setting of
3343 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
3344 assertions. Perl allows it, but it doesn't (in general) work because it can't
3345 calculate the length of the lookbehind. At least, that's the case for Perl
3346 5.8.0 - I've been told they are going to document that it doesn't work in
3347 future.
3348
3349 18. Added an error diagnosis for escapes that PCRE does not support: these are
3350 \L, \l, \N, \P, \p, \U, \u, and \X.
3351
3352 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
3353 reading past the end of the pattern in cases such as /[abcd/.
3354
3355 20. PCRE was getting more memory than necessary for patterns with classes that
3356 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
3357
3358 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
3359 compiling PCRE for use with Virtual Pascal.
3360
3361 22. Small fix to the Makefile to make it work properly if the build is done
3362 outside the source tree.
3363
3364 23. Added a new extension: a condition to go with recursion. If a conditional
3365 subpattern starts with (?(R) the "true" branch is used if recursion has
3366 happened, whereas the "false" branch is used only at the top level.
3367
3368 24. When there was a very long string of literal characters (over 255 bytes
3369 without UTF support, over 250 bytes with UTF support), the computation of how
3370 much memory was required could be incorrect, leading to segfaults or other
3371 strange effects.
3372
3373 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
3374 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
3375 there was a subsequent back reference to those brackets. This meant that, for
3376 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
3377 possible to check for precisely this case. All we can do is abandon the
3378 optimization if .* occurs inside capturing brackets when there are any back
3379 references whatsoever. (See below for a better fix that came later.)
3380
3381 26. The handling of the optimization for finding the first character of a
3382 non-anchored pattern, and for finding a character that is required later in the
3383 match were failing in some cases. This didn't break the matching; it just
3384 failed to optimize when it could. The way this is done has been re-implemented.
3385
3386 27. Fixed typo in error message for invalid (?R item (it said "(?p").
3387
3388 28. Added a new feature that provides some of the functionality that Perl
3389 provides with (?{...}). The facility is termed a "callout". The way it is done
3390 in PCRE is for the caller to provide an optional function, by setting
3391 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
3392 global variable. By default it is unset, which disables all calling out. To get
3393 the function called, the regex must include (?C) at appropriate points. This
3394 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
3395 This provides a means of identifying different callout points. When PCRE
3396 reaches such a point in the regex, if pcre_callout has been set, the external
3397 function is called. It is provided with data in a structure called
3398 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
3399 matching continues; if it returns a non-zero value, the match at the current
3400 point fails. However, backtracking will occur if possible. [This was changed
3401 later and other features added - see item 49 below.]
3402
3403 29. pcretest is upgraded to test the callout functionality. It provides a
3404 callout function that displays information. By default, it shows the start of
3405 the match and the current position in the text. There are some new data escapes
3406 to vary what happens:
3407
3408 \C+ in addition, show current contents of captured substrings
3409 \C- do not supply a callout function
3410 \C!n return 1 when callout number n is reached
3411 \C!n!m return 1 when callout number n is reached for the mth time
3412
3413 30. If pcregrep was called with the -l option and just a single file name, it
3414 output "<stdin>" if a match was found, instead of the file name.
3415
3416 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
3417 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
3418 pcre_exec(). This saves a malloc/free per call. The default value of
3419 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
3420 when configuring.
3421
3422 32. The default maximum size of a compiled pattern is 64K. There have been a
3423 few cases of people hitting this limit. The code now uses macros to handle the
3424 storing of links as offsets within the compiled pattern. It defaults to 2-byte
3425 links, but this can be changed to 3 or 4 bytes by --with-link-size when
3426 configuring. Tests 2 and 5 work only with 2-byte links because they output
3427 debugging information about compiled patterns.
3428
3429 33. Internal code re-arrangements:
3430
3431 (a) Moved the debugging function for printing out a compiled regex into
3432 its own source file (printint.c) and used #include to pull it into
3433 pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
3434 separate copies.
3435
3436 (b) Defined the list of op-code names for debugging as a macro in
3437 internal.h so that it is next to the definition of the opcodes.
3438
3439 (c) Defined a table of op-code lengths for simpler skipping along compiled
3440 code. This is again a macro in internal.h so that it is next to the
3441 definition of the opcodes.
3442
3443 34. Added support for recursive calls to individual subpatterns, along the
3444 lines of Robin Houston's patch (but implemented somewhat differently).
3445
3446 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
3447 allow it to read and process whole directories in Win32. This code was
3448 contributed by Lionel Fourquaux; it has not been tested by me.
3449
3450 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
3451 used to name a group. Names consist of alphanumerics and underscores, and must
3452 be unique. Back references use the syntax (?P=name) and recursive calls use
3453 (?P>name) which is a PCRE extension to the Python extension. Groups still have
3454 numbers. The function pcre_fullinfo() can be used after compilation to extract
3455 a name/number map. There are three relevant calls:
3456
3457 PCRE_INFO_NAMEENTRYSIZE yields the size of each entry in the map
3458 PCRE_INFO_NAMECOUNT yields the number of entries
3459 PCRE_INFO_NAMETABLE yields a pointer to the map.
3460
3461 The map is a vector of fixed-size entries. The size of each entry depends on
3462 the length of the longest name used. The first two bytes of each entry are the
3463 group number, most significant byte first. There follows the corresponding
3464 name, zero terminated. The names are in alphabetical order.
3465
3466 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
3467 case instead of 255. Making it the same both with and without UTF-8 support
3468 means that the same test output works with both.
3469
3470 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
3471 calling malloc() with a zero argument.
3472
3473 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
3474 optimization. I've improved things by keeping a bitmap of backreferences with
3475 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
3476 fact referenced, the optimization can be applied. It is unlikely that a
3477 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
3478 the match to follow \n) will appear inside brackets with a number greater than
3479 31, but if it does, any back reference > 31 suppresses the optimization.
3480
3481 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
3482 of disabling numbered capturing parentheses. Any opening parenthesis that is
3483 not followed by ? behaves as if it were followed by ?: but named parentheses
3484 can still be used for capturing (and they will acquire numbers in the usual
3485 way).
3486
3487 41. Redesigned the return codes from the match() function into yes/no/error so
3488 that errors can be passed back from deep inside the nested calls. A malloc
3489 failure while inside a recursive subpattern call now causes the
3490 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
3491
3492 42. It is now possible to set a limit on the number of times the match()
3493 function is called in a call to pcre_exec(). This facility makes it possible to
3494 limit the amount of recursion and backtracking, though not in a directly
3495 obvious way, because the match() function is used in a number of different
3496 circumstances. The count starts from zero for each position in the subject
3497 string (for non-anchored patterns). The default limit is, for compatibility, a
3498 large number, namely 10 000 000. You can change this in two ways:
3499
3500 (a) When configuring PCRE before making, you can use --with-match-limit=n
3501 to set a default value for the compiled library.
3502
3503 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
3504 a different value is set. See 45 below.
3505
3506 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
3507
3508 43. Added a new function pcre_config(int, void *) to enable run-time extraction
3509 of things that can be changed at compile time. The first argument specifies
3510 what is wanted and the second points to where the information is to be placed.
3511 The current list of available information is:
3512
3513 PCRE_CONFIG_UTF8
3514
3515 The output is an integer that is set to one if UTF-8 support is available;
3516 otherwise it is set to zero.
3517
3518 PCRE_CONFIG_NEWLINE
3519
3520 The output is an integer that it set to the value of the code that is used for
3521 newline. It is either LF (10) or CR (13).
3522
3523 PCRE_CONFIG_LINK_SIZE
3524
3525 The output is an integer that contains the number of bytes used for internal
3526 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
3527
3528 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
3529
3530 The output is an integer that contains the threshold above which the POSIX
3531 interface uses malloc() for output vectors. See item 31 above.
3532
3533 PCRE_CONFIG_MATCH_LIMIT
3534
3535 The output is an unsigned integer that contains the default limit of the number
3536 of match() calls in a pcre_exec() execution. See 42 above.
3537
3538 44. pcretest has been upgraded by the addition of the -C option. This causes it
3539 to extract all the available output from the new pcre_config() function, and to
3540 output it. The program then exits immediately.
3541
3542 45. A need has arisen to pass over additional data with calls to pcre_exec() in
3543 order to support additional features. One way would have been to define
3544 pcre_exec2() (for example) with extra arguments, but this would not have been
3545 extensible, and would also have required all calls to the original function to
3546 be mapped to the new one. Instead, I have chosen to extend the mechanism that
3547 is used for passing in "extra" data from pcre_study().
3548
3549 The pcre_extra structure is now exposed and defined in pcre.h. It currently
3550 contains the following fields:
3551
3552 flags a bitmap indicating which of the following fields are set
3553 study_data opaque data from pcre_study()
3554 match_limit a way of specifying a limit on match() calls for a specific
3555 call to pcre_exec()
3556 callout_data data for callouts (see 49 below)
3557
3558 The flag bits are also defined in pcre.h, and are
3559
3560 PCRE_EXTRA_STUDY_DATA
3561 PCRE_EXTRA_MATCH_LIMIT
3562 PCRE_EXTRA_CALLOUT_DATA
3563
3564 The pcre_study() function now returns one of these new pcre_extra blocks, with
3565 the actual study data pointed to by the study_data field, and the
3566 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
3567 before. That is, this change is entirely upwards-compatible and requires no
3568 change to existing code.
3569
3570 If you want to pass in additional data to pcre_exec(), you can either place it
3571 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
3572 block.
3573
3574 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
3575 data string contains the escape sequence \M, pcretest calls pcre_exec() several
3576 times with different match limits, until it finds the minimum value needed for
3577 pcre_exec() to complete. The value is then output. This can be instructive; for
3578 most simple matches the number is quite small, but for pathological cases it
3579 gets very large very quickly.
3580
3581 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
3582 returns the size of the data block pointed to by the study_data field in a
3583 pcre_extra block, that is, the value that was passed as the argument to
3584 pcre_malloc() when PCRE was getting memory in which to place the information
3585 created by pcre_study(). The fourth argument should point to a size_t variable.
3586 pcretest has been extended so that this information is shown after a successful
3587 pcre_study() call when information about the compiled regex is being displayed.
3588
3589 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
3590 because what follows is always an absolute path. (Later: it turns out that this
3591 is more than cosmetic for MinGW, because it doesn't like empty path
3592 components.)
3593
3594 49. Some changes have been made to the callout feature (see 28 above):
3595
3596 (i) A callout function now has three choices for what it returns:
3597
3598 0 => success, carry on matching
3599 > 0 => failure at this point, but backtrack if possible
3600 < 0 => serious error, return this value from pcre_exec()
3601
3602 Negative values should normally be chosen from the set of PCRE_ERROR_xxx
3603 values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
3604 "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
3605 use by callout functions. It will never be used by PCRE itself.
3606
3607 (ii) The pcre_extra structure (see 45 above) has a void * field called
3608 callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
3609 pcre_callout_block structure has a field of the same name. The contents of
3610 the field passed in the pcre_extra structure are passed to the callout
3611 function in the corresponding field in the callout block. This makes it
3612 easier to use the same callout-containing regex from multiple threads. For
3613 testing, the pcretest program has a new data escape
3614
3615 \C*n pass the number n (may be negative) as callout_data
3616
3617 If the callout function in pcretest receives a non-zero value as
3618 callout_data, it returns that value.
3619
3620 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
3621 there were some redundant $(CFLAGS) in commands that are now specified as
3622 $(LINK), which already includes $(CFLAGS).
3623
3624 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
3625 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
3626 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
3627 one-byte characters throughout. Note that case-insensitive matching applies
3628 only to characters whose values are less than 256. PCRE doesn't support the
3629 notion of cases for higher-valued characters.
3630
3631 (i) A character class whose characters are all within 0-255 is handled as
3632 a bit map, and the map is inverted for negative classes. Previously, a
3633 character > 255 always failed to match such a class; however it should
3634 match if the class was a negative one (e.g. [^ab]). This has been fixed.
3635
3636 (ii) A negated character class with a single character < 255 is coded as
3637 "not this character" (OP_NOT). This wasn't working properly when the test
3638 character was multibyte, either singly or repeated.
3639
3640 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
3641 mode, for example: \x{100}{2,3}.
3642
3643 (iv) The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
3644 singly or repeated) now correctly test multibyte characters. However,
3645 PCRE doesn't recognize any characters with values greater than 255 as
3646 digits, spaces, or word characters. Such characters always match \D, \S,
3647 and \W, and never match \d, \s, or \w.
3648
3649 (v) Classes may now contain characters and character ranges with values
3650 greater than 255. For example: [ab\x{100}-\x{400}].
3651
3652 (vi) pcregrep now has a --utf-8 option (synonym -u) which makes it call
3653 PCRE in UTF-8 mode.
3654
3655 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
3656 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
3657 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
3658 value.)
3659
3660 53. The single man page has become too large. I have therefore split it up into
3661 a number of separate man pages. These also give rise to individual HTML pages;
3662 these are now put in a separate directory, and there is an index.html page that
3663 lists them all. Some hyperlinking between the pages has been installed.
3664
3665 54. Added convenience functions for handling named capturing parentheses.
3666
3667 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
3668 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
3669 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
3670 are faulted.
3671
3672 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
3673 calling configure. These values are used when compiling the dftables.c program
3674 which is run to generate the source of the default character tables. They
3675 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
3676 you will need to set these values.
3677
3678 57. Updated the building process for Windows DLL, as provided by Fred Cox.
3679
3680
3681 Version 3.9 02-Jan-02
3682 ---------------------
3683
3684 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
3685
3686 2. If --disable-static was given, the building process failed when trying to
3687 build pcretest and pcregrep. (For some reason it was using libtool to compile
3688 them, which is not right, as they aren't part of the library.)
3689
3690
3691 Version 3.8 18-Dec-01
3692 ---------------------
3693
3694 1. The experimental UTF-8 code was completely screwed up. It was packing the
3695 bytes in the wrong order. How dumb can you get?
3696
3697
3698 Version 3.7 29-Oct-01
3699 ---------------------
3700
3701 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
3702 This caused pcretest, when used on the test data, to segfault. Unfortunately,
3703 this didn't happen under Solaris 8, where I normally test things.
3704
3705 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
3706 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
3707 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
3708 this fix an hour or so after the initial 3.7 release.)
3709
3710
3711 Version 3.6 23-Oct-01
3712 ---------------------
3713
3714 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
3715 offsets passed as NULL with zero offset count.
3716
3717 2. The config.guess and config.sub files had not been updated when I moved to
3718 the latest autoconf.
3719
3720
3721 Version 3.5 15-Aug-01
3722 ---------------------
3723
3724 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
3725 had been forgotten.
3726
3727 2. By using declared but undefined structures, we can avoid using "void"
3728 definitions in pcre.h while keeping the internal definitions of the structures
3729 private.
3730
3731 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
3732 user point of view, this means that both static and shared libraries are built
3733 by default, but this can be individually controlled. More of the work of
3734 handling this static/shared cases is now inside libtool instead of PCRE's make
3735 file.
3736
3737 4. The pcretest utility is now installed along with pcregrep because it is
3738 useful for users (to test regexs) and by doing this, it automatically gets
3739 relinked by libtool. The documentation has been turned into a man page, so
3740 there are now .1, .txt, and .html versions in /doc.
3741
3742 5. Upgrades to pcregrep:
3743 (i) Added long-form option names like gnu grep.
3744 (ii) Added --help to list all options with an explanatory phrase.
3745 (iii) Added -r, --recursive to recurse into sub-directories.
3746 (iv) Added -f, --file to read patterns from a file.
3747
3748 6. pcre_exec() was referring to its "code" argument before testing that
3749 argument for NULL (and giving an error if it was NULL).
3750
3751 7. Upgraded Makefile.in to allow for compiling in a different directory from
3752 the source directory.
3753
3754 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
3755 options bits, the pointer it was passed was to an int instead of to an unsigned
3756 long int. This mattered only on 64-bit systems.
3757
3758 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
3759 generated) instead of pcre.in, which it its source. Also made the same change
3760 in several of the .c files.
3761
3762 10. A new release of gcc defines printf() as a macro, which broke pcretest
3763 because it had an ifdef in the middle of a string argument for printf(). Fixed
3764 by using separate calls to printf().
3765
3766 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
3767 script, to force use of CR or LF instead of \n in the source. On non-Unix
3768 systems, the value can be set in config.h.
3769
3770 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
3771 absolute limit. Changed the text of the error message to make this clear, and
3772 likewise updated the man page.
3773
3774 13. The limit of 99 on the number of capturing subpatterns has been removed.
3775 The new limit is 65535, which I hope will not be a "real" limit.
3776
3777
3778 Version 3.4 22-Aug-00
3779 ---------------------
3780
3781 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
3782
3783 2. Diagnose condition (?(0) as an error instead of crashing on matching.
3784
3785
3786 Version 3.3 01-Aug-00
3787 ---------------------
3788
3789 1. If an octal character was given, but the value was greater than \377, it
3790 was not getting masked to the least significant bits, as documented. This could
3791 lead to crashes in some systems.
3792
3793 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
3794 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
3795
3796 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
3797 These just pass their arguments on to (pcre_free)(), but they are provided
3798 because some uses of PCRE bind it to non-C systems that can call its functions,
3799 but cannot call free() or pcre_free() directly.
3800
3801 4. Add "make test" as a synonym for "make check". Corrected some comments in
3802 the Makefile.
3803
3804 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
3805 Makefile.
3806
3807 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
3808 command called pgrep for grepping around the active processes.
3809
3810 7. Added the beginnings of support for UTF-8 character strings.
3811
3812 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
3813 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
3814 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
3815 out for the ar command.)
3816
3817
3818 Version 3.2 12-May-00
3819 ---------------------
3820
3821 This is purely a bug fixing release.
3822
3823 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
3824 of ZA. This was just one example of several cases that could provoke this bug,
3825 which was introduced by change 9 of version 2.00. The code for breaking
3826 infinite loops after an iteration that matches an empty string was't working
3827 correctly.
3828
3829 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
3830 when matched against abbab (for example). After matching an empty string, it
3831 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
3832 caused it to match further down the string than it should.
3833
3834 3. The code contained an inclusion of sys/types.h. It isn't clear why this
3835 was there because it doesn't seem to be needed, and it causes trouble on some
3836 systems, as it is not a Standard C header. It has been removed.
3837
3838 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
3839 were reported on the Macintosh. The changes were from
3840
3841 while ((c = *(++ptr)) != 0 && c != '\n');
3842 to
3843 while ((c = *(++ptr)) != 0 && c != '\n') ;
3844
3845 Totally extraordinary, but if that's what it takes...
3846
3847 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
3848 available. Added HAVE_BCOPY and an autoconf test for it; if neither
3849 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
3850 assumes the way PCRE uses memmove() (always moving upwards).
3851
3852 6. PCRE is being used in one environment where strchr() is not available. There
3853 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
3854 faster code anyway.
3855
3856
3857 Version 3.1 09-Feb-00
3858 ---------------------
3859
3860 The only change in this release is the fixing of some bugs in Makefile.in for
3861 the "install" target:
3862
3863 (1) It was failing to install pcreposix.h.
3864
3865 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
3866
3867
3868 Version 3.0 01-Feb-00
3869 ---------------------
3870
3871 1. Add support for the /+ modifier to perltest (to output $` like it does in
3872 pcretest).
3873
3874 2. Add support for the /g modifier to perltest.
3875
3876 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
3877 matches null strings.
3878
3879 4. Fix perltest so that it doesn't do unwanted things when fed an empty
3880 pattern. Perl treats empty patterns specially - it reuses the most recent
3881 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
3882 effect.
3883
3884 5. The POSIX interface was broken in that it was just handing over the POSIX
3885 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
3886 required a bigger vector, with some working space on the end. This means that
3887 the POSIX wrapper now has to get and free some memory, and copy the results.
3888
3889 6. Added some simple autoconf support, placing the test data and the
3890 documentation in separate directories, re-organizing some of the
3891 information files, and making it build pcre-config (a GNU standard). Also added
3892 libtool support for building PCRE as a shared library, which is now the
3893 default.
3894
3895 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
3896 09 are not valid octal constants. Single digits will be used for minor values
3897 less than 10.
3898
3899 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
3900 existing programs that set these in the POSIX interface can use PCRE without
3901 modification.
3902
3903 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
3904 return all that pcre_info() returns, plus additional data. The pcre_info()
3905 function is retained for compatibility, but is considered to be obsolete.
3906
3907 10. Added experimental recursion feature (?R) to handle one common case that
3908 Perl 5.6 will be able to do with (?p{...}).
3909
3910 11. Added support for POSIX character classes like [:alpha:], which Perl is
3911 adopting.
3912
3913
3914 Version 2.08 31-Aug-99
3915 ----------------------
3916
3917 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
3918 trying to match at the startoffset position, but instead was moving forward to
3919 the next newline as if a previous match had failed.
3920
3921 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
3922 and could get into a loop if a null string was matched other than at the start
3923 of the subject.
3924
3925 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
3926 be distinguished at compile time, and for completeness also added PCRE_DATE.
3927
3928 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
3929 in GnuWin32 environments.
3930
3931
3932 Version 2.07 29-Jul-99
3933 ----------------------
3934
3935 1. The documentation is now supplied in plain text form and HTML as well as in
3936 the form of man page sources.
3937
3938 2. C++ compilers don't like assigning (void *) values to other pointer types.
3939 In particular this affects malloc(). Although there is no problem in Standard
3940 C, I've put in casts to keep C++ compilers happy.
3941
3942 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
3943 should be (const char *).
3944
3945 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
3946 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
3947 However, I haven't made this a standard facility. The documentation doesn't
3948 mention it, and the Makefile doesn't support it.
3949
3950 5. The Makefile now contains an "install" target, with editable destinations at
3951 the top of the file. The pcretest program is not installed.
3952
3953 6. pgrep -V now gives the PCRE version number and date.
3954
3955 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
3956 causing the entire string to be ignored, instead of just the last character.
3957
3958 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
3959 non-matching string, it can take a very, very long time, even for strings of
3960 quite modest length, because of the nested recursion. PCRE now does better in
3961 some of these cases. It does this by remembering the last required literal
3962 character in the pattern, and pre-searching the subject to ensure it is present
3963 before running the real match. In other words, it applies a heuristic to detect
3964 some types of certain failure quickly, and in the above example, if presented
3965 with a string that has no trailing " it gives "no match" very quickly.
3966
3967 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
3968 other alternatives are tried instead.
3969
3970
3971 Version 2.06 09-Jun-99
3972 ----------------------
3973
3974 1. Change pcretest's output for amount of store used to show just the code
3975 space, because the remainder (the data block) varies in size between 32-bit and
3976 64-bit systems.
3977
3978 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
3979 start matching at. This allows lookbehinds to work when searching for multiple
3980 occurrences in a string.
3981
3982 3. Added additional options to pcretest for testing multiple occurrences:
3983
3984 /+ outputs the rest of the string that follows a match
3985 /g loops for multiple occurrences, using the new startoffset argument
3986 /G loops for multiple occurrences by passing an incremented pointer
3987
3988 4. PCRE wasn't doing the "first character" optimization for patterns starting
3989 with \b or \B, though it was doing it for other lookbehind assertions. That is,
3990 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
3991 the letter 'x'. On long subject strings, this gives a significant speed-up.
3992
3993
3994 Version 2.05 21-Apr-99
3995 ----------------------
3996
3997 1. Changed the type of magic_number from int to long int so that it works
3998 properly on 16-bit systems.
3999
4000 2. Fixed a bug which caused patterns starting with .* not to work correctly
4001 when the subject string contained newline characters. PCRE was assuming
4002 anchoring for such patterns in all cases, which is not correct because .* will
4003 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
4004 DOTALL is set at top level; otherwise it knows that patterns starting with .*
4005 must be retried after every newline in the subject.
4006
4007
4008 Version 2.04 18-Feb-99
4009 ----------------------
4010
4011 1. For parenthesized subpatterns with repeats whose minimum was zero, the
4012 computation of the store needed to hold the pattern was incorrect (too large).
4013 If such patterns were nested a few deep, this could multiply and become a real
4014 problem.
4015
4016 2. Added /M option to pcretest to show the memory requirement of a specific
4017 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
4018
4019 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
4020 compiled in such a way that the backtracking after subsequent failure was
4021 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
4022 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
4023
4024
4025 Version 2.03 02-Feb-99
4026 ----------------------
4027
4028 1. Fixed typo and small mistake in man page.
4029
4030 2. Added 4th condition (GPL supersedes if conflict) and created separate
4031 LICENCE file containing the conditions.
4032
4033 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
4034 Perl, that is the internal \ allows the delimiter to be included in the
4035 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
4036 the final delimiter, add \ to the end of the pattern (to test the error).
4037
4038 4. Added the convenience functions for extracting substrings after a successful
4039 match. Updated pcretest to make it able to test these functions.
4040
4041
4042 Version 2.02 14-Jan-99
4043 ----------------------
4044
4045 1. Initialized the working variables associated with each extraction so that
4046 their saving and restoring doesn't refer to uninitialized store.
4047
4048 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
4049 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
4050 fix the problem.
4051
4052 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
4053 calls, and wasn't printing the correct value for compiling calls. Increased the
4054 default value of LOOPREPEAT, and the number of significant figures in the
4055 times.
4056
4057 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
4058
4059 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
4060 a building problem on Windows NT with a FAT file system.
4061
4062
4063 Version 2.01 21-Oct-98
4064 ----------------------
4065
4066 1. Changed the API for pcre_compile() to allow for the provision of a pointer
4067 to character tables built by pcre_maketables() in the current locale. If NULL
4068 is passed, the default tables are used.
4069
4070
4071 Version 2.00 24-Sep-98
4072 ----------------------
4073
4074 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
4075 it any more.
4076
4077 2. Allow quantification of (?>) groups, and make it work correctly.
4078
4079 3. The first character computation wasn't working for (?>) groups.
4080
4081 4. Correct the implementation of \Z (it is permitted to match on the \n at the
4082 end of the subject) and add 5.005's \z, which really does match only at the
4083 very end of the subject.
4084
4085 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
4086
4087 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
4088 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
4089 localized options. All options to pcre_study() were also removed.
4090
4091 7. Add other new features from 5.005:
4092
4093 $(?<= positive lookbehind
4094 $(?<! negative lookbehind
4095 (?imsx-imsx) added the unsetting capability
4096 such a setting is global if at outer level; local otherwise
4097 (?imsx-imsx:) non-capturing groups with option setting
4098 (?(cond)re|re) conditional pattern matching
4099
4100 A backreference to itself in a repeated group matches the previous
4101 captured string.
4102
4103 8. General tidying up of studying (both automatic and via "study")
4104 consequential on the addition of new assertions.
4105
4106 9. As in 5.005, unlimited repeated groups that could match an empty substring
4107 are no longer faulted at compile time. Instead, the loop is forcibly broken at
4108 runtime if any iteration does actually match an empty substring.
4109
4110 10. Include the RunTest script in the distribution.
4111
4112 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
4113 discrepancies, some of which were old and were also with respect to 5.004. They
4114 have now been fixed.
4115
4116
4117 Version 1.09 28-Apr-98
4118 ----------------------
4119
4120 1. A negated single character class followed by a quantifier with a minimum
4121 value of one (e.g. [^x]{1,6} ) was not compiled correctly. This could lead to
4122 program crashes, or just wrong answers. This did not apply to negated classes
4123 containing more than one character, or to minima other than one.
4124
4125
4126 Version 1.08 27-Mar-98
4127 ----------------------
4128
4129 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
4130
4131 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
4132 latter must appear before anything that relies on it in the pattern.
4133
4134
4135 Version 1.07 16-Feb-98
4136 ----------------------
4137
4138 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
4139 repeat of a potentially empty string).
4140
4141
4142 Version 1.06 23-Jan-98
4143 ----------------------
4144
4145 1. Added Markus Oberhumer's little patches for C++.
4146
4147 2. Literal strings longer than 255 characters were broken.
4148
4149
4150 Version 1.05 23-Dec-97
4151 ----------------------
4152
4153 1. Negated character classes containing more than one character were failing if
4154 PCRE_CASELESS was set at run time.
4155
4156
4157 Version 1.04 19-Dec-97
4158 ----------------------
4159
4160 1. Corrected the man page, where some "const" qualifiers had been omitted.
4161
4162 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
4163 input syntax.
4164
4165 3. Fixed memory leak which occurred when a regex with back references was
4166 matched with an offsets vector that wasn't big enough. The temporary memory
4167 that is used in this case wasn't being freed if the match failed.
4168
4169 4. Tidied pcretest to ensure it frees memory that it gets.
4170
4171 5. Temporary memory was being obtained in the case where the passed offsets
4172 vector was exactly big enough.
4173
4174 6. Corrected definition of offsetof() from change 5 below.
4175
4176 7. I had screwed up change 6 below and broken the rules for the use of
4177 setjmp(). Now fixed.
4178
4179
4180 Version 1.03 18-Dec-97
4181 ----------------------
4182
4183 1. A erroneous regex with a missing opening parenthesis was correctly
4184 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
4185 on some systems.
4186
4187 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
4188 it was reported that one broken compiler failed on the former because "code" is
4189 also an independent variable.
4190
4191 3. The erroneous regex a[]b caused an array overrun reference.
4192
4193 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
4194 fail on data ending with that character. (It was going on too far, and checking
4195 the next character, typically a binary zero.) This was specific to the
4196 optimized code for single-character negative classes.
4197
4198 5. Added a contributed patch from the TIN world which does the following:
4199
4200 + Add an undef for memmove, in case the the system defines a macro for it.
4201
4202 + Add a definition of offsetof(), in case there isn't one. (I don't know
4203 the reason behind this - offsetof() is part of the ANSI standard - but
4204 it does no harm).
4205
4206 + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
4207 most of the places where whitespace preceded '#'. I have given up and
4208 allowed the remaining 2 cases to be at the margin.
4209
4210 + Rename some variables in pcre to eliminate shadowing. This seems very
4211 pedantic, but does no harm, of course.
4212
4213 6. Moved the call to setjmp() into its own function, to get rid of warnings
4214 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
4215
4216 7. Constructs such as \d{8,} were compiling into the equivalent of
4217 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
4218 outcome, but in this particular case used more store than had been allocated,
4219 which caused the bug to be discovered because it threw up an internal error.
4220
4221 8. The debugging code in both pcre and pcretest for outputting the compiled
4222 form of a regex was going wrong in the case of back references followed by
4223 curly-bracketed repeats.
4224
4225
4226 Version 1.02 12-Dec-97
4227 ----------------------
4228
4229 1. Typos in pcre.3 and comments in the source fixed.
4230
4231 2. Applied a contributed patch to get rid of places where it used to remove
4232 'const' from variables, and fixed some signed/unsigned and uninitialized
4233 variable warnings.
4234
4235 3. Added the "runtest" target to Makefile.
4236
4237 4. Set default compiler flag to -O2 rather than just -O.
4238
4239
4240 Version 1.01 19-Nov-97
4241 ----------------------
4242
4243 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
4244 like /([ab]*)*/, that is, for classes with more than one character in them.
4245
4246 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
4247 as /((?>a*))*/ (a PCRE_EXTRA facility).
4248
4249
4250 Version 1.00 18-Nov-97
4251 ----------------------
4252
4253 1. Added compile-time macros to support systems such as SunOS4 which don't have
4254 memmove() or strerror() but have other things that can be used instead.
4255
4256 2. Arranged that "make clean" removes the executables.
4257
4258
4259 Version 0.99 27-Oct-97
4260 ----------------------
4261
4262 1. Fixed bug in code for optimizing classes with only one character. It was
4263 initializing a 32-byte map regardless, which could cause it to run off the end
4264 of the memory it had got.
4265
4266 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
4267
4268
4269 Version 0.98 22-Oct-97
4270 ----------------------
4271
4272 1. Fixed bug in code for handling temporary memory usage when there are more
4273 back references than supplied space in the ovector. This could cause segfaults.
4274
4275
4276 Version 0.97 21-Oct-97
4277 ----------------------
4278
4279 1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
4280
4281 2. Optimized negated single characters not to use a bit map.
4282
4283 3. Brought error texts together as macro definitions; clarified some of them;
4284 fixed one that was wrong - it said "range out of order" when it meant "invalid
4285 escape sequence".
4286
4287 4. Changed some char * arguments to const char *.
4288
4289 5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
4290
4291 6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
4292 pcretest.
4293
4294
4295 Version 0.96 16-Oct-97
4296 ----------------------
4297
4298 1. Added a simple "pgrep" utility to the distribution.
4299
4300 2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
4301 unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
4302 where "ddd" means "one or more decimal digits".
4303
4304 3. Fixed serious bug. If a pattern had a back reference, but the call to
4305 pcre_exec() didn't supply a large enough ovector to record the related
4306 identifying subpattern, the match always failed. PCRE now remembers the number
4307 of the largest back reference, and gets some temporary memory in which to save
4308 the offsets during matching if necessary, in order to ensure that
4309 backreferences always work.
4310
4311 4. Increased the compatibility with Perl in a number of ways:
4312
4313 (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
4314 to request this handling. The option can be set at compile or exec time.
4315
4316 (b) $ matches before a terminating newline by default; an option
4317 PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
4318 mode). The option can be set at compile or exec time.
4319
4320 (c) The handling of \ followed by a digit other than 0 is now supposed to be
4321 the same as Perl's. If the decimal number it represents is less than 10
4322 or there aren't that many previous left capturing parentheses, an octal
4323 escape is read. Inside a character class, it's always an octal escape,
4324 even if it is a single digit.
4325
4326 (d) An escaped but undefined alphabetic character is taken as a literal,
4327 unless PCRE_EXTRA is set. Currently this just reserves the remaining
4328 escapes.
4329
4330 (e) {0} is now permitted. (The previous item is removed from the compiled
4331 pattern).
4332
4333 5. Changed all the names of code files so that the basic parts are no longer
4334 than 10 characters, and abolished the teeny "globals.c" file.
4335
4336 6. Changed the handling of character classes; they are now done with a 32-byte
4337 bit map always.
4338
4339 7. Added the -d and /D options to pcretest to make it possible to look at the
4340 internals of compilation without having to recompile pcre.
4341
4342
4343 Version 0.95 23-Sep-97
4344 ----------------------
4345
4346 1. Fixed bug in pre-pass concerning escaped "normal" characters such as \x5c or
4347 \x20 at the start of a run of normal characters. These were being treated as
4348 real characters, instead of the source characters being re-checked.
4349
4350
4351 Version 0.94 18-Sep-97
4352 ----------------------
4353
4354 1. The functions are now thread-safe, with the caveat that the global variables
4355 containing pointers to malloc() and free() or alternative functions are the
4356 same for all threads.
4357
4358 2. Get pcre_study() to generate a bitmap of initial characters for non-
4359 anchored patterns when this is possible, and use it if passed to pcre_exec().
4360
4361
4362 Version 0.93 15-Sep-97
4363 ----------------------
4364
4365 1. /(b)|(:+)/ was computing an incorrect first character.
4366
4367 2. Add pcre_study() to the API and the passing of pcre_extra to pcre_exec(),
4368 but not actually doing anything yet.
4369
4370 3. Treat "-" characters in classes that cannot be part of ranges as literals,
4371 as Perl does (e.g. [-az] or [az-]).
4372
4373 4. Set the anchored flag if a branch starts with .* or .*? because that tests
4374 all possible positions.
4375
4376 5. Split up into different modules to avoid including unneeded functions in a
4377 compiled binary. However, compile and exec are still in one module. The "study"
4378 function is split off.
4379
4380 6. The character tables are now in a separate module whose source is generated
4381 by an auxiliary program - but can then be edited by hand if required. There are
4382 now no calls to isalnum(), isspace(), isdigit(), isxdigit(), tolower() or
4383 toupper() in the code.
4384
4385 7. Turn the malloc/free funtions variables into pcre_malloc and pcre_free and
4386 make them global. Abolish the function for setting them, as the caller can now
4387 set them directly.
4388
4389
4390 Version 0.92 11-Sep-97
4391 ----------------------
4392
4393 1. A repeat with a fixed maximum and a minimum of 1 for an ordinary character
4394 (e.g. /a{1,3}/) was broken (I mis-optimized it).
4395
4396 2. Caseless matching was not working in character classes if the characters in
4397 the pattern were in upper case.
4398
4399 3. Make ranges like [W-c] work in the same way as Perl for caseless matching.
4400
4401 4. Make PCRE_ANCHORED public and accept as a compile option.
4402
4403 5. Add an options word to pcre_exec() and accept PCRE_ANCHORED and
4404 PCRE_CASELESS at run time. Add escapes \A and \I to pcretest to cause it to
4405 pass them.
4406
4407 6. Give an error if bad option bits passed at compile or run time.
4408
4409 7. Add PCRE_MULTILINE at compile and exec time, and (?m) as well. Add \M to
4410 pcretest to cause it to pass that flag.
4411
4412 8. Add pcre_info(), to get the number of identifying subpatterns, the stored
4413 options, and the first character, if set.
4414
4415 9. Recognize C+ or C{n,m} where n >= 1 as providing a fixed starting character.
4416
4417
4418 Version 0.91 10-Sep-97
4419 ----------------------
4420
4421 1. PCRE was failing to diagnose unlimited repeats of subpatterns that could
4422 match the empty string as in /(a*)*/. It was looping and ultimately crashing.
4423
4424 2. PCRE was looping on encountering an indefinitely repeated back reference to
4425 a subpattern that had matched an empty string, e.g. /(a|)\1*/. It now does what
4426 Perl does - treats the match as successful.
4427
4428 ****

Properties

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

  ViewVC Help
Powered by ViewVC 1.1.5