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

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log


Revision 892 - (show annotations)
Wed Jan 18 17:23:20 2012 UTC (3 years, 4 months ago) by ph10
File size: 225842 byte(s)
Error occurred while calculating annotation data.
Put top level heap frame on the stack.
1 ChangeLog for PCRE
2 ------------------
3
4 Version 8.30
5 ------------
6
7 1. Renamed "isnumber" as "is_a_number" because in some Mac environments this
8 name is defined in ctype.h.
9
10 2. Fixed a bug in fixed-length calculation for lookbehinds that would show up
11 only in quite long subpatterns.
12
13 3. Removed the function pcre_info(), which has been obsolete and deprecated
14 since it was replaced by pcre_fullinfo() in February 2000.
15
16 4. For a non-anchored pattern, if (*SKIP) was given with a name that did not
17 match a (*MARK), and the match failed at the start of the subject, a
18 reference to memory before the start of the subject could occur. This bug
19 was introduced by fix 17 of release 8.21.
20
21 5. A reference to an unset group with zero minimum repetition was giving
22 totally wrong answers (in non-JavaScript-compatibility mode). For example,
23 /(another)?(\1?)test/ matched against "hello world test". This bug was
24 introduced in release 8.13.
25
26 6. Add support for 16-bit character strings (a large amount of work involving
27 many changes and refactorings).
28
29 7. RunGrepTest failed on msys because \r\n was replaced by whitespace when the
30 command "pattern=`printf 'xxx\r\njkl'`" was run. The pattern is now taken
31 from a file.
32
33 8. Ovector size of 2 is also supported by JIT based pcre_exec (the ovector size
34 rounding is not applied in this particular case).
35
36 9. The invalid Unicode surrogate codepoints U+D800 to U+DFFF are now rejected
37 if they appear, or are escaped, in patterns.
38
39 10. Get rid of a number of -Wunused-but-set-variable warnings.
40
41 11. The pattern /(?=(*:x))(q|)/ matches an empty string, and returns the mark
42 "x". The similar pattern /(?=(*:x))((*:y)q|)/ did not return a mark at all.
43 Oddly, Perl behaves the same way. PCRE has been fixed so that this pattern
44 also returns the mark "x". This bug applied to capturing parentheses,
45 non-capturing parentheses, and atomic parentheses. It also applied to some
46 assertions.
47
48 12. Stephen Kelly's patch to CMakeLists.txt allows it to parse the version
49 information out of configure.ac instead of relying on pcre.h.generic, which
50 is not stored in the repository.
51
52 13. Applied Dmitry V. Levin's patch for a more portable method for linking with
53 -lreadline.
54
55 14. ZH added PCRE_CONFIG_JITTARGET; added its output to pcretest -C.
56
57 15. Applied Graycode's patch to put the top-level frame on the stack rather
58 than the heap when not using the stack for recursion. This gives a
59 performance improvement in many cases when recursion is not deep.
60
61
62 Version 8.21 12-Dec-2011
63 ------------------------
64
65 1. Updating the JIT compiler.
66
67 2. JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
68 are added as well.
69
70 3. Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
71 PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
72 calling _pcre_jit_exec. Some extra comments are added.
73
74 4. (*MARK) settings inside atomic groups that do not contain any capturing
75 parentheses, for example, (?>a(*:m)), were not being passed out. This bug
76 was introduced by change 18 for 8.20.
77
78 5. Supporting of \x, \U and \u in JavaScript compatibility mode based on the
79 ECMA-262 standard.
80
81 6. Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
82 erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
83 This bug was probably introduced by change 9 of 8.13.
84
85 7. While fixing 6 above, I noticed that a number of other items were being
86 incorrectly rejected as "not fixed length". This arose partly because newer
87 opcodes had not been added to the fixed-length checking code. I have (a)
88 corrected the bug and added tests for these items, and (b) arranged for an
89 error to occur if an unknown opcode is encountered while checking for fixed
90 length instead of just assuming "not fixed length". The items that were
91 rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
92 (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
93 repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
94
95 8. A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
96 being incorrectly compiled and would have given unpredicatble results.
97
98 9. A possessively repeated subpattern with minimum repeat count greater than
99 one behaved incorrectly. For example, (A){2,}+ behaved as if it was
100 (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
101 the first (A) could occur when it should not.
102
103 10. Add a cast and remove a redundant test from the code.
104
105 11. JIT should use pcre_malloc/pcre_free for allocation.
106
107 12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
108 best practice nowadays, and helps with cross-compiling. (If the exec_prefix
109 is anything other than /usr, -L is still shown).
110
111 13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
112
113 14. Perl does not support \N without a following name in a [] class; PCRE now
114 also gives an error.
115
116 15. If a forward reference was repeated with an upper limit of around 2000,
117 it caused the error "internal error: overran compiling workspace". The
118 maximum number of forward references (including repeats) was limited by the
119 internal workspace, and dependent on the LINK_SIZE. The code has been
120 rewritten so that the workspace expands (via pcre_malloc) if necessary, and
121 the default depends on LINK_SIZE. There is a new upper limit (for safety)
122 of around 200,000 forward references. While doing this, I also speeded up
123 the filling in of repeated forward references.
124
125 16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
126 incorrectly expecting the subject to contain another "a" after the start.
127
128 17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
129 in the match, the SKIP should be ignored. This was not happening; instead
130 the SKIP was being treated as NOMATCH. For patterns such as
131 /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
132 tested.
133
134 18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
135 now much more compatible with Perl, in particular in cases where the result
136 is a non-match for a non-anchored pattern. For example, if
137 /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
138 "m", where previously it did not return a name. A side effect of this
139 change is that for partial matches, the last encountered mark name is
140 returned, as for non matches. A number of tests that were previously not
141 Perl-compatible have been moved into the Perl-compatible test files. The
142 refactoring has had the pleasing side effect of removing one argument from
143 the match() function, thus reducing its stack requirements.
144
145 19. If the /S+ option was used in pcretest to study a pattern using JIT,
146 subsequent uses of /S (without +) incorrectly behaved like /S+.
147
148 21. Retrieve executable code size support for the JIT compiler and fixing
149 some warnings.
150
151 22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
152 not work when the shorter character appeared right at the end of the
153 subject string.
154
155 23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
156 systems.
157
158 24. Added PCRE_INFO_JITSIZE to pass on the value from (21) above, and also
159 output it when the /M option is used in pcretest.
160
161 25. The CheckMan script was not being included in the distribution. Also, added
162 an explicit "perl" to run Perl scripts from the PrepareRelease script
163 because this is reportedly needed in Windows.
164
165 26. If study data was being save in a file and studying had not found a set of
166 "starts with" bytes for the pattern, the data written to the file (though
167 never used) was taken from uninitialized memory and so caused valgrind to
168 complain.
169
170 27. Updated RunTest.bat as provided by Sheri Pierce.
171
172 28. Fixed a possible uninitialized memory bug in pcre_jit_compile.c.
173
174 29. Computation of memory usage for the table of capturing group names was
175 giving an unnecessarily large value.
176
177
178 Version 8.20 21-Oct-2011
179 ------------------------
180
181 1. Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
182 a POSIX class. After further experiments with Perl, which convinced me that
183 Perl has bugs and confusions, a closing square bracket is no longer allowed
184 in a POSIX name. This bug also affected patterns with classes that started
185 with full stops.
186
187 2. If a pattern such as /(a)b|ac/ is matched against "ac", there is no
188 captured substring, but while checking the failing first alternative,
189 substring 1 is temporarily captured. If the output vector supplied to
190 pcre_exec() was not big enough for this capture, the yield of the function
191 was still zero ("insufficient space for captured substrings"). This cannot
192 be totally fixed without adding another stack variable, which seems a lot
193 of expense for a edge case. However, I have improved the situation in cases
194 such as /(a)(b)x|abc/ matched against "abc", where the return code
195 indicates that fewer than the maximum number of slots in the ovector have
196 been set.
197
198 3. Related to (2) above: when there are more back references in a pattern than
199 slots in the output vector, pcre_exec() uses temporary memory during
200 matching, and copies in the captures as far as possible afterwards. It was
201 using the entire output vector, but this conflicts with the specification
202 that only 2/3 is used for passing back captured substrings. Now it uses
203 only the first 2/3, for compatibility. This is, of course, another edge
204 case.
205
206 4. Zoltan Herczeg's just-in-time compiler support has been integrated into the
207 main code base, and can be used by building with --enable-jit. When this is
208 done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
209 runtime --no-jit option is given.
210
211 5. When the number of matches in a pcre_dfa_exec() run exactly filled the
212 ovector, the return from the function was zero, implying that there were
213 other matches that did not fit. The correct "exactly full" value is now
214 returned.
215
216 6. If a subpattern that was called recursively or as a subroutine contained
217 (*PRUNE) or any other control that caused it to give a non-standard return,
218 invalid errors such as "Error -26 (nested recursion at the same subject
219 position)" or even infinite loops could occur.
220
221 7. If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
222 computing the minimum length on reaching *ACCEPT, and so ended up with the
223 wrong value of 1 rather than 0. Further investigation indicates that
224 computing a minimum subject length in the presence of *ACCEPT is difficult
225 (think back references, subroutine calls), and so I have changed the code
226 so that no minimum is registered for a pattern that contains *ACCEPT.
227
228 8. If (*THEN) was present in the first (true) branch of a conditional group,
229 it was not handled as intended. [But see 16 below.]
230
231 9. Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
232 Sheri Pierce.
233
234 10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
235 the first byte in a match must be "a".
236
237 11. Change 17 for 8.13 increased the recursion depth for patterns like
238 /a(?:.)*?a/ drastically. I've improved things by remembering whether a
239 pattern contains any instances of (*THEN). If it does not, the old
240 optimizations are restored. It would be nice to do this on a per-group
241 basis, but at the moment that is not feasible.
242
243 12. In some environments, the output of pcretest -C is CRLF terminated. This
244 broke RunTest's code that checks for the link size. A single white space
245 character after the value is now allowed for.
246
247 13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
248 For "fr", it uses the Windows-specific input and output files.
249
250 14. If (*THEN) appeared in a group that was called recursively or as a
251 subroutine, it did not work as intended. [But see next item.]
252
253 15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
254 pattern fragments (but not containing any | characters). If A and B are
255 matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
256 was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
257 D. In other words, Perl considers parentheses that do not contain any |
258 characters to be part of a surrounding alternative, whereas PCRE was
259 treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
260 differently. PCRE now behaves in the same way as Perl, except in the case
261 of subroutine/recursion calls such as (?1) which have in any case always
262 been different (but PCRE had them first :-).
263
264 16. Related to 15 above: Perl does not treat the | in a conditional group as
265 creating alternatives. Such a group is treated in the same way as an
266 ordinary group without any | characters when processing (*THEN). PCRE has
267 been changed to match Perl's behaviour.
268
269 17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
270 RunGrepTest script failed.
271
272 18. Change 22 for version 13 caused atomic groups to use more stack. This is
273 inevitable for groups that contain captures, but it can lead to a lot of
274 stack use in large patterns. The old behaviour has been restored for atomic
275 groups that do not contain any capturing parentheses.
276
277 19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
278 suppress the check for a minimum subject length at run time. (If it was
279 given to pcre_exec() or pcre_dfa_exec() it did work.)
280
281 20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
282 fail to work when decoding hex characters in data strings in EBCDIC
283 environments.
284
285 21. It appears that in at least one Mac OS environment, the isxdigit() function
286 is implemented as a macro that evaluates to its argument more than once,
287 contravening the C 90 Standard (I haven't checked a later standard). There
288 was an instance in pcretest which caused it to go wrong when processing
289 \x{...} escapes in subject strings. The has been rewritten to avoid using
290 things like p++ in the argument of isxdigit().
291
292
293 Version 8.13 16-Aug-2011
294 ------------------------
295
296 1. The Unicode data tables have been updated to Unicode 6.0.0.
297
298 2. Two minor typos in pcre_internal.h have been fixed.
299
300 3. Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
301 pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
302 in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
303
304 4. There were a number of related bugs in the code for matching backrefences
305 caselessly in UTF-8 mode when codes for the characters concerned were
306 different numbers of bytes. For example, U+023A and U+2C65 are an upper
307 and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
308 (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
309 code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
310 2-byte code at the end of the subject (it thought there wasn't enough data
311 left).
312
313 5. Comprehensive information about what went wrong is now returned by
314 pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
315 as the output vector has at least 2 elements. The offset of the start of
316 the failing character and a reason code are placed in the vector.
317
318 6. When the UTF-8 string check fails for pcre_compile(), the offset that is
319 now returned is for the first byte of the failing character, instead of the
320 last byte inspected. This is an incompatible change, but I hope it is small
321 enough not to be a problem. It makes the returned offset consistent with
322 pcre_exec() and pcre_dfa_exec().
323
324 7. pcretest now gives a text phrase as well as the error number when
325 pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
326 failure, the offset and reason code are output.
327
328 8. When \R was used with a maximizing quantifier it failed to skip backwards
329 over a \r\n pair if the subsequent match failed. Instead, it just skipped
330 back over a single character (\n). This seems wrong (because it treated the
331 two characters as a single entity when going forwards), conflicts with the
332 documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
333 behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
334 has been changed.
335
336 9. Some internal refactoring has changed the processing so that the handling
337 of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
338 time (the PCRE_DOTALL option was changed this way some time ago: version
339 7.7 change 16). This has made it possible to abolish the OP_OPT op code,
340 which was always a bit of a fudge. It also means that there is one less
341 argument for the match() function, which reduces its stack requirements
342 slightly. This change also fixes an incompatibility with Perl: the pattern
343 (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
344
345 10. More internal refactoring has drastically reduced the number of recursive
346 calls to match() for possessively repeated groups such as (abc)++ when
347 using pcre_exec().
348
349 11. While implementing 10, a number of bugs in the handling of groups were
350 discovered and fixed:
351
352 (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
353 (a|)*(?1) gave a compile-time internal error.
354 ((a|)+)+ did not notice that the outer group could match an empty string.
355 (^a|^)+ was not marked as anchored.
356 (.*a|.*)+ was not marked as matching at start or after a newline.
357
358 12. Yet more internal refactoring has removed another argument from the match()
359 function. Special calls to this function are now indicated by setting a
360 value in a variable in the "match data" data block.
361
362 13. Be more explicit in pcre_study() instead of relying on "default" for
363 opcodes that mean there is no starting character; this means that when new
364 ones are added and accidentally left out of pcre_study(), testing should
365 pick them up.
366
367 14. The -s option of pcretest has been documented for ages as being an old
368 synonym of -m (show memory usage). I have changed it to mean "force study
369 for every regex", that is, assume /S for every regex. This is similar to -i
370 and -d etc. It's slightly incompatible, but I'm hoping nobody is still
371 using it. It makes it easier to run collections of tests with and without
372 study enabled, and thereby test pcre_study() more easily. All the standard
373 tests are now run with and without -s (but some patterns can be marked as
374 "never study" - see 20 below).
375
376 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
377 restoration of the capturing data to the outer values was not happening
378 correctly.
379
380 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
381 empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
382 pattern had matched an empty string, and so incorrectly returned a no
383 match.
384
385 17. There was optimizing code for the last branch of non-capturing parentheses,
386 and also for the obeyed branch of a conditional subexpression, which used
387 tail recursion to cut down on stack usage. Unfortunately, now that there is
388 the possibility of (*THEN) occurring in these branches, tail recursion is
389 no longer possible because the return has to be checked for (*THEN). These
390 two optimizations have therefore been removed. [But see 8.20/11 above.]
391
392 18. If a pattern containing \R was studied, it was assumed that \R always
393 matched two bytes, thus causing the minimum subject length to be
394 incorrectly computed because \R can also match just one byte.
395
396 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
397 was incorrectly computed.
398
399 20. If /S is present twice on a test pattern in pcretest input, it now
400 *disables* studying, thereby overriding the use of -s on the command line
401 (see 14 above). This is necessary for one or two tests to keep the output
402 identical in both cases.
403
404 21. When (*ACCEPT) was used in an assertion that matched an empty string and
405 PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
406
407 22. When an atomic group that contained a capturing parenthesis was
408 successfully matched, but the branch in which it appeared failed, the
409 capturing was not being forgotten if a higher numbered group was later
410 captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
411 group 1 to "a", when in fact it should be unset. This applied to multi-
412 branched capturing and non-capturing groups, repeated or not, and also to
413 positive assertions (capturing in negative assertions does not happen
414 in PCRE) and also to nested atomic groups.
415
416 23. Add the ++ qualifier feature to pcretest, to show the remainder of the
417 subject after a captured substring, to make it easier to tell which of a
418 number of identical substrings has been captured.
419
420 24. The way atomic groups are processed by pcre_exec() has been changed so that
421 if they are repeated, backtracking one repetition now resets captured
422 values correctly. For example, if ((?>(a+)b)+aabab) is matched against
423 "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
424 "aaa". Previously, it would have been "a". As part of this code
425 refactoring, the way recursive calls are handled has also been changed.
426
427 25. If an assertion condition captured any substrings, they were not passed
428 back unless some other capturing happened later. For example, if
429 (?(?=(a))a) was matched against "a", no capturing was returned.
430
431 26. When studying a pattern that contained subroutine calls or assertions,
432 the code for finding the minimum length of a possible match was handling
433 direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
434 group 1 called group 2 while simultaneously a separate group 2 called group
435 1). A stack overflow occurred in this case. I have fixed this by limiting
436 the recursion depth to 10.
437
438 27. Updated RunTest.bat in the distribution to the version supplied by Tom
439 Fortmann. This supports explicit test numbers on the command line, and has
440 argument validation and error reporting.
441
442 28. An instance of \X with an unlimited repeat could fail if at any point the
443 first character it looked at was a mark character.
444
445 29. Some minor code refactoring concerning Unicode properties and scripts
446 should reduce the stack requirement of match() slightly.
447
448 30. Added the '=' option to pcretest to check the setting of unused capturing
449 slots at the end of the pattern, which are documented as being -1, but are
450 not included in the return count.
451
452 31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
453 compiled something random. Now it gives a compile-time error (as does
454 Perl).
455
456 32. A *MARK encountered during the processing of a positive assertion is now
457 recorded and passed back (compatible with Perl).
458
459 33. If --only-matching or --colour was set on a pcregrep call whose pattern
460 had alternative anchored branches, the search for a second match in a line
461 was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
462 matched the line "0102" twice. The same bug affected patterns that started
463 with a backwards assertion. For example /\b01|\b02/ also matched "0102"
464 twice.
465
466 34. Previously, PCRE did not allow quantification of assertions. However, Perl
467 does, and because of capturing effects, quantifying parenthesized
468 assertions may at times be useful. Quantifiers are now allowed for
469 parenthesized assertions.
470
471 35. A minor code tidy in pcre_compile() when checking options for \R usage.
472
473 36. \g was being checked for fancy things in a character class, when it should
474 just be a literal "g".
475
476 37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
477 appearance of a nested POSIX class supersedes an apparent external class.
478 For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
479 unescaped square brackets may also appear as part of class names. For
480 example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
481 more like Perl. (But see 8.20/1 above.)
482
483 38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
484 was because it thought it was \N{name}, which is not supported).
485
486 39. Add minix to OS list not supporting the -S option in pcretest.
487
488 40. PCRE tries to detect cases of infinite recursion at compile time, but it
489 cannot analyze patterns in sufficient detail to catch mutual recursions
490 such as ((?1))((?2)). There is now a runtime test that gives an error if a
491 subgroup is called recursively as a subpattern for a second time at the
492 same position in the subject string. In previous releases this might have
493 been caught by the recursion limit, or it might have run out of stack.
494
495 41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
496 happen only once. PCRE was, however incorrectly giving a compile time error
497 "recursive call could loop indefinitely" because it cannot analyze the
498 pattern in sufficient detail. The compile time test no longer happens when
499 PCRE is compiling a conditional subpattern, but actual runaway loops are
500 now caught at runtime (see 40 above).
501
502 42. It seems that Perl allows any characters other than a closing parenthesis
503 to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
504 has been changed to be the same.
505
506 43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
507 as not to get warnings when autogen.sh is called. Also changed
508 AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
509
510 44. To help people who use pcregrep to scan files containing exceedingly long
511 lines, the following changes have been made:
512
513 (a) The default value of the buffer size parameter has been increased from
514 8K to 20K. (The actual buffer used is three times this size.)
515
516 (b) The default can be changed by ./configure --with-pcregrep-bufsize when
517 PCRE is built.
518
519 (c) A --buffer-size=n option has been added to pcregrep, to allow the size
520 to be set at run time.
521
522 (d) Numerical values in pcregrep options can be followed by K or M, for
523 example --buffer-size=50K.
524
525 (e) If a line being scanned overflows pcregrep's buffer, an error is now
526 given and the return code is set to 2.
527
528 45. Add a pointer to the latest mark to the callout data block.
529
530 46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
531 partial match of an empty string instead of no match. This was specific to
532 the use of ".".
533
534 47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
535 complete match instead of a partial match. This bug was dependent on both
536 the PCRE_UTF8 and PCRE_DOTALL options being set.
537
538 48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
539 starting byte set, because \b was not being ignored.
540
541
542 Version 8.12 15-Jan-2011
543 ------------------------
544
545 1. Fixed some typos in the markup of the man pages, and wrote a script that
546 checks for such things as part of the documentation building process.
547
548 2. On a big-endian 64-bit system, pcregrep did not correctly process the
549 --match-limit and --recursion-limit options (added for 8.11). In
550 particular, this made one of the standard tests fail. (The integer value
551 went into the wrong half of a long int.)
552
553 3. If the --colour option was given to pcregrep with -v (invert match), it
554 did strange things, either producing crazy output, or crashing. It should,
555 of course, ignore a request for colour when reporting lines that do not
556 match.
557
558 4. Another pcregrep bug caused similar problems if --colour was specified with
559 -M (multiline) and the pattern match finished with a line ending.
560
561 5. In pcregrep, when a pattern that ended with a literal newline sequence was
562 matched in multiline mode, the following line was shown as part of the
563 match. This seems wrong, so I have changed it.
564
565 6. Another pcregrep bug in multiline mode, when --colour was specified, caused
566 the check for further matches in the same line (so they could be coloured)
567 to overrun the end of the current line. If another match was found, it was
568 incorrectly shown (and then shown again when found in the next line).
569
570 7. If pcregrep was compiled under Windows, there was a reference to the
571 function pcregrep_exit() before it was defined. I am assuming this was
572 the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
573 reported by a user. I've moved the definition above the reference.
574
575
576 Version 8.11 10-Dec-2010
577 ------------------------
578
579 1. (*THEN) was not working properly if there were untried alternatives prior
580 to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
581 backtracked to try for "b" instead of moving to the next alternative branch
582 at the same level (in this case, to look for "c"). The Perl documentation
583 is clear that when (*THEN) is backtracked onto, it goes to the "next
584 alternative in the innermost enclosing group".
585
586 2. (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
587 such as (A(*COMMIT)B(*THEN)C|D) any failure after matching A should
588 result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
589 (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
590 (*THEN).
591
592 3. If \s appeared in a character class, it removed the VT character from
593 the class, even if it had been included by some previous item, for example
594 in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
595 of \s, but is part of the POSIX "space" class.)
596
597 4. A partial match never returns an empty string (because you can always
598 match an empty string at the end of the subject); however the checking for
599 an empty string was starting at the "start of match" point. This has been
600 changed to the "earliest inspected character" point, because the returned
601 data for a partial match starts at this character. This means that, for
602 example, /(?<=abc)def/ gives a partial match for the subject "abc"
603 (previously it gave "no match").
604
605 5. Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
606 of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
607 previously a full match would be given. However, setting PCRE_PARTIAL_HARD
608 has an implication that the given string is incomplete (because a partial
609 match is preferred over a full match). For this reason, these items now
610 give a partial match in this situation. [Aside: previously, the one case
611 /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
612 match rather than a full match, which was wrong by the old rules, but is
613 now correct.]
614
615 6. There was a bug in the handling of #-introduced comments, recognized when
616 PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
617 If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
618 UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
619 scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
620 but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
621 places in pcre_compile().
622
623 7. Related to (6) above, when pcre_compile() was skipping #-introduced
624 comments when looking ahead for named forward references to subpatterns,
625 the only newline sequence it recognized was NL. It now handles newlines
626 according to the set newline convention.
627
628 8. SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
629 former, but used strtoul(), whereas pcretest avoided strtoul() but did not
630 cater for a lack of strerror(). These oversights have been fixed.
631
632 9. Added --match-limit and --recursion-limit to pcregrep.
633
634 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
635
636 11. When the -o option was used, pcregrep was setting a return code of 1, even
637 when matches were found, and --line-buffered was not being honoured.
638
639 12. Added an optional parentheses number to the -o and --only-matching options
640 of pcregrep.
641
642 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
643 can match an empty string. The code to do it in pcretest and pcredemo
644 needed fixing:
645
646 (a) When the newline convention was "crlf", pcretest got it wrong, skipping
647 only one byte after an empty string match just before CRLF (this case
648 just got forgotten; "any" and "anycrlf" were OK).
649
650 (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
651 mode when an empty string match preceded an ASCII character followed by
652 a non-ASCII character. (The code for advancing by one character rather
653 than one byte was nonsense.)
654
655 (c) The pcredemo.c sample program did not have any code at all to handle
656 the cases when CRLF is a valid newline sequence.
657
658 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
659 as a starting offset was within the subject string. There is now a new
660 error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
661 negative or greater than the length of the string. In order to test this,
662 pcretest is extended to allow the setting of negative starting offsets.
663
664 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
665 starting offset points to the beginning of a UTF-8 character was
666 unnecessarily clumsy. I tidied it up.
667
668 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
669 bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
670
671 17. Nobody had reported that the --include_dir option, which was added in
672 release 7.7 should have been called --include-dir (hyphen, not underscore)
673 for compatibility with GNU grep. I have changed it to --include-dir, but
674 left --include_dir as an undocumented synonym, and the same for
675 --exclude-dir, though that is not available in GNU grep, at least as of
676 release 2.5.4.
677
678 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
679 characters from a string of bytes) have been redefined so as not to use
680 loops, in order to improve performance in some environments. At the same
681 time, I abstracted some of the common code into auxiliary macros to save
682 repetition (this should not affect the compiled code).
683
684 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
685 compile-time error is now given if \c is not followed by an ASCII
686 character, that is, a byte less than 128. (In EBCDIC mode, the code is
687 different, and any byte value is allowed.)
688
689 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
690 START_OPTIMIZE option, which is now allowed at compile time - but just
691 passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
692 to pcregrep and other applications that have no direct access to PCRE
693 options. The new /Y option in pcretest sets this option when calling
694 pcre_compile().
695
696 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
697 back references. Groups containing recursive back references were forced to
698 be atomic by that change, but in the case of named groups, the amount of
699 memory required was incorrectly computed, leading to "Failed: internal
700 error: code overflow". This has been fixed.
701
702 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
703 pcretest.c, to avoid build problems in some Borland environments.
704
705
706 Version 8.10 25-Jun-2010
707 ------------------------
708
709 1. Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
710 THEN.
711
712 2. (*ACCEPT) was not working when inside an atomic group.
713
714 3. Inside a character class, \B is treated as a literal by default, but
715 faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
716 causes the error). The code is unchanged, but I tidied the documentation.
717
718 4. Inside a character class, PCRE always treated \R and \X as literals,
719 whereas Perl faults them if its -w option is set. I have changed PCRE so
720 that it faults them when PCRE_EXTRA is set.
721
722 5. Added support for \N, which always matches any character other than
723 newline. (It is the same as "." when PCRE_DOTALL is not set.)
724
725 6. When compiling pcregrep with newer versions of gcc which may have
726 FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
727 declared with attribute warn_unused_result" were given. Just casting the
728 result to (void) does not stop the warnings; a more elaborate fudge is
729 needed. I've used a macro to implement this.
730
731 7. Minor change to pcretest.c to avoid a compiler warning.
732
733 8. Added four artifical Unicode properties to help with an option to make
734 \s etc use properties (see next item). The new properties are: Xan
735 (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
736
737 9. Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
738 use Unicode properties. (*UCP) at the start of a pattern can be used to set
739 this option. Modified pcretest to add /W to test this facility. Added
740 REG_UCP to make it available via the POSIX interface.
741
742 10. Added --line-buffered to pcregrep.
743
744 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
745 studied, and the match started with a letter with a code point greater than
746 127 whose first byte was different to the first byte of the other case of
747 the letter, the other case of this starting letter was not recognized
748 (#976).
749
750 12. If a pattern that was studied started with a repeated Unicode property
751 test, for example, \p{Nd}+, there was the theoretical possibility of
752 setting up an incorrect bitmap of starting bytes, but fortunately it could
753 not have actually happened in practice until change 8 above was made (it
754 added property types that matched character-matching opcodes).
755
756 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
757 possible starting bytes for non-anchored patterns.
758
759 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
760 \R, and also a number of cases that involve Unicode properties, both
761 explicit and implicit when PCRE_UCP is set.
762
763 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
764 input, it could crash or give wrong results if characters with values
765 greater than 0xc0 were present in the subject string. (Detail: it assumed
766 UTF-8 input when processing these items.)
767
768 16. Added a lot of (int) casts to avoid compiler warnings in systems where
769 size_t is 64-bit (#991).
770
771 17. Added a check for running out of memory when PCRE is compiled with
772 --disable-stack-for-recursion (#990).
773
774 18. If the last data line in a file for pcretest does not have a newline on
775 the end, a newline was missing in the output.
776
777 19. The default pcre_chartables.c file recognizes only ASCII characters (values
778 less than 128) in its various bitmaps. However, there is a facility for
779 generating tables according to the current locale when PCRE is compiled. It
780 turns out that in some environments, 0x85 and 0xa0, which are Unicode space
781 characters, are recognized by isspace() and therefore were getting set in
782 these tables, and indeed these tables seem to approximate to ISO 8859. This
783 caused a problem in UTF-8 mode when pcre_study() was used to create a list
784 of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
785 which of course cannot start UTF-8 characters. I have changed the code so
786 that only real ASCII characters (less than 128) and the correct starting
787 bytes for UTF-8 encodings are set for characters greater than 127 when in
788 UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
789 altogether.)
790
791 20. Added the /T option to pcretest so as to be able to run tests with non-
792 standard character tables, thus making it possible to include the tests
793 used for 19 above in the standard set of tests.
794
795 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
796 reference to a subpattern the other side of a comment that contains an
797 opening parenthesis caused either an internal compiling error, or a
798 reference to the wrong subpattern.
799
800
801 Version 8.02 19-Mar-2010
802 ------------------------
803
804 1. The Unicode data tables have been updated to Unicode 5.2.0.
805
806 2. Added the option --libs-cpp to pcre-config, but only when C++ support is
807 configured.
808
809 3. Updated the licensing terms in the pcregexp.pas file, as agreed with the
810 original author of that file, following a query about its status.
811
812 4. On systems that do not have stdint.h (e.g. Solaris), check for and include
813 inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
814
815 5. A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
816 quantifier applied to a forward-referencing subroutine call, could compile
817 incorrect code or give the error "internal error: previously-checked
818 referenced subpattern not found".
819
820 6. Both MS Visual Studio and Symbian OS have problems with initializing
821 variables to point to external functions. For these systems, therefore,
822 pcre_malloc etc. are now initialized to local functions that call the
823 relevant global functions.
824
825 7. There were two entries missing in the vectors called coptable and poptable
826 in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
827 I've fixed the data, and added a kludgy way of testing at compile time that
828 the lengths are correct (equal to the number of opcodes).
829
830 8. Following on from 7, I added a similar kludge to check the length of the
831 eint vector in pcreposix.c.
832
833 9. Error texts for pcre_compile() are held as one long string to avoid too
834 much relocation at load time. To find a text, the string is searched,
835 counting zeros. There was no check for running off the end of the string,
836 which could happen if a new error number was added without updating the
837 string.
838
839 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
840
841 11. \K was not working if it appeared in an atomic group or in a group that
842 was called as a "subroutine", or in an assertion. Perl 5.11 documents that
843 \K is "not well defined" if used in an assertion. PCRE now accepts it if
844 the assertion is positive, but not if it is negative.
845
846 12. Change 11 fortuitously reduced the size of the stack frame used in the
847 "match()" function of pcre_exec.c by one pointer. Forthcoming
848 implementation of support for (*MARK) will need an extra pointer on the
849 stack; I have reserved it now, so that the stack frame size does not
850 decrease.
851
852 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
853 item in branch that calls a recursion is a subroutine call - as in the
854 second branch in the above example - was incorrectly given the compile-
855 time error "recursive call could loop indefinitely" because pcre_compile()
856 was not correctly checking the subroutine for matching a non-empty string.
857
858 14. The checks for overrunning compiling workspace could trigger after an
859 overrun had occurred. This is a "should never occur" error, but it can be
860 triggered by pathological patterns such as hundreds of nested parentheses.
861 The checks now trigger 100 bytes before the end of the workspace.
862
863 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
864
865
866 Version 8.01 19-Jan-2010
867 ------------------------
868
869 1. If a pattern contained a conditional subpattern with only one branch (in
870 particular, this includes all (*DEFINE) patterns), a call to pcre_study()
871 computed the wrong minimum data length (which is of course zero for such
872 subpatterns). This could cause incorrect "no match" results.
873
874 2. For patterns such as (?i)a(?-i)b|c where an option setting at the start of
875 the pattern is reset in the first branch, pcre_compile() failed with
876 "internal error: code overflow at offset...". This happened only when
877 the reset was to the original external option setting. (An optimization
878 abstracts leading options settings into an external setting, which was the
879 cause of this.)
880
881 3. A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
882 of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
883 assertion pattern did not match (meaning that the assertion was true), it
884 was incorrectly treated as false if the SKIP had been reached during the
885 matching. This also applied to assertions used as conditions.
886
887 4. If an item that is not supported by pcre_dfa_exec() was encountered in an
888 assertion subpattern, including such a pattern used as a condition,
889 unpredictable results occurred, instead of the error return
890 PCRE_ERROR_DFA_UITEM.
891
892 5. The C++ GlobalReplace function was not working like Perl for the special
893 situation when an empty string is matched. It now does the fancy magic
894 stuff that is necessary.
895
896 6. In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
897 removed. (These were left over from very, very early versions of PCRE.)
898
899 7. Some cosmetic changes to the code to make life easier when compiling it
900 as part of something else:
901
902 (a) Change DEBUG to PCRE_DEBUG.
903
904 (b) In pcre_compile(), rename the member of the "branch_chain" structure
905 called "current" as "current_branch", to prevent a collision with the
906 Linux macro when compiled as a kernel module.
907
908 (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
909 prevent a collision with the Linux macro when compiled as a kernel
910 module.
911
912 8. In pcre_compile() there are some checks for integer overflows that used to
913 cast potentially large values to (double). This has been changed to that
914 when building, a check for int64_t is made, and if it is found, it is used
915 instead, thus avoiding the use of floating point arithmetic. (There is no
916 other use of FP in PCRE.) If int64_t is not found, the fallback is to
917 double.
918
919 9. Added two casts to avoid signed/unsigned warnings from VS Studio Express
920 2005 (difference between two addresses compared to an unsigned value).
921
922 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
923 custom one, because of the following reported problem in Windows:
924
925 - libbz2 uses the Pascal calling convention (WINAPI) for the functions
926 under Win32.
927 - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
928 therefore missing the function definition.
929 - The compiler thus generates a "C" signature for the test function.
930 - The linker fails to find the "C" function.
931 - PCRE fails to configure if asked to do so against libbz2.
932
933 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
934 messages were output:
935
936 Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
937 rerunning libtoolize, to keep the correct libtool macros in-tree.
938 Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
939
940 I have done both of these things.
941
942 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
943 most of the time, it *can* run out if it is given a pattern that contains a
944 runaway infinite recursion. I updated the discussion in the pcrestack man
945 page.
946
947 13. Now that we have gone to the x.xx style of version numbers, the minor
948 version may start with zero. Using 08 or 09 is a bad idea because users
949 might check the value of PCRE_MINOR in their code, and 08 or 09 may be
950 interpreted as invalid octal numbers. I've updated the previous comment in
951 configure.ac, and also added a check that gives an error if 08 or 09 are
952 used.
953
954 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
955 causing partial matching to fail when the end of the subject matched \W
956 in a UTF-8 pattern where \W was quantified with a minimum of 3.
957
958 15. There were some discrepancies between the declarations in pcre_internal.h
959 of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
960 their definitions. The declarations used "const uschar *" and the
961 definitions used USPTR. Even though USPTR is normally defined as "const
962 unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
963 reported that: "This difference in casting confuses some C++ compilers, for
964 example, SunCC recognizes above declarations as different functions and
965 generates broken code for hbpcre." I have changed the declarations to use
966 USPTR.
967
968 16. GNU libtool is named differently on some systems. The autogen.sh script now
969 tries several variants such as glibtoolize (MacOSX) and libtoolize1x
970 (FreeBSD).
971
972 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
973 (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
974 comment: "Figure out how to create a longlong from a string: strtoll and
975 equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
976 instance, but it only takes 2 args instead of 3!"
977
978 18. A subtle bug concerned with back references has been fixed by a change of
979 specification, with a corresponding code fix. A pattern such as
980 ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
981 refers, was giving matches when it shouldn't. For example, xa=xaaa would
982 match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
983 same bug. Such groups have to be quantified to be useful, or contained
984 inside another quantified group. (If there's no repetition, the reference
985 can never match.) The problem arises because, having left the group and
986 moved on to the rest of the pattern, a later failure that backtracks into
987 the group uses the captured value from the final iteration of the group
988 rather than the correct earlier one. I have fixed this in PCRE by forcing
989 any group that contains a reference to itself to be an atomic group; that
990 is, there cannot be any backtracking into it once it has completed. This is
991 similar to recursive and subroutine calls.
992
993
994 Version 8.00 19-Oct-09
995 ----------------------
996
997 1. The table for translating pcre_compile() error codes into POSIX error codes
998 was out-of-date, and there was no check on the pcre_compile() error code
999 being within the table. This could lead to an OK return being given in
1000 error.
1001
1002 2. Changed the call to open a subject file in pcregrep from fopen(pathname,
1003 "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
1004 in a Windows environment.
1005
1006 3. The pcregrep --count option prints the count for each file even when it is
1007 zero, as does GNU grep. However, pcregrep was also printing all files when
1008 --files-with-matches was added. Now, when both options are given, it prints
1009 counts only for those files that have at least one match. (GNU grep just
1010 prints the file name in this circumstance, but including the count seems
1011 more useful - otherwise, why use --count?) Also ensured that the
1012 combination -clh just lists non-zero counts, with no names.
1013
1014 4. The long form of the pcregrep -F option was incorrectly implemented as
1015 --fixed_strings instead of --fixed-strings. This is an incompatible change,
1016 but it seems right to fix it, and I didn't think it was worth preserving
1017 the old behaviour.
1018
1019 5. The command line items --regex=pattern and --regexp=pattern were not
1020 recognized by pcregrep, which required --regex pattern or --regexp pattern
1021 (with a space rather than an '='). The man page documented the '=' forms,
1022 which are compatible with GNU grep; these now work.
1023
1024 6. No libpcreposix.pc file was created for pkg-config; there was just
1025 libpcre.pc and libpcrecpp.pc. The omission has been rectified.
1026
1027 7. Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
1028 when UCP support is not needed, by modifying the Python script that
1029 generates it from Unicode data files. This should not matter if the module
1030 is correctly used as a library, but I received one complaint about 50K of
1031 unwanted data. My guess is that the person linked everything into his
1032 program rather than using a library. Anyway, it does no harm.
1033
1034 8. A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
1035 was a minimum greater than 1 for a wide character in a possessive
1036 repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
1037 which had an unlimited repeat of a nested, fixed maximum repeat of a wide
1038 character. Chaos in the form of incorrect output or a compiling loop could
1039 result.
1040
1041 9. The restrictions on what a pattern can contain when partial matching is
1042 requested for pcre_exec() have been removed. All patterns can now be
1043 partially matched by this function. In addition, if there are at least two
1044 slots in the offset vector, the offset of the earliest inspected character
1045 for the match and the offset of the end of the subject are set in them when
1046 PCRE_ERROR_PARTIAL is returned.
1047
1048 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
1049 synonymous with PCRE_PARTIAL, for backwards compatibility, and
1050 PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
1051 and may be more useful for multi-segment matching.
1052
1053 11. Partial matching with pcre_exec() is now more intuitive. A partial match
1054 used to be given if ever the end of the subject was reached; now it is
1055 given only if matching could not proceed because another character was
1056 needed. This makes a difference in some odd cases such as Z(*FAIL) with the
1057 string "Z", which now yields "no match" instead of "partial match". In the
1058 case of pcre_dfa_exec(), "no match" is given if every matching path for the
1059 final character ended with (*FAIL).
1060
1061 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
1062 if the pattern had a "must contain" character that was already found in the
1063 earlier partial match, unless partial matching was again requested. For
1064 example, with the pattern /dog.(body)?/, the "must contain" character is
1065 "g". If the first part-match was for the string "dog", restarting with
1066 "sbody" failed. This bug has been fixed.
1067
1068 13. The string returned by pcre_dfa_exec() after a partial match has been
1069 changed so that it starts at the first inspected character rather than the
1070 first character of the match. This makes a difference only if the pattern
1071 starts with a lookbehind assertion or \b or \B (\K is not supported by
1072 pcre_dfa_exec()). It's an incompatible change, but it makes the two
1073 matching functions compatible, and I think it's the right thing to do.
1074
1075 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
1076 so that the demonstration program is easily available in environments where
1077 PCRE has not been installed from source.
1078
1079 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
1080 libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
1081 library.
1082
1083 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
1084 It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
1085 is not the first non-POSIX option to be added. Clearly some people find
1086 these options useful.
1087
1088 17. If a caller to the POSIX matching function regexec() passes a non-zero
1089 value for nmatch with a NULL value for pmatch, the value of
1090 nmatch is forced to zero.
1091
1092 18. RunGrepTest did not have a test for the availability of the -u option of
1093 the diff command, as RunTest does. It now checks in the same way as
1094 RunTest, and also checks for the -b option.
1095
1096 19. If an odd number of negated classes containing just a single character
1097 interposed, within parentheses, between a forward reference to a named
1098 subpattern and the definition of the subpattern, compilation crashed with
1099 an internal error, complaining that it could not find the referenced
1100 subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
1101 [The bug was that it was starting one character too far in when skipping
1102 over the character class, thus treating the ] as data rather than
1103 terminating the class. This meant it could skip too much.]
1104
1105 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
1106 /g option in pcretest when the pattern contains \K, which makes it possible
1107 to have an empty string match not at the start, even when the pattern is
1108 anchored. Updated pcretest and pcredemo to use this option.
1109
1110 21. If the maximum number of capturing subpatterns in a recursion was greater
1111 than the maximum at the outer level, the higher number was returned, but
1112 with unset values at the outer level. The correct (outer level) value is
1113 now given.
1114
1115 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
1116 PCRE did not set those parentheses (unlike Perl). I have now found a way to
1117 make it do so. The string so far is captured, making this feature
1118 compatible with Perl.
1119
1120 23. The tests have been re-organized, adding tests 11 and 12, to make it
1121 possible to check the Perl 5.10 features against Perl 5.10.
1122
1123 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
1124 pattern matches a fixed length string. PCRE did not allow this; now it
1125 does. Neither allows recursion.
1126
1127 25. I finally figured out how to implement a request to provide the minimum
1128 length of subject string that was needed in order to match a given pattern.
1129 (It was back references and recursion that I had previously got hung up
1130 on.) This code has now been added to pcre_study(); it finds a lower bound
1131 to the length of subject needed. It is not necessarily the greatest lower
1132 bound, but using it to avoid searching strings that are too short does give
1133 some useful speed-ups. The value is available to calling programs via
1134 pcre_fullinfo().
1135
1136 26. While implementing 25, I discovered to my embarrassment that pcretest had
1137 not been passing the result of pcre_study() to pcre_dfa_exec(), so the
1138 study optimizations had never been tested with that matching function.
1139 Oops. What is worse, even when it was passed study data, there was a bug in
1140 pcre_dfa_exec() that meant it never actually used it. Double oops. There
1141 were also very few tests of studied patterns with pcre_dfa_exec().
1142
1143 27. If (?| is used to create subpatterns with duplicate numbers, they are now
1144 allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
1145 on the other side of the coin, they are no longer allowed to have different
1146 names, because these cannot be distinguished in PCRE, and this has caused
1147 confusion. (This is a difference from Perl.)
1148
1149 28. When duplicate subpattern names are present (necessarily with different
1150 numbers, as required by 27 above), and a test is made by name in a
1151 conditional pattern, either for a subpattern having been matched, or for
1152 recursion in such a pattern, all the associated numbered subpatterns are
1153 tested, and the overall condition is true if the condition is true for any
1154 one of them. This is the way Perl works, and is also more like the way
1155 testing by number works.
1156
1157
1158 Version 7.9 11-Apr-09
1159 ---------------------
1160
1161 1. When building with support for bzlib/zlib (pcregrep) and/or readline
1162 (pcretest), all targets were linked against these libraries. This included
1163 libpcre, libpcreposix, and libpcrecpp, even though they do not use these
1164 libraries. This caused unwanted dependencies to be created. This problem
1165 has been fixed, and now only pcregrep is linked with bzlib/zlib and only
1166 pcretest is linked with readline.
1167
1168 2. The "typedef int BOOL" in pcre_internal.h that was included inside the
1169 "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
1170 moved outside it again, because FALSE and TRUE are already defined in AIX,
1171 but BOOL is not.
1172
1173 3. The pcre_config() function was treating the PCRE_MATCH_LIMIT and
1174 PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
1175
1176 4. The pcregrep documentation said spaces were inserted as well as colons (or
1177 hyphens) following file names and line numbers when outputting matching
1178 lines. This is not true; no spaces are inserted. I have also clarified the
1179 wording for the --colour (or --color) option.
1180
1181 5. In pcregrep, when --colour was used with -o, the list of matching strings
1182 was not coloured; this is different to GNU grep, so I have changed it to be
1183 the same.
1184
1185 6. When --colo(u)r was used in pcregrep, only the first matching substring in
1186 each matching line was coloured. Now it goes on to look for further matches
1187 of any of the test patterns, which is the same behaviour as GNU grep.
1188
1189 7. A pattern that could match an empty string could cause pcregrep to loop; it
1190 doesn't make sense to accept an empty string match in pcregrep, so I have
1191 locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
1192 seems to be how GNU grep behaves.
1193
1194 8. The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
1195 start or after a newline", because the conditional assertion was not being
1196 correctly handled. The rule now is that both the assertion and what follows
1197 in the first alternative must satisfy the test.
1198
1199 9. If auto-callout was enabled in a pattern with a conditional group whose
1200 condition was an assertion, PCRE could crash during matching, both with
1201 pcre_exec() and pcre_dfa_exec().
1202
1203 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
1204 used for matching.
1205
1206 11. Unicode property support in character classes was not working for
1207 characters (bytes) greater than 127 when not in UTF-8 mode.
1208
1209 12. Added the -M command line option to pcretest.
1210
1211 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
1212
1213 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
1214
1215 16. Added comments and documentation about mis-use of no_arg in the C++
1216 wrapper.
1217
1218 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
1219 from Martin Jerabek that uses macro names for all relevant character and
1220 string constants.
1221
1222 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
1223 SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
1224 SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
1225 these, but not everybody uses configure.
1226
1227 19. A conditional group that had only one branch was not being correctly
1228 recognized as an item that could match an empty string. This meant that an
1229 enclosing group might also not be so recognized, causing infinite looping
1230 (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
1231 with the subject "ab", where knowledge that the repeated group can match
1232 nothing is needed in order to break the loop.
1233
1234 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
1235 exec(), but without supplying a callout function, matching went wrong.
1236
1237 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
1238 leak if the size of the offset vector was greater than 30. When the vector
1239 is smaller, the saved offsets during recursion go onto a local stack
1240 vector, but for larger vectors malloc() is used. It was failing to free
1241 when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
1242 error, in fact).
1243
1244 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
1245 heapframe that is used only when UTF-8 support is enabled. This caused no
1246 problem, but was untidy.
1247
1248 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
1249 CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
1250 included within another project.
1251
1252 24. Steven Van Ingelgem's patches to add more options to the CMake support,
1253 slightly modified by me:
1254
1255 (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
1256 not building pcregrep.
1257
1258 (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
1259 if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
1260
1261 25. Forward references, both numeric and by name, in patterns that made use of
1262 duplicate group numbers, could behave incorrectly or give incorrect errors,
1263 because when scanning forward to find the reference group, PCRE was not
1264 taking into account the duplicate group numbers. A pattern such as
1265 ^X(?3)(a)(?|(b)|(q))(Y) is an example.
1266
1267 26. Changed a few more instances of "const unsigned char *" to USPTR, making
1268 the feature of a custom pointer more persuasive (as requested by a user).
1269
1270 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
1271 pcretest.c, inside #ifndefs, because it seems they are sometimes already
1272 pre-defined.
1273
1274 28. Added support for (*UTF8) at the start of a pattern.
1275
1276 29. Arrange for flags added by the "release type" setting in CMake to be shown
1277 in the configuration summary.
1278
1279
1280 Version 7.8 05-Sep-08
1281 ---------------------
1282
1283 1. Replaced UCP searching code with optimized version as implemented for Ad
1284 Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
1285 stage table and inline lookup instead of a function, giving speed ups of 2
1286 to 5 times on some simple patterns that I tested. Permission was given to
1287 distribute the MultiStage2.py script that generates the tables (it's not in
1288 the tarball, but is in the Subversion repository).
1289
1290 2. Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
1291 scripts.
1292
1293 3. Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
1294 a group with a zero qualifier. The result of the study could be incorrect,
1295 or the function might crash, depending on the pattern.
1296
1297 4. Caseless matching was not working for non-ASCII characters in back
1298 references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
1299 It now works when Unicode Property Support is available.
1300
1301 5. In pcretest, an escape such as \x{de} in the data was always generating
1302 a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
1303 non-UTF-8 mode. If the value is greater than 255, it gives a warning about
1304 truncation.
1305
1306 6. Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
1307
1308 7. Added two (int) casts to pcregrep when printing the difference of two
1309 pointers, in case they are 64-bit values.
1310
1311 8. Added comments about Mac OS X stack usage to the pcrestack man page and to
1312 test 2 if it fails.
1313
1314 9. Added PCRE_CALL_CONVENTION just before the names of all exported functions,
1315 and a #define of that name to empty if it is not externally set. This is to
1316 allow users of MSVC to set it if necessary.
1317
1318 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
1319 the convenience functions in the pcre_get.c source file.
1320
1321 11. An option change at the start of a pattern that had top-level alternatives
1322 could cause overwriting and/or a crash. This command provoked a crash in
1323 some environments:
1324
1325 printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
1326
1327 This potential security problem was recorded as CVE-2008-2371.
1328
1329 12. For a pattern where the match had to start at the beginning or immediately
1330 after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
1331 pcre_dfa_exec() could read past the end of the passed subject if there was
1332 no match. To help with detecting such bugs (e.g. with valgrind), I modified
1333 pcretest so that it places the subject at the end of its malloc-ed buffer.
1334
1335 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
1336 exec() might read past the end of the data buffer in UTF-8 mode.
1337
1338 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
1339 the data contained the byte 0x85 as part of a UTF-8 character within its
1340 first line. This applied both to normal and DFA matching.
1341
1342 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
1343 /^[^d]*?$/8 failed to match "abc".
1344
1345 16. Added a missing copyright notice to pcrecpp_internal.h.
1346
1347 17. Make it more clear in the documentation that values returned from
1348 pcre_exec() in ovector are byte offsets, not character counts.
1349
1350 18. Tidied a few places to stop certain compilers from issuing warnings.
1351
1352 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
1353 supplied by Stefan Weber. I made a further small update for 7.8 because
1354 there is a change of source arrangements: the pcre_searchfuncs.c module is
1355 replaced by pcre_ucd.c.
1356
1357
1358 Version 7.7 07-May-08
1359 ---------------------
1360
1361 1. Applied Craig's patch to sort out a long long problem: "If we can't convert
1362 a string to a long long, pretend we don't even have a long long." This is
1363 done by checking for the strtoq, strtoll, and _strtoi64 functions.
1364
1365 2. Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
1366 pre-7.6 versions, which defined a global no_arg variable instead of putting
1367 it in the RE class. (See also #8 below.)
1368
1369 3. Remove a line of dead code, identified by coverity and reported by Nuno
1370 Lopes.
1371
1372 4. Fixed two related pcregrep bugs involving -r with --include or --exclude:
1373
1374 (1) The include/exclude patterns were being applied to the whole pathnames
1375 of files, instead of just to the final components.
1376
1377 (2) If there was more than one level of directory, the subdirectories were
1378 skipped unless they satisfied the include/exclude conditions. This is
1379 inconsistent with GNU grep (and could even be seen as contrary to the
1380 pcregrep specification - which I improved to make it absolutely clear).
1381 The action now is always to scan all levels of directory, and just
1382 apply the include/exclude patterns to regular files.
1383
1384 5. Added the --include_dir and --exclude_dir patterns to pcregrep, and used
1385 --exclude_dir in the tests to avoid scanning .svn directories.
1386
1387 6. Applied Craig's patch to the QuoteMeta function so that it escapes the
1388 NUL character as backslash + 0 rather than backslash + NUL, because PCRE
1389 doesn't support NULs in patterns.
1390
1391 7. Added some missing "const"s to declarations of static tables in
1392 pcre_compile.c and pcre_dfa_exec.c.
1393
1394 8. Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
1395 caused by fix #2 above. (Subsequently also a second patch to fix the
1396 first patch. And a third patch - this was a messy problem.)
1397
1398 9. Applied Craig's patch to remove the use of push_back().
1399
1400 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
1401 matching function regexec().
1402
1403 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
1404 which, however, unlike Perl's \g{...}, are subroutine calls, not back
1405 references. PCRE supports relative numbers with this syntax (I don't think
1406 Oniguruma does).
1407
1408 12. Previously, a group with a zero repeat such as (...){0} was completely
1409 omitted from the compiled regex. However, this means that if the group
1410 was called as a subroutine from elsewhere in the pattern, things went wrong
1411 (an internal error was given). Such groups are now left in the compiled
1412 pattern, with a new opcode that causes them to be skipped at execution
1413 time.
1414
1415 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
1416 to the way PCRE behaves:
1417
1418 (a) A lone ] character is dis-allowed (Perl treats it as data).
1419
1420 (b) A back reference to an unmatched subpattern matches an empty string
1421 (Perl fails the current match path).
1422
1423 (c) A data ] in a character class must be notated as \] because if the
1424 first data character in a class is ], it defines an empty class. (In
1425 Perl it is not possible to have an empty class.) The empty class []
1426 never matches; it forces failure and is equivalent to (*FAIL) or (?!).
1427 The negative empty class [^] matches any one character, independently
1428 of the DOTALL setting.
1429
1430 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
1431 non-existent subpattern following a character class starting with ']' and
1432 containing () gave an internal compiling error instead of "reference to
1433 non-existent subpattern". Fortunately, when the pattern did exist, the
1434 compiled code was correct. (When scanning forwards to check for the
1435 existencd of the subpattern, it was treating the data ']' as terminating
1436 the class, so got the count wrong. When actually compiling, the reference
1437 was subsequently set up correctly.)
1438
1439 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
1440 it was being rejected as not supported by pcre_dfa_exec(), even though
1441 other assertions are supported. I have made pcre_dfa_exec() support
1442 (*FAIL).
1443
1444 16. The implementation of 13c above involved the invention of a new opcode,
1445 OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
1446 cannot be changed at match time, I realized I could make a small
1447 improvement to matching performance by compiling OP_ALLANY instead of
1448 OP_ANY for "." when DOTALL was set, and then removing the runtime tests
1449 on the OP_ANY path.
1450
1451 17. Compiling pcretest on Windows with readline support failed without the
1452 following two fixes: (1) Make the unistd.h include conditional on
1453 HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
1454
1455 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
1456 ncurses library to be included for pcretest when ReadLine support is
1457 requested, but also to allow for it to be overridden. This patch came from
1458 Daniel Bergström.
1459
1460 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
1461 as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
1462 any errors with the current Unicode tables. Thanks to Peter Kankowski for
1463 spotting this.
1464
1465
1466 Version 7.6 28-Jan-08
1467 ---------------------
1468
1469 1. A character class containing a very large number of characters with
1470 codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
1471 overflow.
1472
1473 2. Patch to cut out the "long long" test in pcrecpp_unittest when
1474 HAVE_LONG_LONG is not defined.
1475
1476 3. Applied Christian Ehrlicher's patch to update the CMake build files to
1477 bring them up to date and include new features. This patch includes:
1478
1479 - Fixed PH's badly added libz and libbz2 support.
1480 - Fixed a problem with static linking.
1481 - Added pcredemo. [But later removed - see 7 below.]
1482 - Fixed dftables problem and added an option.
1483 - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
1484 HAVE_LONG_LONG.
1485 - Added readline support for pcretest.
1486 - Added an listing of the option settings after cmake has run.
1487
1488 4. A user submitted a patch to Makefile that makes it easy to create
1489 "pcre.dll" under mingw when using Configure/Make. I added stuff to
1490 Makefile.am that cause it to include this special target, without
1491 affecting anything else. Note that the same mingw target plus all
1492 the other distribution libraries and programs are now supported
1493 when configuring with CMake (see 6 below) instead of with
1494 Configure/Make.
1495
1496 5. Applied Craig's patch that moves no_arg into the RE class in the C++ code.
1497 This is an attempt to solve the reported problem "pcrecpp::no_arg is not
1498 exported in the Windows port". It has not yet been confirmed that the patch
1499 solves the problem, but it does no harm.
1500
1501 6. Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
1502 NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
1503 with CMake, and also correct the comment about stack recursion.
1504
1505 7. Remove the automatic building of pcredemo from the ./configure system and
1506 from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
1507 of a program that users should build themselves after PCRE is installed, so
1508 building it automatically is not really right. What is more, it gave
1509 trouble in some build environments.
1510
1511 8. Further tidies to CMakeLists.txt from Sheri and Christian.
1512
1513
1514 Version 7.5 10-Jan-08
1515 ---------------------
1516
1517 1. Applied a patch from Craig: "This patch makes it possible to 'ignore'
1518 values in parens when parsing an RE using the C++ wrapper."
1519
1520 2. Negative specials like \S did not work in character classes in UTF-8 mode.
1521 Characters greater than 255 were excluded from the class instead of being
1522 included.
1523
1524 3. The same bug as (2) above applied to negated POSIX classes such as
1525 [:^space:].
1526
1527 4. PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
1528 defined or documented. It seems to have been a typo for PCRE_STATIC, so
1529 I have changed it.
1530
1531 5. The construct (?&) was not diagnosed as a syntax error (it referenced the
1532 first named subpattern) and a construct such as (?&a) would reference the
1533 first named subpattern whose name started with "a" (in other words, the
1534 length check was missing). Both these problems are fixed. "Subpattern name
1535 expected" is now given for (?&) (a zero-length name), and this patch also
1536 makes it give the same error for \k'' (previously it complained that that
1537 was a reference to a non-existent subpattern).
1538
1539 6. The erroneous patterns (?+-a) and (?-+a) give different error messages;
1540 this is right because (?- can be followed by option settings as well as by
1541 digits. I have, however, made the messages clearer.
1542
1543 7. Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
1544 than the number used in the conditional) now cause a compile-time error.
1545 This is actually not compatible with Perl, which accepts such patterns, but
1546 treats the conditional as always being FALSE (as PCRE used to), but it
1547 seems to me that giving a diagnostic is better.
1548
1549 8. Change "alphameric" to the more common word "alphanumeric" in comments
1550 and messages.
1551
1552 9. Fix two occurrences of "backslash" in comments that should have been
1553 "backspace".
1554
1555 10. Remove two redundant lines of code that can never be obeyed (their function
1556 was moved elsewhere).
1557
1558 11. The program that makes PCRE's Unicode character property table had a bug
1559 which caused it to generate incorrect table entries for sequences of
1560 characters that have the same character type, but are in different scripts.
1561 It amalgamated them into a single range, with the script of the first of
1562 them. In other words, some characters were in the wrong script. There were
1563 thirteen such cases, affecting characters in the following ranges:
1564
1565 U+002b0 - U+002c1
1566 U+0060c - U+0060d
1567 U+0061e - U+00612
1568 U+0064b - U+0065e
1569 U+0074d - U+0076d
1570 U+01800 - U+01805
1571 U+01d00 - U+01d77
1572 U+01d9b - U+01dbf
1573 U+0200b - U+0200f
1574 U+030fc - U+030fe
1575 U+03260 - U+0327f
1576 U+0fb46 - U+0fbb1
1577 U+10450 - U+1049d
1578
1579 12. The -o option (show only the matching part of a line) for pcregrep was not
1580 compatible with GNU grep in that, if there was more than one match in a
1581 line, it showed only the first of them. It now behaves in the same way as
1582 GNU grep.
1583
1584 13. If the -o and -v options were combined for pcregrep, it printed a blank
1585 line for every non-matching line. GNU grep prints nothing, and pcregrep now
1586 does the same. The return code can be used to tell if there were any
1587 non-matching lines.
1588
1589 14. Added --file-offsets and --line-offsets to pcregrep.
1590
1591 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
1592 infinitely looping recursion. The bug was that positive lookaheads were not
1593 being skipped when checking for a possible empty match (negative lookaheads
1594 and both kinds of lookbehind were skipped).
1595
1596 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
1597 inclusion of <windows.h> to before rather than after the definition of
1598 INVALID_FILE_ATTRIBUTES (patch from David Byron).
1599
1600 17. Specifying a possessive quantifier with a specific limit for a Unicode
1601 character property caused pcre_compile() to compile bad code, which led at
1602 runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
1603 are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
1604 caused the error; without that there was no problem.
1605
1606 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
1607
1608 19. Added --enable-pcretest-libreadline.
1609
1610 20. In pcrecpp.cc, the variable 'count' was incremented twice in
1611 RE::GlobalReplace(). As a result, the number of replacements returned was
1612 double what it should be. I removed one of the increments, but Craig sent a
1613 later patch that removed the other one (the right fix) and added unit tests
1614 that check the return values (which was not done before).
1615
1616 21. Several CMake things:
1617
1618 (1) Arranged that, when cmake is used on Unix, the libraries end up with
1619 the names libpcre and libpcreposix, not just pcre and pcreposix.
1620
1621 (2) The above change means that pcretest and pcregrep are now correctly
1622 linked with the newly-built libraries, not previously installed ones.
1623
1624 (3) Added PCRE_SUPPORT_LIBREADLINE, PCRE_SUPPORT_LIBZ, PCRE_SUPPORT_LIBBZ2.
1625
1626 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
1627 crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
1628 UTF-8 newline character). The key issue is that the pattern starts .*;
1629 this means that the match must be either at the beginning, or after a
1630 newline. The bug was in the code for advancing after a failed match and
1631 checking that the new position followed a newline. It was not taking
1632 account of UTF-8 characters correctly.
1633
1634 23. PCRE was behaving differently from Perl in the way it recognized POSIX
1635 character classes. PCRE was not treating the sequence [:...:] as a
1636 character class unless the ... were all letters. Perl, however, seems to
1637 allow any characters between [: and :], though of course it rejects as
1638 unknown any "names" that contain non-letters, because all the known class
1639 names consist only of letters. Thus, Perl gives an error for [[:1234:]],
1640 for example, whereas PCRE did not - it did not recognize a POSIX character
1641 class. This seemed a bit dangerous, so the code has been changed to be
1642 closer to Perl. The behaviour is not identical to Perl, because PCRE will
1643 diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
1644 treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
1645 Perl does, and where it didn't before.
1646
1647 24. Rewrite so as to remove the single use of %n from pcregrep because in some
1648 Windows environments %n is disabled by default.
1649
1650
1651 Version 7.4 21-Sep-07
1652 ---------------------
1653
1654 1. Change 7.3/28 was implemented for classes by looking at the bitmap. This
1655 means that a class such as [\s] counted as "explicit reference to CR or
1656 LF". That isn't really right - the whole point of the change was to try to
1657 help when there was an actual mention of one of the two characters. So now
1658 the change happens only if \r or \n (or a literal CR or LF) character is
1659 encountered.
1660
1661 2. The 32-bit options word was also used for 6 internal flags, but the numbers
1662 of both had grown to the point where there were only 3 bits left.
1663 Fortunately, there was spare space in the data structure, and so I have
1664 moved the internal flags into a new 16-bit field to free up more option
1665 bits.
1666
1667 3. The appearance of (?J) at the start of a pattern set the DUPNAMES option,
1668 but did not set the internal JCHANGED flag - either of these is enough to
1669 control the way the "get" function works - but the PCRE_INFO_JCHANGED
1670 facility is supposed to tell if (?J) was ever used, so now (?J) at the
1671 start sets both bits.
1672
1673 4. Added options (at build time, compile time, exec time) to change \R from
1674 matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
1675
1676 5. doc/pcresyntax.html was missing from the distribution.
1677
1678 6. Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
1679 compatibility, even though it is no longer used.
1680
1681 7. Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
1682 strtoull to pcrecpp.cc to select the available functions in WIN32 when the
1683 windows.h file is present (where different names are used). [This was
1684 reversed later after testing - see 16 below.]
1685
1686 8. Changed all #include <config.h> to #include "config.h". There were also
1687 some further <pcre.h> cases that I changed to "pcre.h".
1688
1689 9. When pcregrep was used with the --colour option, it missed the line ending
1690 sequence off the lines that it output.
1691
1692 10. It was pointed out to me that arrays of string pointers cause lots of
1693 relocations when a shared library is dynamically loaded. A technique of
1694 using a single long string with a table of offsets can drastically reduce
1695 these. I have refactored PCRE in four places to do this. The result is
1696 dramatic:
1697
1698 Originally: 290
1699 After changing UCP table: 187
1700 After changing error message table: 43
1701 After changing table of "verbs" 36
1702 After changing table of Posix names 22
1703
1704 Thanks to the folks working on Gregex for glib for this insight.
1705
1706 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
1707 unicode-properties was also set.
1708
1709 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
1710
1711 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
1712 checked only for CRLF.
1713
1714 14. Added casts to pcretest.c to avoid compiler warnings.
1715
1716 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
1717
1718 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
1719 and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
1720 entirely. This removes changes made in 7 above.
1721
1722 17. The CMake files have been updated, and there is now more information about
1723 building with CMake in the NON-UNIX-USE document.
1724
1725
1726 Version 7.3 28-Aug-07
1727 ---------------------
1728
1729 1. In the rejigging of the build system that eventually resulted in 7.1, the
1730 line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
1731 brackets there is not right, since it causes compilers to look for an
1732 installed pcre.h, not the version that is in the source that is being
1733 compiled (which of course may be different). I have changed it back to:
1734
1735 #include "pcre.h"
1736
1737 I have a vague recollection that the change was concerned with compiling in
1738 different directories, but in the new build system, that is taken care of
1739 by the VPATH setting the Makefile.
1740
1741 2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
1742 when the subject happened to end in the byte 0x85 (e.g. if the last
1743 character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
1744 characters but of course it shouldn't be taken as a newline when it is part
1745 of another character. The bug was that, for an unlimited repeat of . in
1746 not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
1747 characters when looking for a newline.
1748
1749 3. A small performance improvement in the DOTALL UTF-8 mode .* case.
1750
1751 4. Debugging: adjusted the names of opcodes for different kinds of parentheses
1752 in debug output.
1753
1754 5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
1755 long printing in the pcrecpp unittest when running under MinGW.
1756
1757 6. ESC_K was left out of the EBCDIC table.
1758
1759 7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
1760 parentheses; I made it 1000, which seemed large enough. Unfortunately, the
1761 limit also applies to "virtual nesting" when a pattern is recursive, and in
1762 this case 1000 isn't so big. I have been able to remove this limit at the
1763 expense of backing off one optimization in certain circumstances. Normally,
1764 when pcre_exec() would call its internal match() function recursively and
1765 immediately return the result unconditionally, it uses a "tail recursion"
1766 feature to save stack. However, when a subpattern that can match an empty
1767 string has an unlimited repetition quantifier, it no longer makes this
1768 optimization. That gives it a stack frame in which to save the data for
1769 checking that an empty string has been matched. Previously this was taken
1770 from the 1000-entry workspace that had been reserved. So now there is no
1771 explicit limit, but more stack is used.
1772
1773 8. Applied Daniel's patches to solve problems with the import/export magic
1774 syntax that is required for Windows, and which was going wrong for the
1775 pcreposix and pcrecpp parts of the library. These were overlooked when this
1776 problem was solved for the main library.
1777
1778 9. There were some crude static tests to avoid integer overflow when computing
1779 the size of patterns that contain repeated groups with explicit upper
1780 limits. As the maximum quantifier is 65535, the maximum group length was
1781 set at 30,000 so that the product of these two numbers did not overflow a
1782 32-bit integer. However, it turns out that people want to use groups that
1783 are longer than 30,000 bytes (though not repeat them that many times).
1784 Change 7.0/17 (the refactoring of the way the pattern size is computed) has
1785 made it possible to implement the integer overflow checks in a much more
1786 dynamic way, which I have now done. The artificial limitation on group
1787 length has been removed - we now have only the limit on the total length of
1788 the compiled pattern, which depends on the LINK_SIZE setting.
1789
1790 10. Fixed a bug in the documentation for get/copy named substring when
1791 duplicate names are permitted. If none of the named substrings are set, the
1792 functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
1793 empty string.
1794
1795 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
1796 instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
1797 because the ] is interpreted as the first data character and the
1798 terminating ] is not found. PCRE has been made compatible with Perl in this
1799 regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
1800 cause memory overwriting.
1801
1802 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
1803 string has been matched (to stop an infinite loop). It was not recognizing
1804 a conditional subpattern that could match an empty string if that
1805 subpattern was within another subpattern. For example, it looped when
1806 trying to match (((?(1)X|))*) but it was OK with ((?(1)X|)*) where the
1807 condition was not nested. This bug has been fixed.
1808
1809 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
1810 past the start of the subject in the presence of bytes with the top bit
1811 set, for example "\x8aBCD".
1812
1813 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
1814 (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
1815
1816 14. Optimized (?!) to (*FAIL).
1817
1818 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
1819 This restricts code points to be within the range 0 to 0x10FFFF, excluding
1820 the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
1821 full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
1822 does: it's just the validity check that is more restrictive.
1823
1824 16. Inserted checks for integer overflows during escape sequence (backslash)
1825 processing, and also fixed erroneous offset values for syntax errors during
1826 backslash processing.
1827
1828 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
1829 for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
1830
1831 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
1832 caused an overrun.
1833
1834 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
1835 something other than just ASCII characters) inside a group that had an
1836 unlimited repeat caused a loop at compile time (while checking to see
1837 whether the group could match an empty string).
1838
1839 20. Debugging a pattern containing \p or \P could cause a crash. For example,
1840 [\P{Any}] did so. (Error in the code for printing property names.)
1841
1842 21. An orphan \E inside a character class could cause a crash.
1843
1844 22. A repeated capturing bracket such as (A)? could cause a wild memory
1845 reference during compilation.
1846
1847 23. There are several functions in pcre_compile() that scan along a compiled
1848 expression for various reasons (e.g. to see if it's fixed length for look
1849 behind). There were bugs in these functions when a repeated \p or \P was
1850 present in the pattern. These operators have additional parameters compared
1851 with \d, etc, and these were not being taken into account when moving along
1852 the compiled data. Specifically:
1853
1854 (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
1855 length.
1856
1857 (b) An item such as \pL+ within a repeated group could cause crashes or
1858 loops.
1859
1860 (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
1861 "reference to non-existent subpattern" error.
1862
1863 (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
1864
1865 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
1866 characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
1867
1868 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
1869
1870 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
1871 character were causing crashes (broken optimization).
1872
1873 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
1874 \p or \P) caused a compile-time loop.
1875
1876 28. More problems have arisen in unanchored patterns when CRLF is a valid line
1877 break. For example, the unstudied pattern [\r\n]A does not match the string
1878 "\r\nA" because change 7.0/46 below moves the current point on by two
1879 characters after failing to match at the start. However, the pattern \nA
1880 *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
1881 the same is true. There doesn't seem any very clean way out of this, but
1882 what I have chosen to do makes the common cases work: PCRE now takes note
1883 of whether there can be an explicit match for \r or \n anywhere in the
1884 pattern, and if so, 7.0/46 no longer applies. As part of this change,
1885 there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
1886 pattern has explicit CR or LF references.
1887
1888 29. Added (*CR) etc for changing newline setting at start of pattern.
1889
1890
1891 Version 7.2 19-Jun-07
1892 ---------------------
1893
1894 1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
1895 which is apparently normally available under Windows.
1896
1897 2. Re-jig the pcregrep tests with different newline settings in an attempt
1898 to make them independent of the local environment's newline setting.
1899
1900 3. Add code to configure.ac to remove -g from the CFLAGS default settings.
1901
1902 4. Some of the "internals" tests were previously cut out when the link size
1903 was not 2, because the output contained actual offsets. The recent new
1904 "Z" feature of pcretest means that these can be cut out, making the tests
1905 usable with all link sizes.
1906
1907 5. Implemented Stan Switzer's goto replacement for longjmp() when not using
1908 stack recursion. This gives a massive performance boost under BSD, but just
1909 a small improvement under Linux. However, it saves one field in the frame
1910 in all cases.
1911
1912 6. Added more features from the forthcoming Perl 5.10:
1913
1914 (a) (?-n) (where n is a string of digits) is a relative subroutine or
1915 recursion call. It refers to the nth most recently opened parentheses.
1916
1917 (b) (?+n) is also a relative subroutine call; it refers to the nth next
1918 to be opened parentheses.
1919
1920 (c) Conditions that refer to capturing parentheses can be specified
1921 relatively, for example, (?(-2)... or (?(+3)...
1922
1923 (d) \K resets the start of the current match so that everything before
1924 is not part of it.
1925
1926 (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
1927
1928 (f) \g{name} is another synonym - part of Perl 5.10's unification of
1929 reference syntax.
1930
1931 (g) (?| introduces a group in which the numbering of parentheses in each
1932 alternative starts with the same number.
1933
1934 (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
1935
1936 7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
1937 PCRE_INFO_JCHANGED.
1938
1939 8. A pattern such as (.*(.)?)* caused pcre_exec() to fail by either not
1940 terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
1941 for detecting groups that can match an empty string.
1942
1943 9. A pattern with a very large number of alternatives (more than several
1944 hundred) was running out of internal workspace during the pre-compile
1945 phase, where pcre_compile() figures out how much memory will be needed. A
1946 bit of new cunning has reduced the workspace needed for groups with
1947 alternatives. The 1000-alternative test pattern now uses 12 bytes of
1948 workspace instead of running out of the 4096 that are available.
1949
1950 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
1951
1952 11. Applied patch from Google to remove an optimization that didn't quite work.
1953 The report of the bug said:
1954
1955 pcrecpp::RE("a*").FullMatch("aaa") matches, while
1956 pcrecpp::RE("a*?").FullMatch("aaa") does not, and
1957 pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
1958
1959 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
1960 it matched the wrong number of bytes.
1961
1962
1963 Version 7.1 24-Apr-07
1964 ---------------------
1965
1966 1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
1967 that is more "standard", making use of automake and other Autotools. There
1968 is some re-arrangement of the files and adjustment of comments consequent
1969 on this.
1970
1971 2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
1972 for recursive directory scanning broke on some systems because the files
1973 are not scanned in any specific order and on different systems the order
1974 was different. A call to "sort" has been inserted into RunGrepTest for the
1975 approprate test as a short-term fix. In the longer term there may be an
1976 alternative.
1977
1978 3. I had an email from Eric Raymond about problems translating some of PCRE's
1979 man pages to HTML (despite the fact that I distribute HTML pages, some
1980 people do their own conversions for various reasons). The problems
1981 concerned the use of low-level troff macros .br and .in. I have therefore
1982 removed all such uses from the man pages (some were redundant, some could
1983 be replaced by .nf/.fi pairs). The 132html script that I use to generate
1984 HTML has been updated to handle .nf/.fi and to complain if it encounters
1985 .br or .in.
1986
1987 4. Updated comments in configure.ac that get placed in config.h.in and also
1988 arranged for config.h to be included in the distribution, with the name
1989 config.h.generic, for the benefit of those who have to compile without
1990 Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
1991
1992 5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
1993 Weber: (1) pcre_internal.h was missing some function renames; (2) updated
1994 makevp.bat for the current PCRE, using the additional files
1995 makevp_c.txt, makevp_l.txt, and pcregexp.pas.
1996
1997 6. A Windows user reported a minor discrepancy with test 2, which turned out
1998 to be caused by a trailing space on an input line that had got lost in his
1999 copy. The trailing space was an accident, so I've just removed it.
2000
2001 7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
2002 that is needed.
2003
2004 8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
2005 as "const" (a) because they are and (b) because it helps the PHP
2006 maintainers who have recently made a script to detect big data structures
2007 in the php code that should be moved to the .rodata section. I remembered
2008 to update Builducptable as well, so it won't revert if ucptable.h is ever
2009 re-created.
2010
2011 9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
2012 pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
2013 order to be able to cut out the UTF-8 tables in the latter when UTF-8
2014 support is not required. This saves 1.5-2K of code, which is important in
2015 some applications.
2016
2017 Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
2018 so as not to refer to the tables, even though these functions will never be
2019 called when UTF-8 support is disabled. Otherwise there are problems with a
2020 shared library.
2021
2022 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
2023
2024 (a) It was defining its arguments as char * instead of void *.
2025
2026 (b) It was assuming that all moves were upwards in memory; this was true
2027 a long time ago when I wrote it, but is no longer the case.
2028
2029 The emulated memove() is provided for those environments that have neither
2030 memmove() nor bcopy(). I didn't think anyone used it these days, but that
2031 is clearly not the case, as these two bugs were recently reported.
2032
2033 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
2034 and Detrail to create the HTML documentation, the .txt form of the man
2035 pages, and it removes trailing spaces from listed files. It also creates
2036 pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
2037 case, it wraps all the #defines with #ifndefs. This script should be run
2038 before "make dist".
2039
2040 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
2041 with Unicode property support.
2042
2043 (a) For a maximizing quantifier, if the two different cases of the
2044 character were of different lengths in their UTF-8 codings (there are
2045 some cases like this - I found 11), and the matching function had to
2046 back up over a mixture of the two cases, it incorrectly assumed they
2047 were both the same length.
2048
2049 (b) When PCRE was configured to use the heap rather than the stack for
2050 recursion during matching, it was not correctly preserving the data for
2051 the other case of a UTF-8 character when checking ahead for a match
2052 while processing a minimizing repeat. If the check also involved
2053 matching a wide character, but failed, corruption could cause an
2054 erroneous result when trying to check for a repeat of the original
2055 character.
2056
2057 13. Some tidying changes to the testing mechanism:
2058
2059 (a) The RunTest script now detects the internal link size and whether there
2060 is UTF-8 and UCP support by running ./pcretest -C instead of relying on
2061 values substituted by "configure". (The RunGrepTest script already did
2062 this for UTF-8.) The configure.ac script no longer substitutes the
2063 relevant variables.
2064
2065 (b) The debugging options /B and /D in pcretest show the compiled bytecode
2066 with length and offset values. This means that the output is different
2067 for different internal link sizes. Test 2 is skipped for link sizes
2068 other than 2 because of this, bypassing the problem. Unfortunately,
2069 there was also a test in test 3 (the locale tests) that used /B and
2070 failed for link sizes other than 2. Rather than cut the whole test out,
2071 I have added a new /Z option to pcretest that replaces the length and
2072 offset values with spaces. This is now used to make test 3 independent
2073 of link size. (Test 2 will be tidied up later.)
2074
2075 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
2076 segmentation fault instead of returning the appropriate error message.
2077
2078 15. In multiline mode when the newline sequence was set to "any", the pattern
2079 ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
2080 This doesn't seem right; it now treats the CRLF combination as the line
2081 ending, and so does not match in that case. It's only a pattern such as ^$
2082 that would hit this one: something like ^ABC$ would have failed after \r
2083 and then tried again after \r\n.
2084
2085 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
2086 in an attempt to make files that differ only in their line terminators
2087 compare equal. This works on Linux.
2088
2089 17. Under certain error circumstances pcregrep might try to free random memory
2090 as it exited. This is now fixed, thanks to valgrind.
2091
2092 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
2093 "abc\r\n\r\n", it found an unwanted second match after the second \r. This
2094 was because its rules for how to advance for /g after matching an empty
2095 string at the end of a line did not allow for this case. They now check for
2096 it specially.
2097
2098 20. pcretest is supposed to handle patterns and data of any length, by
2099 extending its buffers when necessary. It was getting this wrong when the
2100 buffer for a data line had to be extended.
2101
2102 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
2103 CRLF as a newline sequence.
2104
2105 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
2106 out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
2107 I have nevertheless tidied it up.
2108
2109 23. Added some casts to kill warnings from HP-UX ia64 compiler.
2110
2111 24. Added a man page for pcre-config.
2112
2113
2114 Version 7.0 19-Dec-06
2115 ---------------------
2116
2117 1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
2118 moving to gcc 4.1.1.
2119
2120 2. The -S option for pcretest uses setrlimit(); I had omitted to #include
2121 sys/time.h, which is documented as needed for this function. It doesn't
2122 seem to matter on Linux, but it showed up on some releases of OS X.
2123
2124 3. It seems that there are systems where bytes whose values are greater than
2125 127 match isprint() in the "C" locale. The "C" locale should be the
2126 default when a C program starts up. In most systems, only ASCII printing
2127 characters match isprint(). This difference caused the output from pcretest
2128 to vary, making some of the tests fail. I have changed pcretest so that:
2129
2130 (a) When it is outputting text in the compiled version of a pattern, bytes
2131 other than 32-126 are always shown as hex escapes.
2132
2133 (b) When it is outputting text that is a matched part of a subject string,
2134 it does the same, unless a different locale has been set for the match
2135 (using the /L modifier). In this case, it uses isprint() to decide.
2136
2137 4. Fixed a major bug that caused incorrect computation of the amount of memory
2138 required for a compiled pattern when options that changed within the
2139 pattern affected the logic of the preliminary scan that determines the
2140 length. The relevant options are -x, and -i in UTF-8 mode. The result was
2141 that the computed length was too small. The symptoms of this bug were
2142 either the PCRE error "internal error: code overflow" from pcre_compile(),
2143 or a glibc crash with a message such as "pcretest: free(): invalid next
2144 size (fast)". Examples of patterns that provoked this bug (shown in
2145 pcretest format) are:
2146
2147 /(?-x: )/x
2148 /(?x)(?-x: \s*#\s*)/
2149 /((?i)[\x{c0}])/8
2150 /(?i:[\x{c0}])/8
2151
2152 HOWEVER: Change 17 below makes this fix obsolete as the memory computation
2153 is now done differently.
2154
2155 5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
2156 wrapper classes; (b) implement a new function in the C++ scanner that is
2157 more efficient than the old way of doing things because it avoids levels of
2158 recursion in the regex matching; (c) add a paragraph to the documentation
2159 for the FullMatch() function.
2160
2161 6. The escape sequence \n was being treated as whatever was defined as
2162 "newline". Not only was this contrary to the documentation, which states
2163 that \n is character 10 (hex 0A), but it also went horribly wrong when
2164 "newline" was defined as CRLF. This has been fixed.
2165
2166 7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
2167 was being set to -1 for the "end of line" case (supposedly a value that no
2168 character can have). Though this value is never used (the check for end of
2169 line is "zero bytes in current character"), it caused compiler complaints.
2170 I've changed it to 0xffffffff.
2171
2172 8. In pcre_version.c, the version string was being built by a sequence of
2173 C macros that, in the event of PCRE_PRERELEASE being defined as an empty
2174 string (as it is for production releases) called a macro with an empty
2175 argument. The C standard says the result of this is undefined. The gcc
2176 compiler treats it as an empty string (which was what was wanted) but it is
2177 reported that Visual C gives an error. The source has been hacked around to
2178 avoid this problem.
2179
2180 9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
2181 builds of pcretest, and changed the call to _setmode() to use _O_BINARY
2182 instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
2183 of them did).
2184
2185 10. Originally, pcretest opened its input and output without "b"; then I was
2186 told that "b" was needed in some environments, so it was added for release
2187 5.0 to both the input and output. (It makes no difference on Unix-like
2188 systems.) Later I was told that it is wrong for the input on Windows. I've
2189 now abstracted the modes into two macros, to make it easier to fiddle with
2190 them, and removed "b" from the input mode under Windows.
2191
2192 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
2193
2194 12. Added -help and --help to pcretest as an official way of being reminded
2195 of the options.
2196
2197 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
2198 and pcrecpp.cc because they annoy compilers at high warning levels.
2199
2200 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
2201
2202 15. Fixed an occurrence of == in configure.ac that should have been = (shell
2203 scripts are not C programs :-) and which was not noticed because it works
2204 on Linux.
2205
2206 16. pcretest is supposed to handle any length of pattern and data line (as one
2207 line or as a continued sequence of lines) by extending its input buffer if
2208 necessary. This feature was broken for very long pattern lines, leading to
2209 a string of junk being passed to pcre_compile() if the pattern was longer
2210 than about 50K.
2211
2212 17. I have done a major re-factoring of the way pcre_compile() computes the
2213 amount of memory needed for a compiled pattern. Previously, there was code
2214 that made a preliminary scan of the pattern in order to do this. That was
2215 OK when PCRE was new, but as the facilities have expanded, it has become
2216 harder and harder to keep it in step with the real compile phase, and there
2217 have been a number of bugs (see for example, 4 above). I have now found a
2218 cunning way of running the real compile function in a "fake" mode that
2219 enables it to compute how much memory it would need, while actually only
2220 ever using a few hundred bytes of working memory and without too many
2221 tests of the mode. This should make future maintenance and development
2222 easier. A side effect of this work is that the limit of 200 on the nesting
2223 depth of parentheses has been removed (though this was never a serious
2224 limitation, I suspect). However, there is a downside: pcre_compile() now
2225 runs more slowly than before (30% or more, depending on the pattern). I
2226 hope this isn't a big issue. There is no effect on runtime performance.
2227
2228 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
2229 newline (only possible for the last line of a file) and it was a
2230 pattern that set a locale (followed by /Lsomething), pcretest crashed.
2231
2232 19. Added additional timing features to pcretest. (1) The -tm option now times
2233 matching only, not compiling. (2) Both -t and -tm can be followed, as a
2234 separate command line item, by a number that specifies the number of
2235 repeats to use when timing. The default is 50000; this gives better
2236 precision, but takes uncomfortably long for very large patterns.
2237
2238 20. Extended pcre_study() to be more clever in cases where a branch of a
2239 subpattern has no definite first character. For example, (a*|b*)[cd] would
2240 previously give no result from pcre_study(). Now it recognizes that the
2241 first character must be a, b, c, or d.
2242
2243 21. There was an incorrect error "recursive call could loop indefinitely" if
2244 a subpattern (or the entire pattern) that was being tested for matching an
2245 empty string contained only one non-empty item after a nested subpattern.
2246 For example, the pattern (?>\x{100}*)\d(?R) provoked this error
2247 incorrectly, because the \d was being skipped in the check.
2248
2249 22. The pcretest program now has a new pattern option /B and a command line
2250 option -b, which is equivalent to adding /B to every pattern. This causes
2251 it to show the compiled bytecode, without the additional information that
2252 -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
2253 is the same as /B/I).
2254
2255 23. A new optimization is now able automatically to treat some sequences such
2256 as a*b as a*+b. More specifically, if something simple (such as a character
2257 or a simple class like \d) has an unlimited quantifier, and is followed by
2258 something that cannot possibly match the quantified thing, the quantifier
2259 is automatically "possessified".
2260
2261 24. A recursive reference to a subpattern whose number was greater than 39
2262 went wrong under certain circumstances in UTF-8 mode. This bug could also
2263 have affected the operation of pcre_study().
2264
2265 25. Realized that a little bit of performance could be had by replacing
2266 (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
2267
2268 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
2269
2270 27. Possessive quantifiers such as a++ were previously implemented by turning
2271 them into atomic groups such as ($>a+). Now they have their own opcodes,
2272 which improves performance. This includes the automatically created ones
2273 from 23 above.
2274
2275 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
2276 lookahead was broken if it was not anchored. PCRE was mistakenly expecting
2277 the first matched character to be a colon. This applied both to named and
2278 numbered groups.
2279
2280 29. The ucpinternal.h header file was missing its idempotency #ifdef.
2281
2282 30. I was sent a "project" file called libpcre.a.dev which I understand makes
2283 building PCRE on Windows easier, so I have included it in the distribution.
2284
2285 31. There is now a check in pcretest against a ridiculously large number being
2286 returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
2287 loop, the loop is abandoned.
2288
2289 32. Forward references to subpatterns in conditions such as (?(2)...) where
2290 subpattern 2 is defined later cause pcre_compile() to search forwards in
2291 the pattern for the relevant set of parentheses. This search went wrong
2292 when there were unescaped parentheses in a character class, parentheses
2293 escaped with \Q...\E, or parentheses in a #-comment in /x mode.
2294
2295 33. "Subroutine" calls and backreferences were previously restricted to
2296 referencing subpatterns earlier in the regex. This restriction has now
2297 been removed.
2298
2299 34. Added a number of extra features that are going to be in Perl 5.10. On the
2300 whole, these are just syntactic alternatives for features that PCRE had
2301 previously implemented using the Python syntax or my own invention. The
2302 other formats are all retained for compatibility.
2303
2304 (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
2305 as (?P<name>...). The new forms, as well as being in Perl 5.10, are
2306 also .NET compatible.
2307
2308 (b) A recursion or subroutine call to a named group can now be defined as
2309 (?&name) as well as (?P>name).
2310
2311 (c) A backreference to a named group can now be defined as \k<name> or
2312 \k'name' as well as (?P=name). The new forms, as well as being in Perl
2313 5.10, are also .NET compatible.
2314
2315 (d) A conditional reference to a named group can now use the syntax
2316 (?(<name>) or (?('name') as well as (?(name).
2317
2318 (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
2319 groups (named and numbered) that are never evaluated inline, but can be
2320 called as "subroutines" from elsewhere. In effect, the DEFINE condition
2321 is always false. There may be only one alternative in such a group.
2322
2323 (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
2324 as the simple (?(R). The condition is true only if the most recent
2325 recursion is that of the given number or name. It does not search out
2326 through the entire recursion stack.
2327
2328 (g) The escape \gN or \g{N} has been added, where N is a positive or
2329 negative number, specifying an absolute or relative reference.
2330
2331 35. Tidied to get rid of some further signed/unsigned compiler warnings and
2332 some "unreachable code" warnings.
2333
2334 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
2335 things, this adds five new scripts.
2336
2337 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
2338 There were also incompatibilities regarding the handling of \Q..\E inside
2339 character classes, for example with patterns like [\Qa\E-\Qz\E] where the
2340 hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
2341
2342 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
2343 matches an empty string, and forcibly breaks the loop. There were bugs in
2344 this code in non-simple cases. For a pattern such as ^(a()*)* matched
2345 against aaaa the result was just "a" rather than "aaaa", for example. Two
2346 separate and independent bugs (that affected different cases) have been
2347 fixed.
2348
2349 39. Refactored the code to abolish the use of different opcodes for small
2350 capturing bracket numbers. This is a tidy that I avoided doing when I
2351 removed the limit on the number of capturing brackets for 3.5 back in 2001.
2352 The new approach is not only tidier, it makes it possible to reduce the
2353 memory needed to fix the previous bug (38).
2354
2355 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
2356 sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
2357 processing dot, circumflex, or dollar metacharacters, or #-comments in /x
2358 mode.
2359
2360 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
2361 report.
2362
2363 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
2364 copy construction and assignment in the C++ wrapper.
2365
2366 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
2367 couple of bugs that could have given wrong results in the "--newline=crlf"
2368 case.
2369
2370 44. Added a number of casts and did some reorganization of signed/unsigned int
2371 variables following suggestions from Dair Grant. Also renamed the variable
2372 "this" as "item" because it is a C++ keyword.
2373
2374 45. Arranged for dftables to add
2375
2376 #include "pcre_internal.h"
2377
2378 to pcre_chartables.c because without it, gcc 4.x may remove the array
2379 definition from the final binary if PCRE is built into a static library and
2380 dead code stripping is activated.
2381
2382 46. For an unanchored pattern, if a match attempt fails at the start of a
2383 newline sequence, and the newline setting is CRLF or ANY, and the next two
2384 characters are CRLF, advance by two characters instead of one.
2385
2386
2387 Version 6.7 04-Jul-06
2388 ---------------------
2389
2390 1. In order to handle tests when input lines are enormously long, pcretest has
2391 been re-factored so that it automatically extends its buffers when
2392 necessary. The code is crude, but this _is_ just a test program. The
2393 default size has been increased from 32K to 50K.
2394
2395 2. The code in pcre_study() was using the value of the re argument before
2396 testing it for NULL. (Of course, in any sensible call of the function, it
2397 won't be NULL.)
2398
2399 3. The memmove() emulation function in pcre_internal.h, which is used on
2400 systems that lack both memmove() and bcopy() - that is, hardly ever -
2401 was missing a "static" storage class specifier.
2402
2403 4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
2404 containing an extended class (one that cannot be represented by a bitmap
2405 because it contains high-valued characters or Unicode property items, e.g.
2406 [\pZ]). Almost always one would set UTF-8 mode when processing such a
2407 pattern, but PCRE should not loop if you do not (it no longer does).
2408 [Detail: two cases were found: (a) a repeated subpattern containing an
2409 extended class; (b) a recursive reference to a subpattern that followed a
2410 previous extended class. It wasn't skipping over the extended class
2411 correctly when UTF-8 mode was not set.]
2412
2413 5. A negated single-character class was not being recognized as fixed-length
2414 in lookbehind assertions such as (?<=[^f]), leading to an incorrect
2415 compile error "lookbehind assertion is not fixed length".
2416
2417 6. The RunPerlTest auxiliary script was showing an unexpected difference
2418 between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
2419 write a Perl script that can interpret lines of an input file either as
2420 byte characters or as UTF-8, which is what "perltest" was being required to
2421 do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
2422 can't do is switch easily at run time between having the "use utf8;" pragma
2423 or not. In the end, I fudged it by using the RunPerlTest script to insert
2424 "use utf8;" explicitly for the UTF-8 tests.
2425
2426 7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
2427 the end of the subject string, contrary to the documentation and to what
2428 Perl does. This was true of both matching functions. Now it matches only at
2429 the start of the subject and immediately after *internal* newlines.
2430
2431 8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
2432 a pointer to an int instead of a pointer to an unsigned long int. This
2433 caused problems on 64-bit systems.
2434
2435 9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
2436 instance of the 'standard' template library not being so standard".
2437
2438 10. There was no check on the number of named subpatterns nor the maximum
2439 length of a subpattern name. The product of these values is used to compute
2440 the size of the memory block for a compiled pattern. By supplying a very
2441 long subpattern name and a large number of named subpatterns, the size
2442 computation could be caused to overflow. This is now prevented by limiting
2443 the length of names to 32 characters, and the number of named subpatterns
2444 to 10,000.
2445
2446 11. Subpatterns that are repeated with specific counts have to be replicated in
2447 the compiled pattern. The size of memory for this was computed from the
2448 length of the subpattern and the repeat count. The latter is limited to
2449 65535, but there was no limit on the former, meaning that integer overflow
2450 could in principle occur. The compiled length of a repeated subpattern is
2451 now limited to 30,000 bytes in order to prevent this.
2452
2453 12. Added the optional facility to have named substrings with the same name.
2454
2455 13. Added the ability to use a named substring as a condition, using the
2456 Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
2457 are numbers (not recommended). Forward references are permitted.
2458
2459 14. Added forward references in named backreferences (if you see what I mean).
2460
2461 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
2462 pattern could run off the end of the subject. For example, the pattern
2463 "(?s)(.{1,5})"8 did this with the subject "ab".
2464
2465 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
2466 PCRE_CASELESS was set when matching characters that were quantified with ?
2467 or *.
2468
2469 17. A character class other than a single negated character that had a minimum
2470 but no maximum quantifier - for example [ab]{6,} - was not handled
2471 correctly by pce_dfa_exec(). It would match only one character.
2472
2473 18. A valid (though odd) pattern that looked like a POSIX character
2474 class but used an invalid character after [ (for example [[,abc,]]) caused
2475 pcre_compile() to give the error "Failed: internal error: code overflow" or
2476 in some cases to crash with a glibc free() error. This could even happen if
2477 the pattern terminated after [[ but there just happened to be a sequence of
2478 letters, a binary zero, and a closing ] in the memory that followed.
2479
2480 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
2481 over the years. Originally (before any Unicode support), just the bottom 8
2482 bits were taken. Thus, for example, \500 really meant \100. Nowadays the
2483 output from "man perlunicode" includes this:
2484
2485 The regular expression compiler produces polymorphic opcodes. That
2486 is, the pattern adapts to the data and automatically switches to
2487 the Unicode character scheme when presented with Unicode data--or
2488 instead uses a traditional byte scheme when presented with byte
2489 data.
2490
2491 Sadly, a wide octal escape does not cause a switch, and in a string with
2492 no other multibyte characters, these octal escapes are treated as before.
2493 Thus, in Perl, the pattern /\500/ actually matches \100 but the pattern
2494 /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
2495 Unicode string.
2496
2497 I have not perpetrated such confusion in PCRE. Up till now, it took just
2498 the bottom 8 bits, as in old Perl. I have now made octal escapes with
2499 values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
2500 translate to the appropriate multibyte character.
2501
2502 29. Applied some refactoring to reduce the number of warnings from Microsoft
2503 and Borland compilers. This has included removing the fudge introduced
2504 seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
2505 a warning about an unused variable.
2506
2507 21. PCRE has not included VT (character 0x0b) in the set of whitespace
2508 characters since release 4.0, because Perl (from release 5.004) does not.
2509 [Or at least, is documented not to: some releases seem to be in conflict
2510 with the documentation.] However, when a pattern was studied with
2511 pcre_study() and all its branches started with \s, PCRE still included VT
2512 as a possible starting character. Of course, this did no harm; it just
2513 caused an unnecessary match attempt.
2514
2515 22. Removed a now-redundant internal flag bit that recorded the fact that case
2516 dependency changed within the pattern. This was once needed for "required
2517 byte" processing, but is no longer used. This recovers a now-scarce options
2518 bit. Also moved the least significant internal flag bit to the most-
2519 significant bit of the word, which was not previously used (hangover from
2520 the days when it was an int rather than a uint) to free up another bit for
2521 the future.
2522
2523 23. Added support for CRLF line endings as well as CR and LF. As well as the
2524 default being selectable at build time, it can now be changed at runtime
2525 via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
2526 specify that it is scanning data with non-default line endings.
2527
2528 24. Changed the definition of CXXLINK to make it agree with the definition of
2529 LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
2530
2531 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
2532 recursions. This makes a big different to stack usage for some patterns.
2533
2534 26. If a subpattern containing a named recursion or subroutine reference such
2535 as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
2536 the space required for the compiled pattern went wrong and gave too small a
2537 value. Depending on the environment, this could lead to "Failed: internal
2538 error: code overflow at offset 49" or "glibc detected double free or
2539 corruption" errors.
2540
2541 27. Applied patches from Google (a) to support the new newline modes and (b) to
2542 advance over multibyte UTF-8 characters in GlobalReplace.
2543
2544 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
2545 difference for some implementation of PCRE in some Windows version.
2546
2547 29. Added some extra testing facilities to pcretest:
2548
2549 \q<number> in a data line sets the "match limit" value
2550 \Q<number> in a data line sets the "match recursion limt" value
2551 -S <number> sets the stack size, where <number> is in megabytes
2552
2553 The -S option isn't available for Windows.
2554
2555
2556 Version 6.6 06-Feb-06
2557 ---------------------
2558
2559 1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
2560 in pcreposix.h. I have copied the definition from pcre.h.
2561
2562 2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
2563 because pcre.h is no longer a built file.
2564
2565 3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
2566 not normally included in the compiled code.
2567
2568
2569 Version 6.5 01-Feb-06
2570 ---------------------
2571
2572 1. When using the partial match feature with pcre_dfa_exec(), it was not
2573 anchoring the second and subsequent partial matches at the new starting
2574 point. This could lead to incorrect results. For example, with the pattern
2575 /1234/, partially matching against "123" and then "a4" gave a match.
2576
2577 2. Changes to pcregrep:
2578
2579 (a) All non-match returns from pcre_exec() were being treated as failures
2580 to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
2581 error message is output. Some extra information is given for the
2582 PCRE_ERROR_MATCHLIMIT and PCRE_ERROR_RECURSIONLIMIT errors, which are
2583 probably the only errors that are likely to be caused by users (by
2584 specifying a regex that has nested indefinite repeats, for instance).
2585 If there are more than 20 of these errors, pcregrep is abandoned.
2586
2587 (b) A binary zero was treated as data while matching, but terminated the
2588 output line if it was written out. This has been fixed: binary zeroes
2589 are now no different to any other data bytes.
2590
2591 (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
2592 used to set a locale for matching. The --locale=xxxx long option has
2593 been added (no short equivalent) to specify a locale explicitly on the
2594 pcregrep command, overriding the environment variables.
2595
2596 (d) When -B was used with -n, some line numbers in the output were one less
2597 than they should have been.
2598
2599 (e) Added the -o (--only-matching) option.
2600
2601 (f) If -A or -C was used with -c (count only), some lines of context were
2602 accidentally printed for the final match.
2603
2604 (g) Added the -H (--with-filename) option.
2605
2606 (h) The combination of options -rh failed to suppress file names for files
2607 that were found from directory arguments.
2608
2609 (i) Added the -D (--devices) and -d (--directories) options.
2610
2611 (j) Added the -F (--fixed-strings) option.
2612
2613 (k) Allow "-" to be used as a file name for -f as well as for a data file.
2614
2615 (l) Added the --colo(u)r option.
2616
2617 (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
2618 is not present by default.
2619
2620 3. A nasty bug was discovered in the handling of recursive patterns, that is,
2621 items such as (?R) or (?1), when the recursion could match a number of
2622 alternatives. If it matched one of the alternatives, but subsequently,
2623 outside the recursion, there was a failure, the code tried to back up into
2624 the recursion. However, because of the way PCRE is implemented, this is not
2625 possible, and the result was an incorrect result from the match.
2626
2627 In order to prevent this happening, the specification of recursion has
2628 been changed so that all such subpatterns are automatically treated as
2629 atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
2630
2631 4. I had overlooked the fact that, in some locales, there are characters for
2632 which isalpha() is true but neither isupper() nor islower() are true. In
2633 the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
2634 and ordfeminine) are like this. This affected the treatment of \w and \W
2635 when they appeared in character classes, but not when they appeared outside
2636 a character class. The bit map for "word" characters is now created
2637 separately from the results of isalnum() instead of just taking it from the
2638 upper, lower, and digit maps. (Plus the underscore character, of course.)
2639
2640 5. The above bug also affected the handling of POSIX character classes such as
2641 [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
2642 permanent tables. Instead, the bit maps for such a class were previously
2643 created as the appropriate unions of the upper, lower, and digit bitmaps.
2644 Now they are created by subtraction from the [[:word:]] class, which has
2645 its own bitmap.
2646
2647 6. The [[:blank:]] character class matches horizontal, but not vertical space.
2648 It is created by subtracting the vertical space characters (\x09, \x0a,
2649 \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
2650 subtraction was done in the overall bitmap for a character class, meaning
2651 that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
2652 be recognized. This bug has been fixed.
2653
2654 7. Patches from the folks at Google:
2655
2656 (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
2657 real life, but is still worth protecting against".
2658
2659 (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
2660 regular expressions".
2661
2662 (c) pcre_scanner.cc: avoid use of std::count() because not all systems
2663 have it.
2664
2665 (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
2666 "configure" and the latter not, in order to fix a problem somebody had
2667 with compiling the Arg class on HP-UX.
2668
2669 (e) Improve the error-handling of the C++ wrapper a little bit.
2670
2671 (f) New tests for checking recursion limiting.
2672
2673 8. The pcre_memmove() function, which is used only if the environment does not
2674 have a standard memmove() function (and is therefore rarely compiled),
2675 contained two bugs: (a) use of int instead of size_t, and (b) it was not
2676 returning a result (though PCRE never actually uses the result).
2677
2678 9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
2679 large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
2680 returned instead of calling malloc() with an overflowing number that would
2681 most likely cause subsequent chaos.
2682
2683 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
2684
2685 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
2686 with this option is matched, the nmatch and pmatch options of regexec() are
2687 ignored.
2688
2689 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
2690 provided in case anyone wants to the the POSIX interface with UTF-8
2691 strings.
2692
2693 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
2694 C++ linking (needed for some HP-UX environments).
2695
2696 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
2697 (unused parameter) and in the pcre_printint() function (omitted "default"
2698 switch label when the default is to do nothing).
2699
2700 15. Added some code to make it possible, when PCRE is compiled as a C++
2701 library, to replace subject pointers for pcre_exec() with a smart pointer
2702 class, thus making it possible to process discontinuous strings.
2703
2704 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
2705 much the same function. They were added by different people who were trying
2706 to make PCRE easy to compile on non-Unix systems. It has been suggested
2707 that PCRE_EXPORT be abolished now that there is more automatic apparatus
2708 for compiling on Windows systems. I have therefore replaced it with
2709 PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
2710 defaults to "extern" for C or "extern C" for C++, which works fine on
2711 Unix-like systems. It is now possible to override the value of PCRE_DATA_
2712 SCOPE with something explicit in config.h. In addition:
2713
2714 (a) pcreposix.h still had just "extern" instead of either of these macros;
2715 I have replaced it with PCRE_DATA_SCOPE.
2716
2717 (b) Functions such as _pcre_xclass(), which are internal to the library,
2718 but external in the C sense, all had PCRE_EXPORT in their definitions.
2719 This is apparently wrong for the Windows case, so I have removed it.
2720 (It makes no difference on Unix-like systems.)
2721
2722 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
2723 of recursive calls to match(). This is different to MATCH_LIMIT because
2724 that limits the total number of calls to match(), not all of which increase
2725 the depth of recursion. Limiting the recursion depth limits the amount of
2726 stack (or heap if NO_RECURSE is set) that is used. The default can be set
2727 when PCRE is compiled, and changed at run time. A patch from Google adds
2728 this functionality to the C++ interface.
2729
2730 18. Changes to the handling of Unicode character properties:
2731
2732 (a) Updated the table to Unicode 4.1.0.
2733
2734 (b) Recognize characters that are not in the table as "Cn" (undefined).
2735
2736 (c) I revised the way the table is implemented to a much improved format
2737 which includes recognition of ranges. It now supports the ranges that
2738 are defined in UnicodeData.txt, and it also amalgamates other
2739 characters into ranges. This has reduced the number of entries in the
2740 table from around 16,000 to around 3,000, thus reducing its size
2741 considerably. I realized I did not need to use a tree structure after
2742 all - a binary chop search is just as efficient. Having reduced the
2743 number of entries, I extended their size from 6 bytes to 8 bytes to
2744 allow for more data.
2745
2746 (d) Added support for Unicode script names via properties such as \p{Han}.
2747
2748 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
2749 matching that character.
2750
2751 20. When matching a repeated Unicode property with a minimum greater than zero,
2752 (for example \pL{2,}), PCRE could look past the end of the subject if it
2753 reached it while seeking the minimum number of characters. This could
2754 happen only if some of the characters were more than one byte long, because
2755 there is a check for at least the minimum number of bytes.
2756
2757 21. Refactored the implementation of \p and \P so as to be more general, to
2758 allow for more different types of property in future. This has changed the
2759 compiled form incompatibly. Anybody with saved compiled patterns that use
2760 \p or \P will have to recompile them.
2761
2762 22. Added "Any" and "L&" to the supported property types.
2763
2764 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
2765 but give a compile time error if the value is greater than 0xff.
2766
2767 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
2768 accidentally not being installed or uninstalled.
2769
2770 25. The pcre.h file was built from pcre.h.in, but the only changes that were
2771 made were to insert the current release number. This seemed silly, because
2772 it made things harder for people building PCRE on systems that don't run
2773 "configure". I have turned pcre.h into a distributed file, no longer built
2774 by "configure", with the version identification directly included. There is
2775 no longer a pcre.h.in file.
2776
2777 However, this change necessitated a change to the pcre-config script as
2778 well. It is built from pcre-config.in, and one of the substitutions was the
2779 release number. I have updated configure.ac so that ./configure now finds
2780 the release number by grepping pcre.h.
2781
2782 26. Added the ability to run the tests under valgrind.
2783
2784
2785 Version 6.4 05-Sep-05
2786 ---------------------
2787
2788 1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
2789 "--" to be printed when multiple files were scanned, even when none of the
2790 -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
2791 consider it to be a bug, and have restored the previous behaviour.
2792
2793 2. A couple of code tidies to get rid of compiler warnings.
2794
2795 3. The pcretest program used to cheat by referring to symbols in the library
2796 whose names begin with _pcre_. These are internal symbols that are not
2797 really supposed to be visible externally, and in some environments it is
2798 possible to suppress them. The cheating is now confined to including
2799 certain files from the library's source, which is a bit cleaner.
2800
2801 4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
2802 file's purpose clearer.
2803
2804 5. Reorganized pcre_ucp_findchar().
2805
2806
2807 Version 6.3 15-Aug-05
2808 ---------------------
2809
2810 1. The file libpcre.pc.in did not have general read permission in the tarball.
2811
2812 2. There were some problems when building without C++ support:
2813
2814 (a) If C++ support was not built, "make install" and "make test" still
2815 tried to test it.
2816
2817 (b) There were problems when the value of CXX was explicitly set. Some
2818 changes have been made to try to fix these, and ...
2819
2820 (c) --disable-cpp can now be used to explicitly disable C++ support.
2821
2822 (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
2823 backslash in a target when C++ was disabled. This confuses some
2824 versions of "make", apparently. Using an intermediate variable solves
2825 this. (Same for CPP_LOBJ.)
2826
2827 3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
2828 (non-Windows) now includes $(CFLAGS) because these flags are sometimes
2829 necessary on certain architectures.
2830
2831 4. Added a setting of -export-symbols-regex to the link command to remove
2832 those symbols that are exported in the C sense, but actually are local
2833 within the library, and not documented. Their names all begin with
2834 "_pcre_". This is not a perfect job, because (a) we have to except some
2835 symbols that pcretest ("illegally") uses, and (b) the facility isn't always
2836 available (and never for static libraries). I have made a note to try to
2837 find a way round (a) in the future.
2838
2839
2840 Version 6.2 01-Aug-05
2841 ---------------------
2842
2843 1. There was no test for integer overflow of quantifier values. A construction
2844 such as {1111111111111111} would give undefined results. What is worse, if
2845 a minimum quantifier for a parenthesized subpattern overflowed and became
2846 negative, the calculation of the memory size went wrong. This could have
2847 led to memory overwriting.
2848
2849 2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
2850
2851 3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
2852 operating environments where this matters.
2853
2854 4. Applied Giuseppe Maxia's patch to add additional features for controlling
2855 PCRE options from within the C++ wrapper.
2856
2857 5. Named capturing subpatterns were not being correctly counted when a pattern
2858 was compiled. This caused two problems: (a) If there were more than 100
2859 such subpatterns, the calculation of the memory needed for the whole
2860 compiled pattern went wrong, leading to an overflow error. (b) Numerical
2861 back references of the form \12, where the number was greater than 9, were
2862 not recognized as back references, even though there were sufficient
2863 previous subpatterns.
2864
2865 6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
2866 versions of gcc, e.g. 2.95.4.
2867
2868
2869 Version 6.1 21-Jun-05
2870 ---------------------
2871
2872 1. There was one reference to the variable "posix" in pcretest.c that was not
2873 surrounded by "#if !defined NOPOSIX".
2874
2875 2. Make it possible to compile pcretest without DFA support, UTF8 support, or
2876 the cross-check on the old pcre_info() function, for the benefit of the
2877 cut-down version of PCRE that is currently imported into Exim.
2878
2879 3. A (silly) pattern starting with (?i)(?-i) caused an internal space
2880 allocation error. I've done the easy fix, which wastes 2 bytes for sensible
2881 patterns that start (?i) but I don't think that matters. The use of (?i) is
2882 just an example; this all applies to the other options as well.
2883
2884 4. Since libtool seems to echo the compile commands it is issuing, the output
2885 from "make" can be reduced a bit by putting "@" in front of each libtool
2886 compile command.
2887
2888 5. Patch from the folks at Google for configure.in to be a bit more thorough
2889 in checking for a suitable C++ installation before trying to compile the
2890 C++ stuff. This should fix a reported problem when a compiler was present,
2891 but no suitable headers.
2892
2893 6. The man pages all had just "PCRE" as their title. I have changed them to
2894 be the relevant file name. I have also arranged that these names are
2895 retained in the file doc/pcre.txt, which is a concatenation in text format
2896 of all the man pages except the little individual ones for each function.
2897
2898 7. The NON-UNIX-USE file had not been updated for the different set of source
2899 files that come with release 6. I also added a few comments about the C++
2900 wrapper.
2901
2902
2903 Version 6.0 07-Jun-05
2904 ---------------------
2905
2906 1. Some minor internal re-organization to help with my DFA experiments.
2907
2908 2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
2909 didn't matter for the library itself when fully configured, but did matter
2910 when compiling without UCP support, or within Exim, where the ucp files are
2911 not imported.
2912
2913 3. Refactoring of the library code to split up the various functions into
2914 different source modules. The addition of the new DFA matching code (see
2915 below) to a single monolithic source would have made it really too
2916 unwieldy, quite apart from causing all the code to be include in a
2917 statically linked application, when only some functions are used. This is
2918 relevant even without the DFA addition now that patterns can be compiled in
2919 one application and matched in another.
2920
2921 The downside of splitting up is that there have to be some external
2922 functions and data tables that are used internally in different modules of
2923 the library but which are not part of the API. These have all had their
2924 names changed to start with "_pcre_" so that they are unlikely to clash
2925 with other external names.
2926
2927 4. Added an alternate matching function, pcre_dfa_exec(), which matches using
2928 a different (DFA) algorithm. Although it is slower than the original
2929 function, it does have some advantages for certain types of matching
2930 problem.
2931
2932 5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
2933 including restarting after a partial match.
2934
2935 6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
2936 defined when compiling for Windows was sent to me. I have put it into the
2937 code, though I have no means of testing or verifying it.
2938
2939 7. Added the pcre_refcount() auxiliary function.
2940
2941 8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
2942 match before or at the first newline in the subject string. In pcretest,
2943 the /f option on a pattern can be used to set this.
2944
2945 9. A repeated \w when used in UTF-8 mode with characters greater than 256
2946 would behave wrongly. This has been present in PCRE since release 4.0.
2947
2948 10. A number of changes to the pcregrep command:
2949
2950 (a) Refactored how -x works; insert ^(...)$ instead of setting
2951 PCRE_ANCHORED and checking the length, in preparation for adding
2952 something similar for -w.
2953
2954 (b) Added the -w (match as a word) option.
2955
2956 (c) Refactored the way lines are read and buffered so as to have more
2957 than one at a time available.
2958
2959 (d) Implemented a pcregrep test script.
2960
2961 (e) Added the -M (multiline match) option. This allows patterns to match
2962 over several lines of the subject. The buffering ensures that at least
2963 8K, or the rest of the document (whichever is the shorter) is available
2964 for matching (and similarly the previous 8K for lookbehind assertions).
2965
2966 (f) Changed the --help output so that it now says
2967
2968 -w, --word-regex(p)
2969
2970 instead of two lines, one with "regex" and the other with "regexp"
2971 because that confused at least one person since the short forms are the
2972 same. (This required a bit of code, as the output is generated
2973 automatically from a table. It wasn't just a text change.)
2974
2975 (g) -- can be used to terminate pcregrep options if the next thing isn't an
2976 option but starts with a hyphen. Could be a pattern or a path name
2977 starting with a hyphen, for instance.
2978
2979 (h) "-" can be given as a file name to represent stdin.
2980
2981 (i) When file names are being printed, "(standard input)" is used for
2982 the standard input, for compatibility with GNU grep. Previously
2983 "<stdin>" was used.
2984
2985 (j) The option --label=xxx can be used to supply a name to be used for
2986 stdin when file names are being printed. There is no short form.
2987
2988 (k) Re-factored the options decoding logic because we are going to add
2989 two more options that take data. Such options can now be given in four
2990 different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
2991
2992 (l) Added the -A, -B, and -C options for requesting that lines of context
2993 around matches be printed.
2994
2995 (m) Added the -L option to print the names of files that do not contain
2996 any matching lines, that is, the complement of -l.
2997
2998 (n) The return code is 2 if any file cannot be opened, but pcregrep does
2999 continue to scan other files.
3000
3001 (o) The -s option was incorrectly implemented. For compatibility with other
3002 greps, it now suppresses the error message for a non-existent or non-
3003 accessible file (but not the return code). There is a new option called
3004 -q that suppresses the output of matching lines, which was what -s was
3005 previously doing.
3006
3007 (p) Added --include and --exclude options to specify files for inclusion
3008 and exclusion when recursing.
3009
3010 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
3011 Hopefully, it now does.
3012
3013 12. Missing cast in pcre_study().
3014
3015 13. Added an "uninstall" target to the makefile.
3016
3017 14. Replaced "extern" in the function prototypes in Makefile.in with
3018 "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
3019 world, but is set differently for Windows.
3020
3021 15. Added a second compiling function called pcre_compile2(). The only
3022 difference is that it has an extra argument, which is a pointer to an
3023 integer error code. When there is a compile-time failure, this is set
3024 non-zero, in addition to the error test pointer being set to point to an
3025 error message. The new argument may be NULL if no error number is required
3026 (but then you may as well call pcre_compile(), which is now just a
3027 wrapper). This facility is provided because some applications need a
3028 numeric error indication, but it has also enabled me to tidy up the way
3029 compile-time errors are handled in the POSIX wrapper.
3030
3031 16. Added VPATH=.libs to the makefile; this should help when building with one
3032 prefix path and installing with another. (Or so I'm told by someone who
3033 knows more about this stuff than I do.)
3034
3035 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
3036 passes PCRE_DOTALL to the pcre_compile() function, making the "." character
3037 match everything, including newlines. This is not POSIX-compatible, but
3038 somebody wanted the feature. From pcretest it can be activated by using
3039 both the P and the s flags.
3040
3041 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
3042
3043 19. libpcre.pc was being incorrectly installed as executable.
3044
3045 20. A couple of places in pcretest check for end-of-line by looking for '\n';
3046 it now also looks for '\r' so that it will work unmodified on Windows.
3047
3048 21. Added Google's contributed C++ wrapper to the distribution.
3049
3050 22. Added some untidy missing memory free() calls in pcretest, to keep
3051 Electric Fence happy when testing.
3052
3053
3054
3055 Version 5.0 13-Sep-04
3056 ---------------------
3057
3058 1. Internal change: literal characters are no longer packed up into items
3059 containing multiple characters in a single byte-string. Each character
3060 is now matched using a separate opcode. However, there may be more than one
3061 byte in the character in UTF-8 mode.
3062
3063 2. The pcre_callout_block structure has two new fields: pattern_position and
3064 next_item_length. These contain the offset in the pattern to the next match
3065 item, and its length, respectively.
3066
3067 3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
3068 insertion of callouts before each pattern item. Added the /C option to
3069 pcretest to make use of this.
3070
3071 4. On the advice of a Windows user, the lines
3072
3073 #if defined(_WIN32) || defined(WIN32)
3074 _setmode( _fileno( stdout ), 0x8000 );
3075 #endif /* defined(_WIN32) || defined(WIN32) */
3076
3077 have been added to the source of pcretest. This apparently does useful
3078 magic in relation to line terminators.
3079
3080 5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
3081 for the benefit of those environments where the "b" makes a difference.
3082
3083 6. The icc compiler has the same options as gcc, but "configure" doesn't seem
3084 to know about it. I have put a hack into configure.in that adds in code
3085 to set GCC=yes if CC=icc. This seems to end up at a point in the
3086 generated configure script that is early enough to affect the setting of
3087 compiler options, which is what is needed, but I have no means of testing
3088 whether it really works. (The user who reported this had patched the
3089 generated configure script, which of course I cannot do.)
3090
3091 LATER: After change 22 below (new libtool files), the configure script
3092 seems to know about icc (and also ecc). Therefore, I have commented out
3093 this hack in configure.in.
3094
3095 7. Added support for pkg-config (2 patches were sent in).
3096
3097 8. Negated POSIX character classes that used a combination of internal tables
3098 were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
3099 [[:^ascii]]. Typically, they would match almost any characters. The other
3100 POSIX classes were not broken in this way.
3101
3102 9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
3103 to find the match, as PCRE was deluded into thinking that the match had to
3104 start at the start point or following a newline. The same bug applied to
3105 patterns with negative forward assertions or any backward assertions
3106 preceding ".*" at the start, unless the pattern required a fixed first
3107 character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
3108
3109 10. In UTF-8 mode, when moving forwards in the subject after a failed match
3110 starting at the last subject character, bytes beyond the end of the subject
3111 string were read.
3112
3113 11. Renamed the variable "class" as "classbits" to make life easier for C++
3114 users. (Previously there was a macro definition, but it apparently wasn't
3115 enough.)
3116
3117 12. Added the new field "tables" to the extra data so that tables can be passed
3118 in at exec time, or the internal tables can be re-selected. This allows
3119 a compiled regex to be saved and re-used at a later time by a different
3120 program that might have everything at different addresses.
3121
3122 13. Modified the pcre-config script so that, when run on Solaris, it shows a
3123 -R library as well as a -L library.
3124
3125 14. The debugging options of pcretest (-d on the command line or D on a
3126 pattern) showed incorrect output for anything following an extended class
3127 that contained multibyte characters and which was followed by a quantifier.
3128
3129 15. Added optional support for general category Unicode character properties
3130 via the \p, \P, and \X escapes. Unicode property support implies UTF-8
3131 support. It adds about 90K to the size of the library. The meanings of the
3132 inbuilt class escapes such as \d and \s have NOT been changed.
3133
3134 16. Updated pcredemo.c to include calls to free() to release the memory for the
3135 compiled pattern.
3136
3137 17. The generated file chartables.c was being created in the source directory
3138 instead of in the building directory. This caused the build to fail if the
3139 source directory was different from the building directory, and was
3140 read-only.
3141
3142 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
3143 file. No doubt somebody will tell me if they don't make sense... Also added
3144 Dan Mooney's comments about building on OpenVMS.
3145
3146 19. Added support for partial matching via the PCRE_PARTIAL option for
3147 pcre_exec() and the \P data escape in pcretest.
3148
3149 20. Extended pcretest with 3 new pattern features:
3150
3151 (i) A pattern option of the form ">rest-of-line" causes pcretest to
3152 write the compiled pattern to the file whose name is "rest-of-line".
3153 This is a straight binary dump of the data, with the saved pointer to
3154 the character tables forced to be NULL. The study data, if any, is
3155 written too. After writing, pcretest reads a new pattern.
3156
3157 (ii) If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
3158 compiled pattern from the given file. There must not be any
3159 occurrences of "<" in the file name (pretty unlikely); if there are,
3160 pcretest will instead treat the initial "<" as a pattern delimiter.
3161 After reading in the pattern, pcretest goes on to read data lines as
3162 usual.
3163
3164 (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
3165 and 16-bit fields in a compiled pattern, to simulate a pattern that
3166 was compiled on a host of opposite endianness.
3167
3168 21. The pcre-exec() function can now cope with patterns that were compiled on
3169 hosts of opposite endianness, with this restriction:
3170
3171 As for any compiled expression that is saved and used later, the tables
3172 pointer field cannot be preserved; the extra_data field in the arguments
3173 to pcre_exec() should be used to pass in a tables address if a value
3174 other than the default internal tables were used at compile time.
3175
3176 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
3177 now diagnosed as an error. Previously, most of the time, a negative number
3178 would have been treated as zero, but if in addition "ovector" was passed as
3179 NULL, a crash could occur.
3180
3181 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
3182 new versions from the libtool 1.5 distribution (the last one is a copy of
3183 a file called libtool.m4). This seems to have fixed the need to patch
3184 "configure" to support Darwin 1.3 (which I used to do). However, I still
3185 had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
3186 workstation).
3187
3188 24. Changed the PCRE licence to be the more standard "BSD" licence.
3189
3190
3191 Version 4.5 01-Dec-03
3192 ---------------------
3193
3194 1. There has been some re-arrangement of the code for the match() function so
3195 that it can be compiled in a version that does not call itself recursively.
3196 Instead, it keeps those local variables that need separate instances for
3197 each "recursion" in a frame on the heap, and gets/frees frames whenever it
3198 needs to "recurse". Keeping track of where control must go is done by means
3199 of setjmp/longjmp. The whole thing is implemented by a set of macros that
3200 hide most of the details from the main code, and operates only if
3201 NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
3202 "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
3203 operating.
3204
3205 To make it easier for callers to provide specially tailored get/free
3206 functions for this usage, two new functions, pcre_stack_malloc, and
3207 pcre_stack_free, are used. They are always called in strict stacking order,
3208 and the size of block requested is always the same.
3209
3210 The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
3211 PCRE has been compiled to use the stack or the heap for recursion. The
3212 -C option of pcretest uses this to show which version is compiled.
3213
3214 A new data escape \S, is added to pcretest; it causes the amounts of store
3215 obtained and freed by both kinds of malloc/free at match time to be added
3216 to the output.
3217
3218 2. Changed the locale test to use "fr_FR" instead of "fr" because that's
3219 what's available on my current Linux desktop machine.
3220
3221 3. When matching a UTF-8 string, the test for a valid string at the start has
3222 been extended. If start_offset is not zero, PCRE now checks that it points
3223 to a byte that is the start of a UTF-8 character. If not, it returns
3224 PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
3225 this is necessary because there may be backward assertions in the pattern.
3226 When matching the same subject several times, it may save resources to use
3227 PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
3228
3229 4. The code for checking the validity of UTF-8 strings has been tightened so
3230 that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
3231 containing "overlong sequences".
3232
3233 5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
3234 I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
3235 should have been "&", but it just so happened that all the cases this let
3236 through by mistake were picked up later in the function.
3237
3238 6. I had used a variable called "isblank" - this is a C99 function, causing
3239 some compilers to warn. To avoid this, I renamed it (as "blankclass").
3240
3241 7. Cosmetic: (a) only output another newline at the end of pcretest if it is
3242 prompting; (b) run "./pcretest /dev/null" at the start of the test script
3243 so the version is shown; (c) stop "make test" echoing "./RunTest".
3244
3245 8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
3246
3247 9. The prototype for memmove() for systems that don't have it was using
3248 size_t, but the inclusion of the header that defines size_t was later. I've
3249 moved the #includes for the C headers earlier to avoid this.
3250
3251 10. Added some adjustments to the code to make it easier to compiler on certain
3252 special systems:
3253
3254 (a) Some "const" qualifiers were missing.
3255 (b) Added the macro EXPORT before all exported functions; by default this
3256 is defined to be empty.
3257 (c) Changed the dftables auxiliary program (that builds chartables.c) so
3258 that it reads its output file name as an argument instead of writing
3259 to the standard output and assuming this can be redirected.
3260
3261 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
3262 class containing characters with values greater than 255, PCRE compilation
3263 went into a loop.
3264
3265 12. A recursive reference to a subpattern that was within another subpattern
3266 that had a minimum quantifier of zero caused PCRE to crash. For example,
3267 (x(y(?2))z)? provoked this bug with a subject that got as far as the
3268 recursion. If the recursively-called subpattern itself had a zero repeat,
3269 that was OK.
3270
3271 13. In pcretest, the buffer for reading a data line was set at 30K, but the
3272 buffer into which it was copied (for escape processing) was still set at
3273 1024, so long lines caused crashes.
3274
3275 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
3276 "internal error: code overflow...". This applied to any character class
3277 that was followed by a possessive quantifier.
3278
3279 15. Modified the Makefile to add libpcre.la as a prerequisite for
3280 libpcreposix.la because I was told this is needed for a parallel build to
3281 work.
3282
3283 16. If a pattern that contained .* following optional items at the start was
3284 studied, the wrong optimizing data was generated, leading to matching
3285 errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
3286 matching string must start with a or b or c. The correct conclusion for
3287 this pattern is that a match can start with any character.
3288
3289
3290 Version 4.4 13-Aug-03
3291 ---------------------
3292
3293 1. In UTF-8 mode, a character class containing characters with values between
3294 127 and 255 was not handled correctly if the compiled pattern was studied.
3295 In fixing this, I have also improved the studying algorithm for such
3296 classes (slightly).
3297
3298 2. Three internal functions had redundant arguments passed to them. Removal
3299 might give a very teeny performance improvement.
3300
3301 3. Documentation bug: the value of the capture_top field in a callout is *one
3302 more than* the number of the hightest numbered captured substring.
3303
3304 4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
3305 in incorrectly linking with a previously installed version. They now link
3306 explicitly with libpcre.la.
3307
3308 5. configure.in no longer needs to recognize Cygwin specially.
3309
3310 6. A problem in pcre.in for Windows platforms is fixed.
3311
3312 7. If a pattern was successfully studied, and the -d (or /D) flag was given to
3313 pcretest, it used to include the size of the study block as part of its
3314 output. Unfortunately, the structure contains a field that has a different
3315 size on different hardware architectures. This meant that the tests that
3316 showed this size failed. As the block is currently always of a fixed size,
3317 this information isn't actually particularly useful in pcretest output, so
3318 I have just removed it.
3319
3320 8. Three pre-processor statements accidentally did not start in column 1.
3321 Sadly, there are *still* compilers around that complain, even though
3322 standard C has not required this for well over a decade. Sigh.
3323
3324 9. In pcretest, the code for checking callouts passed small integers in the
3325 callout_data field, which is a void * field. However, some picky compilers
3326 complained about the casts involved for this on 64-bit systems. Now
3327 pcretest passes the address of the small integer instead, which should get
3328 rid of the warnings.
3329
3330 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
3331 both compile and run time, and gives an error if an invalid UTF-8 sequence
3332 is found. There is a option for disabling this check in cases where the
3333 string is known to be correct and/or the maximum performance is wanted.
3334
3335 11. In response to a bug report, I changed one line in Makefile.in from
3336
3337 -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
3338 to
3339 -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
3340
3341 to look similar to other lines, but I have no way of telling whether this
3342 is the right thing to do, as I do not use Windows. No doubt I'll get told
3343 if it's wrong...
3344
3345
3346 Version 4.3 21-May-03
3347 ---------------------
3348
3349 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
3350 Makefile.
3351
3352 2. Some refactoring to improve the quality of the code:
3353
3354 (i) The utf8_table... variables are now declared "const".
3355
3356 (ii) The code for \cx, which used the "case flipping" table to upper case
3357 lower case letters, now just substracts 32. This is ASCII-specific,
3358 but the whole concept of \cx is ASCII-specific, so it seems
3359 reasonable.
3360
3361 (iii) PCRE was using its character types table to recognize decimal and
3362 hexadecimal digits in the pattern. This is silly, because it handles
3363 only 0-9, a-f, and A-F, but the character types table is locale-
3364 specific, which means strange things might happen. A private
3365 table is now used for this - though it costs 256 bytes, a table is
3366 much faster than multiple explicit tests. Of course, the standard
3367 character types table is still used for matching digits in subject
3368 strings against \d.
3369
3370 (iv) Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
3371 ending in _t are). So I've renamed it as ESC_tee.
3372
3373 3. The first argument for regexec() in the POSIX wrapper should have been
3374 defined as "const".
3375
3376 4. Changed pcretest to use malloc() for its buffers so that they can be
3377 Electric Fenced for debugging.
3378
3379 5. There were several places in the code where, in UTF-8 mode, PCRE would try
3380 to read one or more bytes before the start of the subject string. Often this
3381 had no effect on PCRE's behaviour, but in some circumstances it could
3382 provoke a segmentation fault.
3383
3384 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
3385 to try to read one or more bytes before the start of the subject string.
3386
3387 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
3388 UTF-8 support could misbehave in various ways if the subject string
3389 contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
3390 area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
3391 back over UTF-8 characters.)
3392
3393
3394 Version 4.2 14-Apr-03
3395 ---------------------
3396
3397 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
3398
3399 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
3400 [ON_WINDOWS]: new variable, "#" on non-Windows platforms
3401 [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
3402 [WIN_PREFIX]: new variable, "cyg" for Cygwin
3403 * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
3404 and BUILD_EXEEXT
3405 Note: automatic setting of the BUILD variables is not yet working
3406 set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
3407 compile-time but not at link-time
3408 [LINK]: use for linking executables only
3409 make different versions for Windows and non-Windows
3410 [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
3411 libraries
3412 [LINK_FOR_BUILD]: new variable
3413 [OBJEXT]: use throughout
3414 [EXEEXT]: use throughout
3415 <winshared>: new target
3416 <wininstall>: new target
3417 <dftables.o>: use native compiler
3418 <dftables>: use native linker
3419 <install>: handle Windows platform correctly
3420 <clean>: ditto
3421 <check>: ditto
3422 copy DLL to top builddir before testing
3423
3424 As part of these changes, -no-undefined was removed again. This was reported
3425 to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
3426 in any case.
3427
3428 3. Some tidies to get rid of compiler warnings:
3429
3430 . In the match_data structure, match_limit was an unsigned long int, whereas
3431 match_call_count was an int. I've made them both unsigned long ints.
3432
3433 . In pcretest the fact that a const uschar * doesn't automatically cast to
3434 a void * provoked a warning.
3435
3436 . Turning on some more compiler warnings threw up some "shadow" variables
3437 and a few more missing casts.
3438
3439 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
3440 option, a class that contained a single character with a value between 128
3441 and 255 (e.g. /[\xFF]/) caused PCRE to crash.
3442
3443 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
3444 option, a class that contained several characters, but with at least one
3445 whose value was between 128 and 255 caused PCRE to crash.
3446
3447
3448 Version 4.1 12-Mar-03
3449 ---------------------
3450
3451 1. Compiling with gcc -pedantic found a couple of places where casts were
3452 needed, and a string in dftables.c that was longer than standard compilers are
3453 required to support.
3454
3455 2. Compiling with Sun's compiler found a few more places where the code could
3456 be tidied up in order to avoid warnings.
3457
3458 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
3459 first of these names is deprecated in the latest Autoconf in favour of the name
3460 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
3461 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
3462 analogy I have changed it to CFLAGS_FOR_BUILD.
3463
3464 4. Added -no-undefined to the linking command in the Makefile, because this is
3465 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
3466 linking step for the pcreposix library.
3467
3468 5. PCRE was failing to diagnose the case of two named groups with the same
3469 name.
3470
3471 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
3472 literal character that is needed in the subject for a match, and scans along to
3473 ensure that it is present before embarking on the full matching process. This
3474 saves time in cases of nested unlimited repeats that are never going to match.
3475 Problem: the scan can take a lot of time if the subject is very long (e.g.
3476 megabytes), thus penalizing straightforward matches. It is now done only if the
3477 amount of subject to be scanned is less than 1000 bytes.
3478
3479 7. A lesser problem with the same optimization is that it was recording the
3480 first character of an anchored pattern as "needed", thus provoking a search
3481 right along the subject, even when the first match of the pattern was going to
3482 fail. The "needed" character is now not set for anchored patterns, unless it
3483 follows something in the pattern that is of non-fixed length. Thus, it still
3484 fulfils its original purpose of finding quick non-matches in cases of nested
3485 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
3486
3487
3488 Version 4.0 17-Feb-03
3489 ---------------------
3490
3491 1. If a comment in an extended regex that started immediately after a meta-item
3492 extended to the end of string, PCRE compiled incorrect data. This could lead to
3493 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
3494
3495 2. Moved to autoconf 2.53 and libtool 1.4.2.
3496
3497 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
3498 the special perltest8 script is no longer needed - all the tests can be run
3499 from a single perltest script.
3500
3501 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
3502 by \s. It has now been removed in PCRE. This means it isn't recognized as
3503 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
3504 class [:space:] *does* include VT, thereby creating a mess.
3505
3506 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
3507 space and tab.
3508
3509 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
3510 its new features into the main test script, reducing the number of scripts.
3511
3512 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
3513 were backward compatible, and made the (?i) apply to the whole pattern, as if
3514 /i were given. Now it behaves more logically, and applies the option setting
3515 only to what follows. PCRE has been changed to follow suit. However, if it
3516 finds options settings right at the start of the pattern, it extracts them into
3517 the global options, as before. Thus, they show up in the info data.
3518
3519 8. Added support for the \Q...\E escape sequence. Characters in between are
3520 treated as literals. This is slightly different from Perl in that $ and @ are
3521 also handled as literals inside the quotes. In Perl, they will cause variable
3522 interpolation. Note the following examples:
3523
3524 Pattern PCRE matches Perl matches
3525
3526 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
3527 \Qabc\$xyz\E abc\$xyz abc\$xyz
3528 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
3529
3530 For compatibility with Perl, \Q...\E sequences are recognized inside character
3531 classes as well as outside them.
3532
3533 9. Re-organized 3 code statements in pcretest to avoid "overflow in
3534 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
3535 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
3536 signed/unsigned warnings.
3537
3538 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
3539 option for pcretest, so I've replaced it by a simple function that does just
3540 that job.
3541
3542 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
3543 "pcregrep -".
3544
3545 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
3546 Java package. This provides some syntactic sugar for simple cases of what my
3547 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
3548 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
3549 item, you can use this simplified notation. Note that only makes sense with
3550 greedy quantifiers. Consequently, the use of the possessive quantifier forces
3551 greediness, whatever the setting of the PCRE_UNGREEDY option.
3552
3553 13. A change of greediness default within a pattern was not taking effect at
3554 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
3555 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
3556 was abstracted outside.
3557
3558 14. PCRE now supports the \G assertion. It is true when the current matching
3559 position is at the start point of the match. This differs from \A when the
3560 starting offset is non-zero. Used with the /g option of pcretest (or similar
3561 code), it works in the same way as it does for Perl's /g option. If all
3562 alternatives of a regex begin with \G, the expression is anchored to the start
3563 match position, and the "anchored" flag is set in the compiled expression.
3564
3565 15. Some bugs concerning the handling of certain option changes within patterns
3566 have been fixed. These applied to options other than (?ims). For example,
3567 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
3568 been the other way round. Some of this was related to change 7 above.
3569
3570 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
3571 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
3572 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
3573 POSIX classes only within a class (e.g. /[[:alpha:]]/).
3574
3575 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
3576 mode. Unlike ".", it always matches newline, whatever the setting of
3577 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
3578 assertions. Perl allows it, but it doesn't (in general) work because it can't
3579 calculate the length of the lookbehind. At least, that's the case for Perl
3580 5.8.0 - I've been told they are going to document that it doesn't work in
3581 future.
3582
3583 18. Added an error diagnosis for escapes that PCRE does not support: these are
3584 \L, \l, \N, \P, \p, \U, \u, and \X.
3585
3586 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
3587 reading past the end of the pattern in cases such as /[abcd/.
3588
3589 20. PCRE was getting more memory than necessary for patterns with classes that
3590 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
3591
3592 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
3593 compiling PCRE for use with Virtual Pascal.
3594
3595 22. Small fix to the Makefile to make it work properly if the build is done
3596 outside the source tree.
3597
3598 23. Added a new extension: a condition to go with recursion. If a conditional
3599 subpattern starts with (?(R) the "true" branch is used if recursion has
3600 happened, whereas the "false" branch is used only at the top level.
3601
3602 24. When there was a very long string of literal characters (over 255 bytes
3603 without UTF support, over 250 bytes with UTF support), the computation of how
3604 much memory was required could be incorrect, leading to segfaults or other
3605 strange effects.
3606
3607 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
3608 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
3609 there was a subsequent back reference to those brackets. This meant that, for
3610 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
3611 possible to check for precisely this case. All we can do is abandon the
3612 optimization if .* occurs inside capturing brackets when there are any back
3613 references whatsoever. (See below for a better fix that came later.)
3614
3615 26. The handling of the optimization for finding the first character of a
3616 non-anchored pattern, and for finding a character that is required later in the
3617 match were failing in some cases. This didn't break the matching; it just
3618 failed to optimize when it could. The way this is done has been re-implemented.
3619
3620 27. Fixed typo in error message for invalid (?R item (it said "(?p").
3621
3622 28. Added a new feature that provides some of the functionality that Perl
3623 provides with (?{...}). The facility is termed a "callout". The way it is done
3624 in PCRE is for the caller to provide an optional function, by setting
3625 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
3626 global variable. By default it is unset, which disables all calling out. To get
3627 the function called, the regex must include (?C) at appropriate points. This
3628 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
3629 This provides a means of identifying different callout points. When PCRE
3630 reaches such a point in the regex, if pcre_callout has been set, the external
3631 function is called. It is provided with data in a structure called
3632 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
3633 matching continues; if it returns a non-zero value, the match at the current
3634 point fails. However, backtracking will occur if possible. [This was changed
3635 later and other features added - see item 49 below.]
3636
3637 29. pcretest is upgraded to test the callout functionality. It provides a
3638 callout function that displays information. By default, it shows the start of
3639 the match and the current position in the text. There are some new data escapes
3640 to vary what happens:
3641
3642 \C+ in addition, show current contents of captured substrings
3643 \C- do not supply a callout function
3644 \C!n return 1 when callout number n is reached
3645 \C!n!m return 1 when callout number n is reached for the mth time
3646
3647 30. If pcregrep was called with the -l option and just a single file name, it
3648 output "<stdin>" if a match was found, instead of the file name.
3649
3650 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
3651 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
3652 pcre_exec(). This saves a malloc/free per call. The default value of
3653 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
3654 when configuring.
3655
3656 32. The default maximum size of a compiled pattern is 64K. There have been a
3657 few cases of people hitting this limit. The code now uses macros to handle the
3658 storing of links as offsets within the compiled pattern. It defaults to 2-byte
3659 links, but this can be changed to 3 or 4 bytes by --with-link-size when
3660 configuring. Tests 2 and 5 work only with 2-byte links because they output
3661 debugging information about compiled patterns.
3662
3663 33. Internal code re-arrangements:
3664
3665 (a) Moved the debugging function for printing out a compiled regex into
3666 its own source file (printint.c) and used #include to pull it into
3667 pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
3668 separate copies.
3669
3670 (b) Defined the list of op-code names for debugging as a macro in
3671 internal.h so that it is next to the definition of the opcodes.
3672
3673 (c) Defined a table of op-code lengths for simpler skipping along compiled
3674 code. This is again a macro in internal.h so that it is next to the
3675 definition of the opcodes.
3676
3677 34. Added support for recursive calls to individual subpatterns, along the
3678 lines of Robin Houston's patch (but implemented somewhat differently).
3679
3680 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
3681 allow it to read and process whole directories in Win32. This code was
3682 contributed by Lionel Fourquaux; it has not been tested by me.
3683
3684 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
3685 used to name a group. Names consist of alphanumerics and underscores, and must
3686 be unique. Back references use the syntax (?P=name) and recursive calls use
3687 (?P>name) which is a PCRE extension to the Python extension. Groups still have
3688 numbers. The function pcre_fullinfo() can be used after compilation to extract
3689 a name/number map. There are three relevant calls:
3690
3691 PCRE_INFO_NAMEENTRYSIZE yields the size of each entry in the map
3692 PCRE_INFO_NAMECOUNT yields the number of entries
3693 PCRE_INFO_NAMETABLE yields a pointer to the map.
3694
3695 The map is a vector of fixed-size entries. The size of each entry depends on
3696 the length of the longest name used. The first two bytes of each entry are the
3697 group number, most significant byte first. There follows the corresponding
3698 name, zero terminated. The names are in alphabetical order.
3699
3700 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
3701 case instead of 255. Making it the same both with and without UTF-8 support
3702 means that the same test output works with both.
3703
3704 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
3705 calling malloc() with a zero argument.
3706
3707 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
3708 optimization. I've improved things by keeping a bitmap of backreferences with
3709 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
3710 fact referenced, the optimization can be applied. It is unlikely that a
3711 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
3712 the match to follow \n) will appear inside brackets with a number greater than
3713 31, but if it does, any back reference > 31 suppresses the optimization.
3714
3715 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
3716 of disabling numbered capturing parentheses. Any opening parenthesis that is
3717 not followed by ? behaves as if it were followed by ?: but named parentheses
3718 can still be used for capturing (and they will acquire numbers in the usual
3719 way).
3720
3721 41. Redesigned the return codes from the match() function into yes/no/error so
3722 that errors can be passed back from deep inside the nested calls. A malloc
3723 failure while inside a recursive subpattern call now causes the
3724 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
3725
3726 42. It is now possible to set a limit on the number of times the match()
3727 function is called in a call to pcre_exec(). This facility makes it possible to
3728 limit the amount of recursion and backtracking, though not in a directly
3729 obvious way, because the match() function is used in a number of different
3730 circumstances. The count starts from zero for each position in the subject
3731 string (for non-anchored patterns). The default limit is, for compatibility, a
3732 large number, namely 10 000 000. You can change this in two ways:
3733
3734 (a) When configuring PCRE before making, you can use --with-match-limit=n
3735 to set a default value for the compiled library.
3736
3737 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
3738 a different value is set. See 45 below.
3739
3740 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
3741
3742 43. Added a new function pcre_config(int, void *) to enable run-time extraction
3743 of things that can be changed at compile time. The first argument specifies
3744 what is wanted and the second points to where the information is to be placed.
3745 The current list of available information is:
3746
3747 PCRE_CONFIG_UTF8
3748
3749 The output is an integer that is set to one if UTF-8 support is available;
3750 otherwise it is set to zero.
3751
3752 PCRE_CONFIG_NEWLINE
3753
3754 The output is an integer that it set to the value of the code that is used for
3755 newline. It is either LF (10) or CR (13).
3756
3757 PCRE_CONFIG_LINK_SIZE
3758
3759 The output is an integer that contains the number of bytes used for internal
3760 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
3761
3762 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
3763
3764 The output is an integer that contains the threshold above which the POSIX
3765 interface uses malloc() for output vectors. See item 31 above.
3766
3767 PCRE_CONFIG_MATCH_LIMIT
3768
3769 The output is an unsigned integer that contains the default limit of the number
3770 of match() calls in a pcre_exec() execution. See 42 above.
3771
3772 44. pcretest has been upgraded by the addition of the -C option. This causes it
3773 to extract all the available output from the new pcre_config() function, and to
3774 output it. The program then exits immediately.
3775
3776 45. A need has arisen to pass over additional data with calls to pcre_exec() in
3777 order to support additional features. One way would have been to define
3778 pcre_exec2() (for example) with extra arguments, but this would not have been
3779 extensible, and would also have required all calls to the original function to
3780 be mapped to the new one. Instead, I have chosen to extend the mechanism that
3781 is used for passing in "extra" data from pcre_study().
3782
3783 The pcre_extra structure is now exposed and defined in pcre.h. It currently
3784 contains the following fields:
3785
3786 flags a bitmap indicating which of the following fields are set
3787 study_data opaque data from pcre_study()
3788 match_limit a way of specifying a limit on match() calls for a specific
3789 call to pcre_exec()
3790 callout_data data for callouts (see 49 below)
3791
3792 The flag bits are also defined in pcre.h, and are
3793
3794 PCRE_EXTRA_STUDY_DATA
3795 PCRE_EXTRA_MATCH_LIMIT
3796 PCRE_EXTRA_CALLOUT_DATA
3797
3798 The pcre_study() function now returns one of these new pcre_extra blocks, with
3799 the actual study data pointed to by the study_data field, and the
3800 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
3801 before. That is, this change is entirely upwards-compatible and requires no
3802 change to existing code.
3803
3804 If you want to pass in additional data to pcre_exec(), you can either place it
3805 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
3806 block.
3807
3808 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
3809 data string contains the escape sequence \M, pcretest calls pcre_exec() several
3810 times with different match limits, until it finds the minimum value needed for
3811 pcre_exec() to complete. The value is then output. This can be instructive; for
3812 most simple matches the number is quite small, but for pathological cases it
3813 gets very large very quickly.
3814
3815 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
3816 returns the size of the data block pointed to by the study_data field in a
3817 pcre_extra block, that is, the value that was passed as the argument to
3818 pcre_malloc() when PCRE was getting memory in which to place the information
3819 created by pcre_study(). The fourth argument should point to a size_t variable.
3820 pcretest has been extended so that this information is shown after a successful
3821 pcre_study() call when information about the compiled regex is being displayed.
3822
3823 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
3824 because what follows is always an absolute path. (Later: it turns out that this
3825 is more than cosmetic for MinGW, because it doesn't like empty path
3826 components.)
3827
3828 49. Some changes have been made to the callout feature (see 28 above):
3829
3830 (i) A callout function now has three choices for what it returns:
3831
3832 0 => success, carry on matching
3833 > 0 => failure at this point, but backtrack if possible
3834 < 0 => serious error, return this value from pcre_exec()
3835
3836 Negative values should normally be chosen from the set of PCRE_ERROR_xxx
3837 values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
3838 "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
3839 use by callout functions. It will never be used by PCRE itself.
3840
3841 (ii) The pcre_extra structure (see 45 above) has a void * field called
3842 callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
3843 pcre_callout_block structure has a field of the same name. The contents of
3844 the field passed in the pcre_extra structure are passed to the callout
3845 function in the corresponding field in the callout block. This makes it
3846 easier to use the same callout-containing regex from multiple threads. For
3847 testing, the pcretest program has a new data escape
3848
3849 \C*n pass the number n (may be negative) as callout_data
3850
3851 If the callout function in pcretest receives a non-zero value as
3852 callout_data, it returns that value.
3853
3854 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
3855 there were some redundant $(CFLAGS) in commands that are now specified as
3856 $(LINK), which already includes $(CFLAGS).
3857
3858 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
3859 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
3860 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
3861 one-byte characters throughout. Note that case-insensitive matching applies
3862 only to characters whose values are less than 256. PCRE doesn't support the
3863 notion of cases for higher-valued characters.
3864
3865 (i) A character class whose characters are all within 0-255 is handled as
3866 a bit map, and the map is inverted for negative classes. Previously, a
3867 character > 255 always failed to match such a class; however it should
3868 match if the class was a negative one (e.g. [^ab]). This has been fixed.
3869
3870 (ii) A negated character class with a single character < 255 is coded as
3871 "not this character" (OP_NOT). This wasn't working properly when the test
3872 character was multibyte, either singly or repeated.
3873
3874 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
3875 mode, for example: \x{100}{2,3}.
3876
3877 (iv) The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
3878 singly or repeated) now correctly test multibyte characters. However,
3879 PCRE doesn't recognize any characters with values greater than 255 as
3880 digits, spaces, or word characters. Such characters always match \D, \S,
3881 and \W, and never match \d, \s, or \w.
3882
3883 (v) Classes may now contain characters and character ranges with values
3884 greater than 255. For example: [ab\x{100}-\x{400}].
3885
3886 (vi) pcregrep now has a --utf-8 option (synonym -u) which makes it call
3887 PCRE in UTF-8 mode.
3888
3889 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
3890 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
3891 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
3892 value.)
3893
3894 53. The single man page has become too large. I have therefore split it up into
3895 a number of separate man pages. These also give rise to individual HTML pages;
3896 these are now put in a separate directory, and there is an index.html page that
3897 lists them all. Some hyperlinking between the pages has been installed.
3898
3899 54. Added convenience functions for handling named capturing parentheses.
3900
3901 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
3902 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
3903 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
3904 are faulted.
3905
3906 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
3907 calling configure. These values are used when compiling the dftables.c program
3908 which is run to generate the source of the default character tables. They
3909 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
3910 you will need to set these values.
3911
3912 57. Updated the building process for Windows DLL, as provided by Fred Cox.
3913
3914
3915 Version 3.9 02-Jan-02
3916 ---------------------
3917
3918 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
3919
3920 2. If --disable-static was given, the building process failed when trying to
3921 build pcretest and pcregrep. (For some reason it was using libtool to compile
3922 them, which is not right, as they aren't part of the library.)
3923
3924
3925 Version 3.8 18-Dec-01
3926 ---------------------
3927
3928 1. The experimental UTF-8 code was completely screwed up. It was packing the
3929 bytes in the wrong order. How dumb can you get?
3930
3931
3932 Version 3.7 29-Oct-01
3933 ---------------------
3934
3935 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
3936 This caused pcretest, when used on the test data, to segfault. Unfortunately,
3937 this didn't happen under Solaris 8, where I normally test things.
3938
3939 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
3940 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
3941 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
3942 this fix an hour or so after the initial 3.7 release.)
3943
3944
3945 Version 3.6 23-Oct-01
3946 ---------------------
3947
3948 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
3949 offsets passed as NULL with zero offset count.
3950
3951 2. The config.guess and config.sub files had not been updated when I moved to
3952 the latest autoconf.
3953
3954
3955 Version 3.5 15-Aug-01
3956 ---------------------
3957
3958 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
3959 had been forgotten.
3960
3961 2. By using declared but undefined structures, we can avoid using "void"
3962 definitions in pcre.h while keeping the internal definitions of the structures
3963 private.
3964
3965 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
3966 user point of view, this means that both static and shared libraries are built
3967 by default, but this can be individually controlled. More of the work of
3968 handling this static/shared cases is now inside libtool instead of PCRE's make
3969 file.
3970
3971 4. The pcretest utility is now installed along with pcregrep because it is
3972 useful for users (to test regexs) and by doing this, it automatically gets
3973 relinked by libtool. The documentation has been turned into a man page, so
3974 there are now .1, .txt, and .html versions in /doc.
3975
3976 5. Upgrades to pcregrep:
3977 (i) Added long-form option names like gnu grep.
3978 (ii) Added --help to list all options with an explanatory phrase.
3979 (iii) Added -r, --recursive to recurse into sub-directories.
3980 (iv) Added -f, --file to read patterns from a file.
3981
3982 6. pcre_exec() was referring to its "code" argument before testing that
3983 argument for NULL (and giving an error if it was NULL).
3984
3985 7. Upgraded Makefile.in to allow for compiling in a different directory from
3986 the source directory.
3987
3988 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
3989 options bits, the pointer it was passed was to an int instead of to an unsigned
3990 long int. This mattered only on 64-bit systems.
3991
3992 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
3993 generated) instead of pcre.in, which it its source. Also made the same change
3994 in several of the .c files.
3995
3996 10. A new release of gcc defines printf() as a macro, which broke pcretest
3997 because it had an ifdef in the middle of a string argument for printf(). Fixed
3998 by using separate calls to printf().
3999
4000 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
4001 script, to force use of CR or LF instead of \n in the source. On non-Unix
4002 systems, the value can be set in config.h.
4003
4004 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
4005 absolute limit. Changed the text of the error message to make this clear, and
4006 likewise updated the man page.
4007
4008 13. The limit of 99 on the number of capturing subpatterns has been removed.
4009 The new limit is 65535, which I hope will not be a "real" limit.
4010
4011
4012 Version 3.4 22-Aug-00
4013 ---------------------
4014
4015 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
4016
4017 2. Diagnose condition (?(0) as an error instead of crashing on matching.
4018
4019
4020 Version 3.3 01-Aug-00
4021 ---------------------
4022
4023 1. If an octal character was given, but the value was greater than \377, it
4024 was not getting masked to the least significant bits, as documented. This could
4025 lead to crashes in some systems.
4026
4027 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
4028 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
4029
4030 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
4031 These just pass their arguments on to (pcre_free)(), but they are provided
4032 because some uses of PCRE bind it to non-C systems that can call its functions,
4033 but cannot call free() or pcre_free() directly.
4034
4035 4. Add "make test" as a synonym for "make check". Corrected some comments in
4036 the Makefile.
4037
4038 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
4039 Makefile.
4040
4041 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
4042 command called pgrep for grepping around the active processes.
4043
4044 7. Added the beginnings of support for UTF-8 character strings.
4045
4046 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
4047 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
4048 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
4049 out for the ar command.)
4050
4051
4052 Version 3.2 12-May-00
4053 ---------------------
4054
4055 This is purely a bug fixing release.
4056
4057 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
4058 of ZA. This was just one example of several cases that could provoke this bug,
4059 which was introduced by change 9 of version 2.00. The code for breaking
4060 infinite loops after an iteration that matches an empty string was't working
4061 correctly.
4062
4063 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
4064 when matched against abbab (for example). After matching an empty string, it
4065 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
4066 caused it to match further down the string than it should.
4067
4068 3. The code contained an inclusion of sys/types.h. It isn't clear why this
4069 was there because it doesn't seem to be needed, and it causes trouble on some
4070 systems, as it is not a Standard C header. It has been removed.
4071
4072 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
4073 were reported on the Macintosh. The changes were from
4074
4075 while ((c = *(++ptr)) != 0 && c != '\n');
4076 to
4077 while ((c = *(++ptr)) != 0 && c != '\n') ;
4078
4079 Totally extraordinary, but if that's what it takes...
4080
4081 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
4082 available. Added HAVE_BCOPY and an autoconf test for it; if neither
4083 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
4084 assumes the way PCRE uses memmove() (always moving upwards).
4085
4086 6. PCRE is being used in one environment where strchr() is not available. There
4087 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
4088 faster code anyway.
4089
4090
4091 Version 3.1 09-Feb-00
4092 ---------------------
4093
4094 The only change in this release is the fixing of some bugs in Makefile.in for
4095 the "install" target:
4096
4097 (1) It was failing to install pcreposix.h.
4098
4099 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
4100
4101
4102 Version 3.0 01-Feb-00
4103 ---------------------
4104
4105 1. Add support for the /+ modifier to perltest (to output $` like it does in
4106 pcretest).
4107
4108 2. Add support for the /g modifier to perltest.
4109
4110 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
4111 matches null strings.
4112
4113 4. Fix perltest so that it doesn't do unwanted things when fed an empty
4114 pattern. Perl treats empty patterns specially - it reuses the most recent
4115 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
4116 effect.
4117
4118 5. The POSIX interface was broken in that it was just handing over the POSIX
4119 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
4120 required a bigger vector, with some working space on the end. This means that
4121 the POSIX wrapper now has to get and free some memory, and copy the results.
4122
4123 6. Added some simple autoconf support, placing the test data and the
4124 documentation in separate directories, re-organizing some of the
4125 information files, and making it build pcre-config (a GNU standard). Also added
4126 libtool support for building PCRE as a shared library, which is now the
4127 default.
4128
4129 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
4130 09 are not valid octal constants. Single digits will be used for minor values
4131 less than 10.
4132
4133 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
4134 existing programs that set these in the POSIX interface can use PCRE without
4135 modification.
4136
4137 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
4138 return all that pcre_info() returns, plus additional data. The pcre_info()
4139 function is retained for compatibility, but is considered to be obsolete.
4140
4141 10. Added experimental recursion feature (?R) to handle one common case that
4142 Perl 5.6 will be able to do with (?p{...}).
4143
4144 11. Added support for POSIX character classes like [:alpha:], which Perl is
4145 adopting.
4146
4147
4148 Version 2.08 31-Aug-99
4149 ----------------------
4150
4151 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
4152 trying to match at the startoffset position, but instead was moving forward to
4153 the next newline as if a previous match had failed.
4154
4155 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
4156 and could get into a loop if a null string was matched other than at the start
4157 of the subject.
4158
4159 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
4160 be distinguished at compile time, and for completeness also added PCRE_DATE.
4161
4162 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
4163 in GnuWin32 environments.
4164
4165
4166 Version 2.07 29-Jul-99
4167 ----------------------
4168
4169 1. The documentation is now supplied in plain text form and HTML as well as in
4170 the form of man page sources.
4171
4172 2. C++ compilers don't like assigning (void *) values to other pointer types.
4173 In particular this affects malloc(). Although there is no problem in Standard
4174 C, I've put in casts to keep C++ compilers happy.
4175
4176 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
4177 should be (const char *).
4178
4179 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
4180 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
4181 However, I haven't made this a standard facility. The documentation doesn't
4182 mention it, and the Makefile doesn't support it.
4183
4184 5. The Makefile now contains an "install" target, with editable destinations at
4185 the top of the file. The pcretest program is not installed.
4186
4187 6. pgrep -V now gives the PCRE version number and date.
4188
4189 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
4190 causing the entire string to be ignored, instead of just the last character.
4191
4192 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
4193 non-matching string, it can take a very, very long time, even for strings of
4194 quite modest length, because of the nested recursion. PCRE now does better in
4195 some of these cases. It does this by remembering the last required literal
4196 character in the pattern, and pre-searching the subject to ensure it is present
4197 before running the real match. In other words, it applies a heuristic to detect
4198 some types of certain failure quickly, and in the above example, if presented
4199 with a string that has no trailing " it gives "no match" very quickly.
4200
4201 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
4202 other alternatives are tried instead.
4203
4204
4205 Version 2.06 09-Jun-99
4206 ----------------------
4207
4208 1. Change pcretest's output for amount of store used to show just the code
4209 space, because the remainder (the data block) varies in size between 32-bit and
4210 64-bit systems.
4211
4212 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
4213 start matching at. This allows lookbehinds to work when searching for multiple
4214 occurrences in a string.
4215
4216 3. Added additional options to pcretest for testing multiple occurrences:
4217
4218 /+ outputs the rest of the string that follows a match
4219 /g loops for multiple occurrences, using the new startoffset argument
4220 /G loops for multiple occurrences by passing an incremented pointer
4221
4222 4. PCRE wasn't doing the "first character" optimization for patterns starting
4223 with \b or \B, though it was doing it for other lookbehind assertions. That is,
4224 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
4225 the letter 'x'. On long subject strings, this gives a significant speed-up.
4226
4227
4228 Version 2.05 21-Apr-99
4229 ----------------------
4230
4231 1. Changed the type of magic_number from int to long int so that it works
4232 properly on 16-bit systems.
4233
4234 2. Fixed a bug which caused patterns starting with .* not to work correctly
4235 when the subject string contained newline characters. PCRE was assuming
4236 anchoring for such patterns in all cases, which is not correct because .* will
4237 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
4238 DOTALL is set at top level; otherwise it knows that patterns starting with .*
4239 must be retried after every newline in the subject.
4240
4241
4242 Version 2.04 18-Feb-99
4243 ----------------------
4244
4245 1. For parenthesized subpatterns with repeats whose minimum was zero, the
4246 computation of the store needed to hold the pattern was incorrect (too large).
4247 If such patterns were nested a few deep, this could multiply and become a real
4248 problem.
4249
4250 2. Added /M option to pcretest to show the memory requirement of a specific
4251 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
4252
4253 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
4254 compiled in such a way that the backtracking after subsequent failure was
4255 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
4256 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
4257
4258
4259 Version 2.03 02-Feb-99
4260 ----------------------
4261
4262 1. Fixed typo and small mistake in man page.
4263
4264 2. Added 4th condition (GPL supersedes if conflict) and created separate
4265 LICENCE file containing the conditions.
4266
4267 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
4268 Perl, that is the internal \ allows the delimiter to be included in the
4269 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
4270 the final delimiter, add \ to the end of the pattern (to test the error).
4271
4272 4. Added the convenience functions for extracting substrings after a successful
4273 match. Updated pcretest to make it able to test these functions.
4274
4275
4276 Version 2.02 14-Jan-99
4277 ----------------------
4278
4279 1. Initialized the working variables associated with each extraction so that
4280 their saving and restoring doesn't refer to uninitialized store.
4281
4282 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
4283 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
4284 fix the problem.
4285
4286 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
4287 calls, and wasn't printing the correct value for compiling calls. Increased the
4288 default value of LOOPREPEAT, and the number of significant figures in the
4289 times.
4290
4291 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
4292
4293 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
4294 a building problem on Windows NT with a FAT file system.
4295
4296
4297 Version 2.01 21-Oct-98
4298 ----------------------
4299
4300 1. Changed the API for pcre_compile() to allow for the provision of a pointer
4301 to character tables built by pcre_maketables() in the current locale. If NULL
4302 is passed, the default tables are used.
4303
4304
4305 Version 2.00 24-Sep-98
4306 ----------------------
4307
4308 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
4309 it any more.
4310
4311 2. Allow quantification of (?>) groups, and make it work correctly.
4312
4313 3. The first character computation wasn't working for (?>) groups.
4314
4315 4. Correct the implementation of \Z (it is permitted to match on the \n at the
4316 end of the subject) and add 5.005's \z, which really does match only at the
4317 very end of the subject.
4318
4319 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
4320
4321 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
4322 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
4323 localized options. All options to pcre_study() were also removed.
4324
4325 7. Add other new features from 5.005:
4326
4327 $(?<= positive lookbehind
4328 $(?<! negative lookbehind
4329 (?imsx-imsx) added the unsetting capability
4330 such a setting is global if at outer level; local otherwise
4331 (?imsx-imsx:) non-capturing groups with option setting
4332 (?(cond)re|re) conditional pattern matching
4333
4334 A backreference to itself in a repeated group matches the previous
4335 captured string.
4336
4337 8. General tidying up of studying (both automatic and via "study")
4338 consequential on the addition of new assertions.
4339
4340 9. As in 5.005, unlimited repeated groups that could match an empty substring
4341 are no longer faulted at compile time. Instead, the loop is forcibly broken at
4342 runtime if any iteration does actually match an empty substring.
4343
4344 10. Include the RunTest script in the distribution.
4345
4346 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
4347 discrepancies, some of which were old and were also with respect to 5.004. They
4348 have now been fixed.
4349
4350
4351 Version 1.09 28-Apr-98
4352 ----------------------
4353
4354 1. A negated single character class followed by a quantifier with a minimum
4355 value of one (e.g. [^x]{1,6} ) was not compiled correctly. This could lead to
4356 program crashes, or just wrong answers. This did not apply to negated classes
4357 containing more than one character, or to minima other than one.
4358
4359
4360 Version 1.08 27-Mar-98
4361 ----------------------
4362
4363 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
4364
4365 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
4366 latter must appear before anything that relies on it in the pattern.
4367
4368
4369 Version 1.07 16-Feb-98
4370 ----------------------
4371
4372 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
4373 repeat of a potentially empty string).
4374
4375
4376 Version 1.06 23-Jan-98
4377 ----------------------
4378
4379 1. Added Markus Oberhumer's little patches for C++.
4380
4381 2. Literal strings longer than 255 characters were broken.
4382
4383
4384 Version 1.05 23-Dec-97
4385 ----------------------
4386
4387 1. Negated character classes containing more than one character were failing if
4388 PCRE_CASELESS was set at run time.
4389
4390
4391 Version 1.04 19-Dec-97
4392 ----------------------
4393
4394 1. Corrected the man page, where some "const" qualifiers had been omitted.
4395
4396 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
4397 input syntax.
4398
4399 3. Fixed memory leak which occurred when a regex with back references was
4400 matched with an offsets vector that wasn't big enough. The temporary memory
4401 that is used in this case wasn't being freed if the match failed.
4402
4403 4. Tidied pcretest to ensure it frees memory that it gets.
4404
4405 5. Temporary memory was being obtained in the case where the passed offsets
4406 vector was exactly big enough.
4407
4408 6. Corrected definition of offsetof() from change 5 below.
4409
4410 7. I had screwed up change 6 below and broken the rules for the use of
4411 setjmp(). Now fixed.
4412
4413
4414 Version 1.03 18-Dec-97
4415 ----------------------
4416
4417 1. A erroneous regex with a missing opening parenthesis was correctly
4418 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
4419 on some systems.
4420
4421 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
4422 it was reported that one broken compiler failed on the former because "code" is
4423 also an independent variable.
4424
4425 3. The erroneous regex a[]b caused an array overrun reference.
4426
4427 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
4428 fail on data ending with that character. (It was going on too far, and checking
4429 the next character, typically a binary zero.) This was specific to the
4430 optimized code for single-character negative classes.
4431
4432 5. Added a contributed patch from the TIN world which does the following:
4433
4434 + Add an undef for memmove, in case the the system defines a macro for it.
4435
4436 + Add a definition of offsetof(), in case there isn't one. (I don't know
4437 the reason behind this - offsetof() is part of the ANSI standard - but
4438 it does no harm).
4439
4440 + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
4441 most of the places where whitespace preceded '#'. I have given up and
4442 allowed the remaining 2 cases to be at the margin.
4443
4444 + Rename some variables in pcre to eliminate shadowing. This seems very
4445 pedantic, but does no harm, of course.
4446
4447 6. Moved the call to setjmp() into its own function, to get rid of warnings
4448 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
4449
4450 7. Constructs such as \d{8,} were compiling into the equivalent of
4451 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
4452 outcome, but in this particular case used more store than had been allocated,
4453 which caused the bug to be discovered because it threw up an internal error.
4454
4455 8. The debugging code in both pcre and pcretest for outputting the compiled
4456 form of a regex was going wrong in the case of back references followed by
4457 curly-bracketed repeats.
4458
4459
4460 Version 1.02 12-Dec-97
4461 ----------------------
4462
4463 1. Typos in pcre.3 and comments in the source fixed.
4464
4465 2. Applied a contributed patch to get rid of places where it used to remove
4466 'const' from variables, and fixed some signed/unsigned and uninitialized
4467 variable warnings.
4468
4469 3. Added the "runtest" target to Makefile.
4470
4471 4. Set default compiler flag to -O2 rather than just -O.
4472
4473
4474 Version 1.01 19-Nov-97
4475 ----------------------
4476
4477 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
4478 like /([ab]*)*/, that is, for classes with more than one character in them.
4479
4480 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
4481 as /((?>a*))*/ (a PCRE_EXTRA facility).
4482
4483
4484 Version 1.00 18-Nov-97
4485 ----------------------
4486
4487 1. Added compile-time macros to support systems such as SunOS4 which don't have
4488 memmove() or strerror() but have other things that can be used instead.
4489
4490 2. Arranged that "make clean" removes the executables.
4491
4492
4493 Version 0.99 27-Oct-97
4494 ----------------------
4495
4496 1. Fixed bug in code for optimizing classes with only one character. It was
4497 initializing a 32-byte map regardless, which could cause it to run off the end
4498 of the memory it had got.
4499
4500 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
4501
4502
4503 Version 0.98 22-Oct-97
4504 ----------------------
4505
4506 1. Fixed bug in code for handling temporary memory usage when there are more
4507 back references than supplied space in the ovector. This could cause segfaults.
4508
4509
4510 Version 0.97 21-Oct-97
4511 ----------------------
4512
4513 1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
4514
4515 2. Optimized negated single characters not to use a bit map.
4516
4517 3. Brought error texts together as macro definitions; clarified some of them;
4518 fixed one that was wrong - it said "range out of order" when it meant "invalid
4519 escape sequence".
4520
4521 4. Changed some char * arguments to const char *.
4522
4523 5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
4524
4525 6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
4526 pcretest.
4527
4528
4529 Version 0.96 16-Oct-97
4530 ----------------------
4531
4532 1. Added a simple "pgrep" utility to the distribution.
4533
4534 2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
4535 unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
4536 where "ddd" means "one or more decimal digits".
4537
4538 3. Fixed serious bug. If a pattern had a back reference, but the call to
4539 pcre_exec() didn't supply a large enough ovector to record the related
4540 identifying subpattern, the match always failed. PCRE now remembers the number
4541 of the largest back reference, and gets some temporary memory in which to save
4542 the offsets during matching if necessary, in order to ensure that
4543 backreferences always work.
4544
4545 4. Increased the compatibility with Perl in a number of ways:
4546
4547 (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
4548 to request this handling. The option can be set at compile or exec time.
4549
4550 (b) $ matches before a terminating newline by default; an option
4551 PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
4552 mode). The option can be set at compile or exec time.
4553
4554 (c) The handling of \ followed by a digit other than 0 is now supposed to be
4555 the same as Perl's. If the decimal number it represents is less than 10
4556 or there aren't that many previous left capturing parentheses, an octal
4557 escape is read. Inside a character class, it's always an octal escape,
4558 even if it is a single digit.
4559
4560 (d) An escaped but undefined alphabetic character is taken as a literal,
4561 unless PCRE_EXTRA is set. Currently this just reserves the remaining
4562 escapes.
4563
4564 (e) {0} is now permitted. (The previous item is removed from the compiled
4565 pattern).
4566
4567 5. Changed all the names of code files so that the basic parts are no longer
4568 than 10 characters, and abolished the teeny "globals.c" file.
4569
4570 6. Changed the handling of character classes; they are now done with a 32-byte
4571 bit map always.
4572
4573 7. Added the -d and /D options to pcretest to make it possible to look at the
4574 internals of compilation without having to recompile pcre.
4575
4576
4577 Version 0.95 23-Sep-97
4578 ----------------------
4579
4580 1. Fixed bug in pre-pass concerning escaped "normal" characters such as \x5c or
4581 \x20 at the start of a run of normal characters. These were being treated as
4582 real characters, instead of the source characters being re-checked.
4583
4584
4585 Version 0.94 18-Sep-97
4586 ----------------------
4587
4588 1. The functions are now thread-safe, with the caveat that the global variables
4589 containing pointers to malloc() and free() or alternative functions are the
4590 same for all threads.
4591
4592 2. Get pcre_study() to generate a bitmap of initial characters for non-
4593 anchored patterns when this is possible, and use it if passed to pcre_exec().
4594
4595
4596 Version 0.93 15-Sep-97
4597 ----------------------
4598
4599 1. /(b)|(:+)/ was computing an incorrect first character.
4600
4601 2. Add pcre_study() to the API and the passing of pcre_extra to pcre_exec(),
4602 but not actually doing anything yet.
4603
4604 3. Treat "-" characters in classes that cannot be part of ranges as literals,
4605 as Perl does (e.g. [-az] or [az-]).
4606
4607 4. Set the anchored flag if a branch starts with .* or .*? because that tests
4608 all possible positions.
4609
4610 5. Split up into different modules to avoid including unneeded functions in a
4611 compiled binary. However, compile and exec are still in one module. The "study"
4612 function is split off.
4613
4614 6. The character tables are now in a separate module whose source is generated
4615 by an auxiliary program - but can then be edited by hand if required. There are
4616 now no calls to isalnum(), isspace(), isdigit(), isxdigit(), tolower() or
4617 toupper() in the code.
4618
4619 7. Turn the malloc/free funtions variables into pcre_malloc and pcre_free and
4620 make them global. Abolish the function for setting them, as the caller can now
4621 set them directly.
4622
4623
4624 Version 0.92 11-Sep-97
4625 ----------------------
4626
4627 1. A repeat with a fixed maximum and a minimum of 1 for an ordinary character
4628 (e.g. /a{1,3}/) was broken (I mis-optimized it).
4629
4630 2. Caseless matching was not working in character classes if the characters in
4631 the pattern were in upper case.
4632
4633 3. Make ranges like [W-c] work in the same way as Perl for caseless matching.
4634
4635 4. Make PCRE_ANCHORED public and accept as a compile option.
4636
4637 5. Add an options word to pcre_exec() and accept PCRE_ANCHORED and
4638 PCRE_CASELESS at run time. Add escapes \A and \I to pcretest to cause it to
4639 pass them.
4640
4641 6. Give an error if bad option bits passed at compile or run time.
4642
4643 7. Add PCRE_MULTILINE at compile and exec time, and (?m) as well. Add \M to
4644 pcretest to cause it to pass that flag.
4645
4646 8. Add pcre_info(), to get the number of identifying subpatterns, the stored
4647 options, and the first character, if set.
4648
4649 9. Recognize C+ or C{n,m} where n >= 1 as providing a fixed starting character.
4650
4651
4652 Version 0.91 10-Sep-97
4653 ----------------------
4654
4655 1. PCRE was failing to diagnose unlimited repeats of subpatterns that could
4656 match the empty string as in /(a*)*/. It was looping and ultimately crashing.
4657
4658 2. PCRE was looping on encountering an indefinitely repeated back reference to
4659 a subpattern that had matched an empty string, e.g. /(a|)\1*/. It now does what
4660 Perl does - treats the match as successful.
4661
4662 ****

Properties

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

  ViewVC Help
Powered by ViewVC 1.1.5