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

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log


Revision 784 - (show annotations)
Mon Dec 5 12:33:44 2011 UTC (3 years, 8 months ago) by ph10
File size: 222875 byte(s)
Error occurred while calculating annotation data.
Tidies for 8.21-RC1 release.
1 ChangeLog for PCRE
2 ------------------
3
4 Version 8.21 05-Dec-2011
5 ------------------------
6
7 1. Updating the JIT compiler.
8
9 2. JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
10 are added as well.
11
12 3. Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
13 PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
14 calling _pcre_jit_exec. Some extra comments are added.
15
16 4. (*MARK) settings inside atomic groups that do not contain any capturing
17 parentheses, for example, (?>a(*:m)), were not being passed out. This bug
18 was introduced by change 18 for 8.20.
19
20 5. Supporting of \x, \U and \u in JavaScript compatibility mode based on the
21 ECMA-262 standard.
22
23 6. Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
24 erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
25 This bug was probably introduced by change 9 of 8.13.
26
27 7. While fixing 6 above, I noticed that a number of other items were being
28 incorrectly rejected as "not fixed length". This arose partly because newer
29 opcodes had not been added to the fixed-length checking code. I have (a)
30 corrected the bug and added tests for these items, and (b) arranged for an
31 error to occur if an unknown opcode is encountered while checking for fixed
32 length instead of just assuming "not fixed length". The items that were
33 rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
34 (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
35 repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
36
37 8. A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
38 being incorrectly compiled and would have given unpredicatble results.
39
40 9. A possessively repeated subpattern with minimum repeat count greater than
41 one behaved incorrectly. For example, (A){2,}+ behaved as if it was
42 (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
43 the first (A) could occur when it should not.
44
45 10. Add a cast and remove a redundant test from the code.
46
47 11. JIT should use pcre_malloc/pcre_free for allocation.
48
49 12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
50 best practice nowadays, and helps with cross-compiling. (If the exec_prefix
51 is anything other than /usr, -L is still shown).
52
53 13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
54
55 14. Perl does not support \N without a following name in a [] class; PCRE now
56 also gives an error.
57
58 15. If a forward reference was repeated with an upper limit of around 2000,
59 it caused the error "internal error: overran compiling workspace". The
60 maximum number of forward references (including repeats) was limited by the
61 internal workspace, and dependent on the LINK_SIZE. The code has been
62 rewritten so that the workspace expands (via pcre_malloc) if necessary, and
63 the default depends on LINK_SIZE. There is a new upper limit (for safety)
64 of around 200,000 forward references. While doing this, I also speeded up
65 the filling in of repeated forward references.
66
67 16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
68 incorrectly expecting the subject to contain another "a" after the start.
69
70 17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
71 in the match, the SKIP should be ignored. This was not happening; instead
72 the SKIP was being treated as NOMATCH. For patterns such as
73 /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
74 tested.
75
76 18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
77 now much more compatible with Perl, in particular in cases where the result
78 is a non-match for a non-anchored pattern. For example, if
79 /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
80 "m", where previously it did not return a name. A side effect of this
81 change is that for partial matches, the last encountered mark name is
82 returned, as for non matches. A number of tests that were previously not
83 Perl-compatible have been moved into the Perl-compatible test files. The
84 refactoring has had the pleasing side effect of removing one argument from
85 the match() function, thus reducing its stack requirements.
86
87 19. If the /S+ option was used in pcretest to study a pattern using JIT,
88 subsequent uses of /S (without +) incorrectly behaved like /S+.
89
90 21. Retrieve executable code size support for the JIT compiler and fixing
91 some warnings.
92
93 22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
94 not work when the shorter character appeared right at the end of the
95 subject string.
96
97 23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
98 systems.
99
100 24. Added PCRE_INFO_JITSIZE to pass on the value from (21) above, and also
101 output it when the /M option is used in pcretest.
102
103 25. The CheckMan script was not being included in the distribution. Also, added
104 an explicit "perl" to run Perl scripts from the PrepareRelease script
105 because this is reportedly needed in Windows.
106
107
108 Version 8.20 21-Oct-2011
109 ------------------------
110
111 1. Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
112 a POSIX class. After further experiments with Perl, which convinced me that
113 Perl has bugs and confusions, a closing square bracket is no longer allowed
114 in a POSIX name. This bug also affected patterns with classes that started
115 with full stops.
116
117 2. If a pattern such as /(a)b|ac/ is matched against "ac", there is no
118 captured substring, but while checking the failing first alternative,
119 substring 1 is temporarily captured. If the output vector supplied to
120 pcre_exec() was not big enough for this capture, the yield of the function
121 was still zero ("insufficient space for captured substrings"). This cannot
122 be totally fixed without adding another stack variable, which seems a lot
123 of expense for a edge case. However, I have improved the situation in cases
124 such as /(a)(b)x|abc/ matched against "abc", where the return code
125 indicates that fewer than the maximum number of slots in the ovector have
126 been set.
127
128 3. Related to (2) above: when there are more back references in a pattern than
129 slots in the output vector, pcre_exec() uses temporary memory during
130 matching, and copies in the captures as far as possible afterwards. It was
131 using the entire output vector, but this conflicts with the specification
132 that only 2/3 is used for passing back captured substrings. Now it uses
133 only the first 2/3, for compatibility. This is, of course, another edge
134 case.
135
136 4. Zoltan Herczeg's just-in-time compiler support has been integrated into the
137 main code base, and can be used by building with --enable-jit. When this is
138 done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
139 runtime --no-jit option is given.
140
141 5. When the number of matches in a pcre_dfa_exec() run exactly filled the
142 ovector, the return from the function was zero, implying that there were
143 other matches that did not fit. The correct "exactly full" value is now
144 returned.
145
146 6. If a subpattern that was called recursively or as a subroutine contained
147 (*PRUNE) or any other control that caused it to give a non-standard return,
148 invalid errors such as "Error -26 (nested recursion at the same subject
149 position)" or even infinite loops could occur.
150
151 7. If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
152 computing the minimum length on reaching *ACCEPT, and so ended up with the
153 wrong value of 1 rather than 0. Further investigation indicates that
154 computing a minimum subject length in the presence of *ACCEPT is difficult
155 (think back references, subroutine calls), and so I have changed the code
156 so that no minimum is registered for a pattern that contains *ACCEPT.
157
158 8. If (*THEN) was present in the first (true) branch of a conditional group,
159 it was not handled as intended. [But see 16 below.]
160
161 9. Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
162 Sheri Pierce.
163
164 10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
165 the first byte in a match must be "a".
166
167 11. Change 17 for 8.13 increased the recursion depth for patterns like
168 /a(?:.)*?a/ drastically. I've improved things by remembering whether a
169 pattern contains any instances of (*THEN). If it does not, the old
170 optimizations are restored. It would be nice to do this on a per-group
171 basis, but at the moment that is not feasible.
172
173 12. In some environments, the output of pcretest -C is CRLF terminated. This
174 broke RunTest's code that checks for the link size. A single white space
175 character after the value is now allowed for.
176
177 13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
178 For "fr", it uses the Windows-specific input and output files.
179
180 14. If (*THEN) appeared in a group that was called recursively or as a
181 subroutine, it did not work as intended. [But see next item.]
182
183 15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
184 pattern fragments (but not containing any | characters). If A and B are
185 matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
186 was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
187 D. In other words, Perl considers parentheses that do not contain any |
188 characters to be part of a surrounding alternative, whereas PCRE was
189 treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
190 differently. PCRE now behaves in the same way as Perl, except in the case
191 of subroutine/recursion calls such as (?1) which have in any case always
192 been different (but PCRE had them first :-).
193
194 16. Related to 15 above: Perl does not treat the | in a conditional group as
195 creating alternatives. Such a group is treated in the same way as an
196 ordinary group without any | characters when processing (*THEN). PCRE has
197 been changed to match Perl's behaviour.
198
199 17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
200 RunGrepTest script failed.
201
202 18. Change 22 for version 13 caused atomic groups to use more stack. This is
203 inevitable for groups that contain captures, but it can lead to a lot of
204 stack use in large patterns. The old behaviour has been restored for atomic
205 groups that do not contain any capturing parentheses.
206
207 19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
208 suppress the check for a minimum subject length at run time. (If it was
209 given to pcre_exec() or pcre_dfa_exec() it did work.)
210
211 20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
212 fail to work when decoding hex characters in data strings in EBCDIC
213 environments.
214
215 21. It appears that in at least one Mac OS environment, the isxdigit() function
216 is implemented as a macro that evaluates to its argument more than once,
217 contravening the C 90 Standard (I haven't checked a later standard). There
218 was an instance in pcretest which caused it to go wrong when processing
219 \x{...} escapes in subject strings. The has been rewritten to avoid using
220 things like p++ in the argument of isxdigit().
221
222
223 Version 8.13 16-Aug-2011
224 ------------------------
225
226 1. The Unicode data tables have been updated to Unicode 6.0.0.
227
228 2. Two minor typos in pcre_internal.h have been fixed.
229
230 3. Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
231 pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
232 in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
233
234 4. There were a number of related bugs in the code for matching backrefences
235 caselessly in UTF-8 mode when codes for the characters concerned were
236 different numbers of bytes. For example, U+023A and U+2C65 are an upper
237 and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
238 (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
239 code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
240 2-byte code at the end of the subject (it thought there wasn't enough data
241 left).
242
243 5. Comprehensive information about what went wrong is now returned by
244 pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
245 as the output vector has at least 2 elements. The offset of the start of
246 the failing character and a reason code are placed in the vector.
247
248 6. When the UTF-8 string check fails for pcre_compile(), the offset that is
249 now returned is for the first byte of the failing character, instead of the
250 last byte inspected. This is an incompatible change, but I hope it is small
251 enough not to be a problem. It makes the returned offset consistent with
252 pcre_exec() and pcre_dfa_exec().
253
254 7. pcretest now gives a text phrase as well as the error number when
255 pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
256 failure, the offset and reason code are output.
257
258 8. When \R was used with a maximizing quantifier it failed to skip backwards
259 over a \r\n pair if the subsequent match failed. Instead, it just skipped
260 back over a single character (\n). This seems wrong (because it treated the
261 two characters as a single entity when going forwards), conflicts with the
262 documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
263 behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
264 has been changed.
265
266 9. Some internal refactoring has changed the processing so that the handling
267 of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
268 time (the PCRE_DOTALL option was changed this way some time ago: version
269 7.7 change 16). This has made it possible to abolish the OP_OPT op code,
270 which was always a bit of a fudge. It also means that there is one less
271 argument for the match() function, which reduces its stack requirements
272 slightly. This change also fixes an incompatibility with Perl: the pattern
273 (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
274
275 10. More internal refactoring has drastically reduced the number of recursive
276 calls to match() for possessively repeated groups such as (abc)++ when
277 using pcre_exec().
278
279 11. While implementing 10, a number of bugs in the handling of groups were
280 discovered and fixed:
281
282 (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
283 (a|)*(?1) gave a compile-time internal error.
284 ((a|)+)+ did not notice that the outer group could match an empty string.
285 (^a|^)+ was not marked as anchored.
286 (.*a|.*)+ was not marked as matching at start or after a newline.
287
288 12. Yet more internal refactoring has removed another argument from the match()
289 function. Special calls to this function are now indicated by setting a
290 value in a variable in the "match data" data block.
291
292 13. Be more explicit in pcre_study() instead of relying on "default" for
293 opcodes that mean there is no starting character; this means that when new
294 ones are added and accidentally left out of pcre_study(), testing should
295 pick them up.
296
297 14. The -s option of pcretest has been documented for ages as being an old
298 synonym of -m (show memory usage). I have changed it to mean "force study
299 for every regex", that is, assume /S for every regex. This is similar to -i
300 and -d etc. It's slightly incompatible, but I'm hoping nobody is still
301 using it. It makes it easier to run collections of tests with and without
302 study enabled, and thereby test pcre_study() more easily. All the standard
303 tests are now run with and without -s (but some patterns can be marked as
304 "never study" - see 20 below).
305
306 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
307 restoration of the capturing data to the outer values was not happening
308 correctly.
309
310 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
311 empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
312 pattern had matched an empty string, and so incorrectly returned a no
313 match.
314
315 17. There was optimizing code for the last branch of non-capturing parentheses,
316 and also for the obeyed branch of a conditional subexpression, which used
317 tail recursion to cut down on stack usage. Unfortunately, now that there is
318 the possibility of (*THEN) occurring in these branches, tail recursion is
319 no longer possible because the return has to be checked for (*THEN). These
320 two optimizations have therefore been removed. [But see 8.20/11 above.]
321
322 18. If a pattern containing \R was studied, it was assumed that \R always
323 matched two bytes, thus causing the minimum subject length to be
324 incorrectly computed because \R can also match just one byte.
325
326 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
327 was incorrectly computed.
328
329 20. If /S is present twice on a test pattern in pcretest input, it now
330 *disables* studying, thereby overriding the use of -s on the command line
331 (see 14 above). This is necessary for one or two tests to keep the output
332 identical in both cases.
333
334 21. When (*ACCEPT) was used in an assertion that matched an empty string and
335 PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
336
337 22. When an atomic group that contained a capturing parenthesis was
338 successfully matched, but the branch in which it appeared failed, the
339 capturing was not being forgotten if a higher numbered group was later
340 captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
341 group 1 to "a", when in fact it should be unset. This applied to multi-
342 branched capturing and non-capturing groups, repeated or not, and also to
343 positive assertions (capturing in negative assertions does not happen
344 in PCRE) and also to nested atomic groups.
345
346 23. Add the ++ qualifier feature to pcretest, to show the remainder of the
347 subject after a captured substring, to make it easier to tell which of a
348 number of identical substrings has been captured.
349
350 24. The way atomic groups are processed by pcre_exec() has been changed so that
351 if they are repeated, backtracking one repetition now resets captured
352 values correctly. For example, if ((?>(a+)b)+aabab) is matched against
353 "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
354 "aaa". Previously, it would have been "a". As part of this code
355 refactoring, the way recursive calls are handled has also been changed.
356
357 25. If an assertion condition captured any substrings, they were not passed
358 back unless some other capturing happened later. For example, if
359 (?(?=(a))a) was matched against "a", no capturing was returned.
360
361 26. When studying a pattern that contained subroutine calls or assertions,
362 the code for finding the minimum length of a possible match was handling
363 direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
364 group 1 called group 2 while simultaneously a separate group 2 called group
365 1). A stack overflow occurred in this case. I have fixed this by limiting
366 the recursion depth to 10.
367
368 27. Updated RunTest.bat in the distribution to the version supplied by Tom
369 Fortmann. This supports explicit test numbers on the command line, and has
370 argument validation and error reporting.
371
372 28. An instance of \X with an unlimited repeat could fail if at any point the
373 first character it looked at was a mark character.
374
375 29. Some minor code refactoring concerning Unicode properties and scripts
376 should reduce the stack requirement of match() slightly.
377
378 30. Added the '=' option to pcretest to check the setting of unused capturing
379 slots at the end of the pattern, which are documented as being -1, but are
380 not included in the return count.
381
382 31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
383 compiled something random. Now it gives a compile-time error (as does
384 Perl).
385
386 32. A *MARK encountered during the processing of a positive assertion is now
387 recorded and passed back (compatible with Perl).
388
389 33. If --only-matching or --colour was set on a pcregrep call whose pattern
390 had alternative anchored branches, the search for a second match in a line
391 was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
392 matched the line "0102" twice. The same bug affected patterns that started
393 with a backwards assertion. For example /\b01|\b02/ also matched "0102"
394 twice.
395
396 34. Previously, PCRE did not allow quantification of assertions. However, Perl
397 does, and because of capturing effects, quantifying parenthesized
398 assertions may at times be useful. Quantifiers are now allowed for
399 parenthesized assertions.
400
401 35. A minor code tidy in pcre_compile() when checking options for \R usage.
402
403 36. \g was being checked for fancy things in a character class, when it should
404 just be a literal "g".
405
406 37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
407 appearance of a nested POSIX class supersedes an apparent external class.
408 For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
409 unescaped square brackets may also appear as part of class names. For
410 example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
411 more like Perl. (But see 8.20/1 above.)
412
413 38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
414 was because it thought it was \N{name}, which is not supported).
415
416 39. Add minix to OS list not supporting the -S option in pcretest.
417
418 40. PCRE tries to detect cases of infinite recursion at compile time, but it
419 cannot analyze patterns in sufficient detail to catch mutual recursions
420 such as ((?1))((?2)). There is now a runtime test that gives an error if a
421 subgroup is called recursively as a subpattern for a second time at the
422 same position in the subject string. In previous releases this might have
423 been caught by the recursion limit, or it might have run out of stack.
424
425 41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
426 happen only once. PCRE was, however incorrectly giving a compile time error
427 "recursive call could loop indefinitely" because it cannot analyze the
428 pattern in sufficient detail. The compile time test no longer happens when
429 PCRE is compiling a conditional subpattern, but actual runaway loops are
430 now caught at runtime (see 40 above).
431
432 42. It seems that Perl allows any characters other than a closing parenthesis
433 to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
434 has been changed to be the same.
435
436 43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
437 as not to get warnings when autogen.sh is called. Also changed
438 AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
439
440 44. To help people who use pcregrep to scan files containing exceedingly long
441 lines, the following changes have been made:
442
443 (a) The default value of the buffer size parameter has been increased from
444 8K to 20K. (The actual buffer used is three times this size.)
445
446 (b) The default can be changed by ./configure --with-pcregrep-bufsize when
447 PCRE is built.
448
449 (c) A --buffer-size=n option has been added to pcregrep, to allow the size
450 to be set at run time.
451
452 (d) Numerical values in pcregrep options can be followed by K or M, for
453 example --buffer-size=50K.
454
455 (e) If a line being scanned overflows pcregrep's buffer, an error is now
456 given and the return code is set to 2.
457
458 45. Add a pointer to the latest mark to the callout data block.
459
460 46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
461 partial match of an empty string instead of no match. This was specific to
462 the use of ".".
463
464 47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
465 complete match instead of a partial match. This bug was dependent on both
466 the PCRE_UTF8 and PCRE_DOTALL options being set.
467
468 48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
469 starting byte set, because \b was not being ignored.
470
471
472 Version 8.12 15-Jan-2011
473 ------------------------
474
475 1. Fixed some typos in the markup of the man pages, and wrote a script that
476 checks for such things as part of the documentation building process.
477
478 2. On a big-endian 64-bit system, pcregrep did not correctly process the
479 --match-limit and --recursion-limit options (added for 8.11). In
480 particular, this made one of the standard tests fail. (The integer value
481 went into the wrong half of a long int.)
482
483 3. If the --colour option was given to pcregrep with -v (invert match), it
484 did strange things, either producing crazy output, or crashing. It should,
485 of course, ignore a request for colour when reporting lines that do not
486 match.
487
488 4. Another pcregrep bug caused similar problems if --colour was specified with
489 -M (multiline) and the pattern match finished with a line ending.
490
491 5. In pcregrep, when a pattern that ended with a literal newline sequence was
492 matched in multiline mode, the following line was shown as part of the
493 match. This seems wrong, so I have changed it.
494
495 6. Another pcregrep bug in multiline mode, when --colour was specified, caused
496 the check for further matches in the same line (so they could be coloured)
497 to overrun the end of the current line. If another match was found, it was
498 incorrectly shown (and then shown again when found in the next line).
499
500 7. If pcregrep was compiled under Windows, there was a reference to the
501 function pcregrep_exit() before it was defined. I am assuming this was
502 the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
503 reported by a user. I've moved the definition above the reference.
504
505
506 Version 8.11 10-Dec-2010
507 ------------------------
508
509 1. (*THEN) was not working properly if there were untried alternatives prior
510 to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
511 backtracked to try for "b" instead of moving to the next alternative branch
512 at the same level (in this case, to look for "c"). The Perl documentation
513 is clear that when (*THEN) is backtracked onto, it goes to the "next
514 alternative in the innermost enclosing group".
515
516 2. (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
517 such as (A(*COMMIT)B(*THEN)C|D) any failure after matching A should
518 result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
519 (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
520 (*THEN).
521
522 3. If \s appeared in a character class, it removed the VT character from
523 the class, even if it had been included by some previous item, for example
524 in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
525 of \s, but is part of the POSIX "space" class.)
526
527 4. A partial match never returns an empty string (because you can always
528 match an empty string at the end of the subject); however the checking for
529 an empty string was starting at the "start of match" point. This has been
530 changed to the "earliest inspected character" point, because the returned
531 data for a partial match starts at this character. This means that, for
532 example, /(?<=abc)def/ gives a partial match for the subject "abc"
533 (previously it gave "no match").
534
535 5. Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
536 of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
537 previously a full match would be given. However, setting PCRE_PARTIAL_HARD
538 has an implication that the given string is incomplete (because a partial
539 match is preferred over a full match). For this reason, these items now
540 give a partial match in this situation. [Aside: previously, the one case
541 /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
542 match rather than a full match, which was wrong by the old rules, but is
543 now correct.]
544
545 6. There was a bug in the handling of #-introduced comments, recognized when
546 PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
547 If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
548 UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
549 scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
550 but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
551 places in pcre_compile().
552
553 7. Related to (6) above, when pcre_compile() was skipping #-introduced
554 comments when looking ahead for named forward references to subpatterns,
555 the only newline sequence it recognized was NL. It now handles newlines
556 according to the set newline convention.
557
558 8. SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
559 former, but used strtoul(), whereas pcretest avoided strtoul() but did not
560 cater for a lack of strerror(). These oversights have been fixed.
561
562 9. Added --match-limit and --recursion-limit to pcregrep.
563
564 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
565
566 11. When the -o option was used, pcregrep was setting a return code of 1, even
567 when matches were found, and --line-buffered was not being honoured.
568
569 12. Added an optional parentheses number to the -o and --only-matching options
570 of pcregrep.
571
572 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
573 can match an empty string. The code to do it in pcretest and pcredemo
574 needed fixing:
575
576 (a) When the newline convention was "crlf", pcretest got it wrong, skipping
577 only one byte after an empty string match just before CRLF (this case
578 just got forgotten; "any" and "anycrlf" were OK).
579
580 (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
581 mode when an empty string match preceded an ASCII character followed by
582 a non-ASCII character. (The code for advancing by one character rather
583 than one byte was nonsense.)
584
585 (c) The pcredemo.c sample program did not have any code at all to handle
586 the cases when CRLF is a valid newline sequence.
587
588 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
589 as a starting offset was within the subject string. There is now a new
590 error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
591 negative or greater than the length of the string. In order to test this,
592 pcretest is extended to allow the setting of negative starting offsets.
593
594 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
595 starting offset points to the beginning of a UTF-8 character was
596 unnecessarily clumsy. I tidied it up.
597
598 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
599 bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
600
601 17. Nobody had reported that the --include_dir option, which was added in
602 release 7.7 should have been called --include-dir (hyphen, not underscore)
603 for compatibility with GNU grep. I have changed it to --include-dir, but
604 left --include_dir as an undocumented synonym, and the same for
605 --exclude-dir, though that is not available in GNU grep, at least as of
606 release 2.5.4.
607
608 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
609 characters from a string of bytes) have been redefined so as not to use
610 loops, in order to improve performance in some environments. At the same
611 time, I abstracted some of the common code into auxiliary macros to save
612 repetition (this should not affect the compiled code).
613
614 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
615 compile-time error is now given if \c is not followed by an ASCII
616 character, that is, a byte less than 128. (In EBCDIC mode, the code is
617 different, and any byte value is allowed.)
618
619 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
620 START_OPTIMIZE option, which is now allowed at compile time - but just
621 passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
622 to pcregrep and other applications that have no direct access to PCRE
623 options. The new /Y option in pcretest sets this option when calling
624 pcre_compile().
625
626 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
627 back references. Groups containing recursive back references were forced to
628 be atomic by that change, but in the case of named groups, the amount of
629 memory required was incorrectly computed, leading to "Failed: internal
630 error: code overflow". This has been fixed.
631
632 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
633 pcretest.c, to avoid build problems in some Borland environments.
634
635
636 Version 8.10 25-Jun-2010
637 ------------------------
638
639 1. Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
640 THEN.
641
642 2. (*ACCEPT) was not working when inside an atomic group.
643
644 3. Inside a character class, \B is treated as a literal by default, but
645 faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
646 causes the error). The code is unchanged, but I tidied the documentation.
647
648 4. Inside a character class, PCRE always treated \R and \X as literals,
649 whereas Perl faults them if its -w option is set. I have changed PCRE so
650 that it faults them when PCRE_EXTRA is set.
651
652 5. Added support for \N, which always matches any character other than
653 newline. (It is the same as "." when PCRE_DOTALL is not set.)
654
655 6. When compiling pcregrep with newer versions of gcc which may have
656 FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
657 declared with attribute warn_unused_result" were given. Just casting the
658 result to (void) does not stop the warnings; a more elaborate fudge is
659 needed. I've used a macro to implement this.
660
661 7. Minor change to pcretest.c to avoid a compiler warning.
662
663 8. Added four artifical Unicode properties to help with an option to make
664 \s etc use properties (see next item). The new properties are: Xan
665 (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
666
667 9. Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
668 use Unicode properties. (*UCP) at the start of a pattern can be used to set
669 this option. Modified pcretest to add /W to test this facility. Added
670 REG_UCP to make it available via the POSIX interface.
671
672 10. Added --line-buffered to pcregrep.
673
674 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
675 studied, and the match started with a letter with a code point greater than
676 127 whose first byte was different to the first byte of the other case of
677 the letter, the other case of this starting letter was not recognized
678 (#976).
679
680 12. If a pattern that was studied started with a repeated Unicode property
681 test, for example, \p{Nd}+, there was the theoretical possibility of
682 setting up an incorrect bitmap of starting bytes, but fortunately it could
683 not have actually happened in practice until change 8 above was made (it
684 added property types that matched character-matching opcodes).
685
686 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
687 possible starting bytes for non-anchored patterns.
688
689 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
690 \R, and also a number of cases that involve Unicode properties, both
691 explicit and implicit when PCRE_UCP is set.
692
693 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
694 input, it could crash or give wrong results if characters with values
695 greater than 0xc0 were present in the subject string. (Detail: it assumed
696 UTF-8 input when processing these items.)
697
698 16. Added a lot of (int) casts to avoid compiler warnings in systems where
699 size_t is 64-bit (#991).
700
701 17. Added a check for running out of memory when PCRE is compiled with
702 --disable-stack-for-recursion (#990).
703
704 18. If the last data line in a file for pcretest does not have a newline on
705 the end, a newline was missing in the output.
706
707 19. The default pcre_chartables.c file recognizes only ASCII characters (values
708 less than 128) in its various bitmaps. However, there is a facility for
709 generating tables according to the current locale when PCRE is compiled. It
710 turns out that in some environments, 0x85 and 0xa0, which are Unicode space
711 characters, are recognized by isspace() and therefore were getting set in
712 these tables, and indeed these tables seem to approximate to ISO 8859. This
713 caused a problem in UTF-8 mode when pcre_study() was used to create a list
714 of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
715 which of course cannot start UTF-8 characters. I have changed the code so
716 that only real ASCII characters (less than 128) and the correct starting
717 bytes for UTF-8 encodings are set for characters greater than 127 when in
718 UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
719 altogether.)
720
721 20. Added the /T option to pcretest so as to be able to run tests with non-
722 standard character tables, thus making it possible to include the tests
723 used for 19 above in the standard set of tests.
724
725 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
726 reference to a subpattern the other side of a comment that contains an
727 opening parenthesis caused either an internal compiling error, or a
728 reference to the wrong subpattern.
729
730
731 Version 8.02 19-Mar-2010
732 ------------------------
733
734 1. The Unicode data tables have been updated to Unicode 5.2.0.
735
736 2. Added the option --libs-cpp to pcre-config, but only when C++ support is
737 configured.
738
739 3. Updated the licensing terms in the pcregexp.pas file, as agreed with the
740 original author of that file, following a query about its status.
741
742 4. On systems that do not have stdint.h (e.g. Solaris), check for and include
743 inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
744
745 5. A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
746 quantifier applied to a forward-referencing subroutine call, could compile
747 incorrect code or give the error "internal error: previously-checked
748 referenced subpattern not found".
749
750 6. Both MS Visual Studio and Symbian OS have problems with initializing
751 variables to point to external functions. For these systems, therefore,
752 pcre_malloc etc. are now initialized to local functions that call the
753 relevant global functions.
754
755 7. There were two entries missing in the vectors called coptable and poptable
756 in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
757 I've fixed the data, and added a kludgy way of testing at compile time that
758 the lengths are correct (equal to the number of opcodes).
759
760 8. Following on from 7, I added a similar kludge to check the length of the
761 eint vector in pcreposix.c.
762
763 9. Error texts for pcre_compile() are held as one long string to avoid too
764 much relocation at load time. To find a text, the string is searched,
765 counting zeros. There was no check for running off the end of the string,
766 which could happen if a new error number was added without updating the
767 string.
768
769 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
770
771 11. \K was not working if it appeared in an atomic group or in a group that
772 was called as a "subroutine", or in an assertion. Perl 5.11 documents that
773 \K is "not well defined" if used in an assertion. PCRE now accepts it if
774 the assertion is positive, but not if it is negative.
775
776 12. Change 11 fortuitously reduced the size of the stack frame used in the
777 "match()" function of pcre_exec.c by one pointer. Forthcoming
778 implementation of support for (*MARK) will need an extra pointer on the
779 stack; I have reserved it now, so that the stack frame size does not
780 decrease.
781
782 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
783 item in branch that calls a recursion is a subroutine call - as in the
784 second branch in the above example - was incorrectly given the compile-
785 time error "recursive call could loop indefinitely" because pcre_compile()
786 was not correctly checking the subroutine for matching a non-empty string.
787
788 14. The checks for overrunning compiling workspace could trigger after an
789 overrun had occurred. This is a "should never occur" error, but it can be
790 triggered by pathological patterns such as hundreds of nested parentheses.
791 The checks now trigger 100 bytes before the end of the workspace.
792
793 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
794
795
796 Version 8.01 19-Jan-2010
797 ------------------------
798
799 1. If a pattern contained a conditional subpattern with only one branch (in
800 particular, this includes all (*DEFINE) patterns), a call to pcre_study()
801 computed the wrong minimum data length (which is of course zero for such
802 subpatterns). This could cause incorrect "no match" results.
803
804 2. For patterns such as (?i)a(?-i)b|c where an option setting at the start of
805 the pattern is reset in the first branch, pcre_compile() failed with
806 "internal error: code overflow at offset...". This happened only when
807 the reset was to the original external option setting. (An optimization
808 abstracts leading options settings into an external setting, which was the
809 cause of this.)
810
811 3. A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
812 of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
813 assertion pattern did not match (meaning that the assertion was true), it
814 was incorrectly treated as false if the SKIP had been reached during the
815 matching. This also applied to assertions used as conditions.
816
817 4. If an item that is not supported by pcre_dfa_exec() was encountered in an
818 assertion subpattern, including such a pattern used as a condition,
819 unpredictable results occurred, instead of the error return
820 PCRE_ERROR_DFA_UITEM.
821
822 5. The C++ GlobalReplace function was not working like Perl for the special
823 situation when an empty string is matched. It now does the fancy magic
824 stuff that is necessary.
825
826 6. In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
827 removed. (These were left over from very, very early versions of PCRE.)
828
829 7. Some cosmetic changes to the code to make life easier when compiling it
830 as part of something else:
831
832 (a) Change DEBUG to PCRE_DEBUG.
833
834 (b) In pcre_compile(), rename the member of the "branch_chain" structure
835 called "current" as "current_branch", to prevent a collision with the
836 Linux macro when compiled as a kernel module.
837
838 (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
839 prevent a collision with the Linux macro when compiled as a kernel
840 module.
841
842 8. In pcre_compile() there are some checks for integer overflows that used to
843 cast potentially large values to (double). This has been changed to that
844 when building, a check for int64_t is made, and if it is found, it is used
845 instead, thus avoiding the use of floating point arithmetic. (There is no
846 other use of FP in PCRE.) If int64_t is not found, the fallback is to
847 double.
848
849 9. Added two casts to avoid signed/unsigned warnings from VS Studio Express
850 2005 (difference between two addresses compared to an unsigned value).
851
852 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
853 custom one, because of the following reported problem in Windows:
854
855 - libbz2 uses the Pascal calling convention (WINAPI) for the functions
856 under Win32.
857 - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
858 therefore missing the function definition.
859 - The compiler thus generates a "C" signature for the test function.
860 - The linker fails to find the "C" function.
861 - PCRE fails to configure if asked to do so against libbz2.
862
863 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
864 messages were output:
865
866 Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
867 rerunning libtoolize, to keep the correct libtool macros in-tree.
868 Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
869
870 I have done both of these things.
871
872 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
873 most of the time, it *can* run out if it is given a pattern that contains a
874 runaway infinite recursion. I updated the discussion in the pcrestack man
875 page.
876
877 13. Now that we have gone to the x.xx style of version numbers, the minor
878 version may start with zero. Using 08 or 09 is a bad idea because users
879 might check the value of PCRE_MINOR in their code, and 08 or 09 may be
880 interpreted as invalid octal numbers. I've updated the previous comment in
881 configure.ac, and also added a check that gives an error if 08 or 09 are
882 used.
883
884 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
885 causing partial matching to fail when the end of the subject matched \W
886 in a UTF-8 pattern where \W was quantified with a minimum of 3.
887
888 15. There were some discrepancies between the declarations in pcre_internal.h
889 of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
890 their definitions. The declarations used "const uschar *" and the
891 definitions used USPTR. Even though USPTR is normally defined as "const
892 unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
893 reported that: "This difference in casting confuses some C++ compilers, for
894 example, SunCC recognizes above declarations as different functions and
895 generates broken code for hbpcre." I have changed the declarations to use
896 USPTR.
897
898 16. GNU libtool is named differently on some systems. The autogen.sh script now
899 tries several variants such as glibtoolize (MacOSX) and libtoolize1x
900 (FreeBSD).
901
902 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
903 (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
904 comment: "Figure out how to create a longlong from a string: strtoll and
905 equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
906 instance, but it only takes 2 args instead of 3!"
907
908 18. A subtle bug concerned with back references has been fixed by a change of
909 specification, with a corresponding code fix. A pattern such as
910 ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
911 refers, was giving matches when it shouldn't. For example, xa=xaaa would
912 match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
913 same bug. Such groups have to be quantified to be useful, or contained
914 inside another quantified group. (If there's no repetition, the reference
915 can never match.) The problem arises because, having left the group and
916 moved on to the rest of the pattern, a later failure that backtracks into
917 the group uses the captured value from the final iteration of the group
918 rather than the correct earlier one. I have fixed this in PCRE by forcing
919 any group that contains a reference to itself to be an atomic group; that
920 is, there cannot be any backtracking into it once it has completed. This is
921 similar to recursive and subroutine calls.
922
923
924 Version 8.00 19-Oct-09
925 ----------------------
926
927 1. The table for translating pcre_compile() error codes into POSIX error codes
928 was out-of-date, and there was no check on the pcre_compile() error code
929 being within the table. This could lead to an OK return being given in
930 error.
931
932 2. Changed the call to open a subject file in pcregrep from fopen(pathname,
933 "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
934 in a Windows environment.
935
936 3. The pcregrep --count option prints the count for each file even when it is
937 zero, as does GNU grep. However, pcregrep was also printing all files when
938 --files-with-matches was added. Now, when both options are given, it prints
939 counts only for those files that have at least one match. (GNU grep just
940 prints the file name in this circumstance, but including the count seems
941 more useful - otherwise, why use --count?) Also ensured that the
942 combination -clh just lists non-zero counts, with no names.
943
944 4. The long form of the pcregrep -F option was incorrectly implemented as
945 --fixed_strings instead of --fixed-strings. This is an incompatible change,
946 but it seems right to fix it, and I didn't think it was worth preserving
947 the old behaviour.
948
949 5. The command line items --regex=pattern and --regexp=pattern were not
950 recognized by pcregrep, which required --regex pattern or --regexp pattern
951 (with a space rather than an '='). The man page documented the '=' forms,
952 which are compatible with GNU grep; these now work.
953
954 6. No libpcreposix.pc file was created for pkg-config; there was just
955 libpcre.pc and libpcrecpp.pc. The omission has been rectified.
956
957 7. Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
958 when UCP support is not needed, by modifying the Python script that
959 generates it from Unicode data files. This should not matter if the module
960 is correctly used as a library, but I received one complaint about 50K of
961 unwanted data. My guess is that the person linked everything into his
962 program rather than using a library. Anyway, it does no harm.
963
964 8. A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
965 was a minimum greater than 1 for a wide character in a possessive
966 repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
967 which had an unlimited repeat of a nested, fixed maximum repeat of a wide
968 character. Chaos in the form of incorrect output or a compiling loop could
969 result.
970
971 9. The restrictions on what a pattern can contain when partial matching is
972 requested for pcre_exec() have been removed. All patterns can now be
973 partially matched by this function. In addition, if there are at least two
974 slots in the offset vector, the offset of the earliest inspected character
975 for the match and the offset of the end of the subject are set in them when
976 PCRE_ERROR_PARTIAL is returned.
977
978 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
979 synonymous with PCRE_PARTIAL, for backwards compatibility, and
980 PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
981 and may be more useful for multi-segment matching.
982
983 11. Partial matching with pcre_exec() is now more intuitive. A partial match
984 used to be given if ever the end of the subject was reached; now it is
985 given only if matching could not proceed because another character was
986 needed. This makes a difference in some odd cases such as Z(*FAIL) with the
987 string "Z", which now yields "no match" instead of "partial match". In the
988 case of pcre_dfa_exec(), "no match" is given if every matching path for the
989 final character ended with (*FAIL).
990
991 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
992 if the pattern had a "must contain" character that was already found in the
993 earlier partial match, unless partial matching was again requested. For
994 example, with the pattern /dog.(body)?/, the "must contain" character is
995 "g". If the first part-match was for the string "dog", restarting with
996 "sbody" failed. This bug has been fixed.
997
998 13. The string returned by pcre_dfa_exec() after a partial match has been
999 changed so that it starts at the first inspected character rather than the
1000 first character of the match. This makes a difference only if the pattern
1001 starts with a lookbehind assertion or \b or \B (\K is not supported by
1002 pcre_dfa_exec()). It's an incompatible change, but it makes the two
1003 matching functions compatible, and I think it's the right thing to do.
1004
1005 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
1006 so that the demonstration program is easily available in environments where
1007 PCRE has not been installed from source.
1008
1009 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
1010 libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
1011 library.
1012
1013 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
1014 It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
1015 is not the first non-POSIX option to be added. Clearly some people find
1016 these options useful.
1017
1018 17. If a caller to the POSIX matching function regexec() passes a non-zero
1019 value for nmatch with a NULL value for pmatch, the value of
1020 nmatch is forced to zero.
1021
1022 18. RunGrepTest did not have a test for the availability of the -u option of
1023 the diff command, as RunTest does. It now checks in the same way as
1024 RunTest, and also checks for the -b option.
1025
1026 19. If an odd number of negated classes containing just a single character
1027 interposed, within parentheses, between a forward reference to a named
1028 subpattern and the definition of the subpattern, compilation crashed with
1029 an internal error, complaining that it could not find the referenced
1030 subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
1031 [The bug was that it was starting one character too far in when skipping
1032 over the character class, thus treating the ] as data rather than
1033 terminating the class. This meant it could skip too much.]
1034
1035 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
1036 /g option in pcretest when the pattern contains \K, which makes it possible
1037 to have an empty string match not at the start, even when the pattern is
1038 anchored. Updated pcretest and pcredemo to use this option.
1039
1040 21. If the maximum number of capturing subpatterns in a recursion was greater
1041 than the maximum at the outer level, the higher number was returned, but
1042 with unset values at the outer level. The correct (outer level) value is
1043 now given.
1044
1045 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
1046 PCRE did not set those parentheses (unlike Perl). I have now found a way to
1047 make it do so. The string so far is captured, making this feature
1048 compatible with Perl.
1049
1050 23. The tests have been re-organized, adding tests 11 and 12, to make it
1051 possible to check the Perl 5.10 features against Perl 5.10.
1052
1053 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
1054 pattern matches a fixed length string. PCRE did not allow this; now it
1055 does. Neither allows recursion.
1056
1057 25. I finally figured out how to implement a request to provide the minimum
1058 length of subject string that was needed in order to match a given pattern.
1059 (It was back references and recursion that I had previously got hung up
1060 on.) This code has now been added to pcre_study(); it finds a lower bound
1061 to the length of subject needed. It is not necessarily the greatest lower
1062 bound, but using it to avoid searching strings that are too short does give
1063 some useful speed-ups. The value is available to calling programs via
1064 pcre_fullinfo().
1065
1066 26. While implementing 25, I discovered to my embarrassment that pcretest had
1067 not been passing the result of pcre_study() to pcre_dfa_exec(), so the
1068 study optimizations had never been tested with that matching function.
1069 Oops. What is worse, even when it was passed study data, there was a bug in
1070 pcre_dfa_exec() that meant it never actually used it. Double oops. There
1071 were also very few tests of studied patterns with pcre_dfa_exec().
1072
1073 27. If (?| is used to create subpatterns with duplicate numbers, they are now
1074 allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
1075 on the other side of the coin, they are no longer allowed to have different
1076 names, because these cannot be distinguished in PCRE, and this has caused
1077 confusion. (This is a difference from Perl.)
1078
1079 28. When duplicate subpattern names are present (necessarily with different
1080 numbers, as required by 27 above), and a test is made by name in a
1081 conditional pattern, either for a subpattern having been matched, or for
1082 recursion in such a pattern, all the associated numbered subpatterns are
1083 tested, and the overall condition is true if the condition is true for any
1084 one of them. This is the way Perl works, and is also more like the way
1085 testing by number works.
1086
1087
1088 Version 7.9 11-Apr-09
1089 ---------------------
1090
1091 1. When building with support for bzlib/zlib (pcregrep) and/or readline
1092 (pcretest), all targets were linked against these libraries. This included
1093 libpcre, libpcreposix, and libpcrecpp, even though they do not use these
1094 libraries. This caused unwanted dependencies to be created. This problem
1095 has been fixed, and now only pcregrep is linked with bzlib/zlib and only
1096 pcretest is linked with readline.
1097
1098 2. The "typedef int BOOL" in pcre_internal.h that was included inside the
1099 "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
1100 moved outside it again, because FALSE and TRUE are already defined in AIX,
1101 but BOOL is not.
1102
1103 3. The pcre_config() function was treating the PCRE_MATCH_LIMIT and
1104 PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
1105
1106 4. The pcregrep documentation said spaces were inserted as well as colons (or
1107 hyphens) following file names and line numbers when outputting matching
1108 lines. This is not true; no spaces are inserted. I have also clarified the
1109 wording for the --colour (or --color) option.
1110
1111 5. In pcregrep, when --colour was used with -o, the list of matching strings
1112 was not coloured; this is different to GNU grep, so I have changed it to be
1113 the same.
1114
1115 6. When --colo(u)r was used in pcregrep, only the first matching substring in
1116 each matching line was coloured. Now it goes on to look for further matches
1117 of any of the test patterns, which is the same behaviour as GNU grep.
1118
1119 7. A pattern that could match an empty string could cause pcregrep to loop; it
1120 doesn't make sense to accept an empty string match in pcregrep, so I have
1121 locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
1122 seems to be how GNU grep behaves.
1123
1124 8. The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
1125 start or after a newline", because the conditional assertion was not being
1126 correctly handled. The rule now is that both the assertion and what follows
1127 in the first alternative must satisfy the test.
1128
1129 9. If auto-callout was enabled in a pattern with a conditional group whose
1130 condition was an assertion, PCRE could crash during matching, both with
1131 pcre_exec() and pcre_dfa_exec().
1132
1133 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
1134 used for matching.
1135
1136 11. Unicode property support in character classes was not working for
1137 characters (bytes) greater than 127 when not in UTF-8 mode.
1138
1139 12. Added the -M command line option to pcretest.
1140
1141 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
1142
1143 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
1144
1145 16. Added comments and documentation about mis-use of no_arg in the C++
1146 wrapper.
1147
1148 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
1149 from Martin Jerabek that uses macro names for all relevant character and
1150 string constants.
1151
1152 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
1153 SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
1154 SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
1155 these, but not everybody uses configure.
1156
1157 19. A conditional group that had only one branch was not being correctly
1158 recognized as an item that could match an empty string. This meant that an
1159 enclosing group might also not be so recognized, causing infinite looping
1160 (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
1161 with the subject "ab", where knowledge that the repeated group can match
1162 nothing is needed in order to break the loop.
1163
1164 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
1165 exec(), but without supplying a callout function, matching went wrong.
1166
1167 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
1168 leak if the size of the offset vector was greater than 30. When the vector
1169 is smaller, the saved offsets during recursion go onto a local stack
1170 vector, but for larger vectors malloc() is used. It was failing to free
1171 when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
1172 error, in fact).
1173
1174 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
1175 heapframe that is used only when UTF-8 support is enabled. This caused no
1176 problem, but was untidy.
1177
1178 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
1179 CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
1180 included within another project.
1181
1182 24. Steven Van Ingelgem's patches to add more options to the CMake support,
1183 slightly modified by me:
1184
1185 (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
1186 not building pcregrep.
1187
1188 (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
1189 if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
1190
1191 25. Forward references, both numeric and by name, in patterns that made use of
1192 duplicate group numbers, could behave incorrectly or give incorrect errors,
1193 because when scanning forward to find the reference group, PCRE was not
1194 taking into account the duplicate group numbers. A pattern such as
1195 ^X(?3)(a)(?|(b)|(q))(Y) is an example.
1196
1197 26. Changed a few more instances of "const unsigned char *" to USPTR, making
1198 the feature of a custom pointer more persuasive (as requested by a user).
1199
1200 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
1201 pcretest.c, inside #ifndefs, because it seems they are sometimes already
1202 pre-defined.
1203
1204 28. Added support for (*UTF8) at the start of a pattern.
1205
1206 29. Arrange for flags added by the "release type" setting in CMake to be shown
1207 in the configuration summary.
1208
1209
1210 Version 7.8 05-Sep-08
1211 ---------------------
1212
1213 1. Replaced UCP searching code with optimized version as implemented for Ad
1214 Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
1215 stage table and inline lookup instead of a function, giving speed ups of 2
1216 to 5 times on some simple patterns that I tested. Permission was given to
1217 distribute the MultiStage2.py script that generates the tables (it's not in
1218 the tarball, but is in the Subversion repository).
1219
1220 2. Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
1221 scripts.
1222
1223 3. Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
1224 a group with a zero qualifier. The result of the study could be incorrect,
1225 or the function might crash, depending on the pattern.
1226
1227 4. Caseless matching was not working for non-ASCII characters in back
1228 references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
1229 It now works when Unicode Property Support is available.
1230
1231 5. In pcretest, an escape such as \x{de} in the data was always generating
1232 a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
1233 non-UTF-8 mode. If the value is greater than 255, it gives a warning about
1234 truncation.
1235
1236 6. Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
1237
1238 7. Added two (int) casts to pcregrep when printing the difference of two
1239 pointers, in case they are 64-bit values.
1240
1241 8. Added comments about Mac OS X stack usage to the pcrestack man page and to
1242 test 2 if it fails.
1243
1244 9. Added PCRE_CALL_CONVENTION just before the names of all exported functions,
1245 and a #define of that name to empty if it is not externally set. This is to
1246 allow users of MSVC to set it if necessary.
1247
1248 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
1249 the convenience functions in the pcre_get.c source file.
1250
1251 11. An option change at the start of a pattern that had top-level alternatives
1252 could cause overwriting and/or a crash. This command provoked a crash in
1253 some environments:
1254
1255 printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
1256
1257 This potential security problem was recorded as CVE-2008-2371.
1258
1259 12. For a pattern where the match had to start at the beginning or immediately
1260 after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
1261 pcre_dfa_exec() could read past the end of the passed subject if there was
1262 no match. To help with detecting such bugs (e.g. with valgrind), I modified
1263 pcretest so that it places the subject at the end of its malloc-ed buffer.
1264
1265 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
1266 exec() might read past the end of the data buffer in UTF-8 mode.
1267
1268 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
1269 the data contained the byte 0x85 as part of a UTF-8 character within its
1270 first line. This applied both to normal and DFA matching.
1271
1272 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
1273 /^[^d]*?$/8 failed to match "abc".
1274
1275 16. Added a missing copyright notice to pcrecpp_internal.h.
1276
1277 17. Make it more clear in the documentation that values returned from
1278 pcre_exec() in ovector are byte offsets, not character counts.
1279
1280 18. Tidied a few places to stop certain compilers from issuing warnings.
1281
1282 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
1283 supplied by Stefan Weber. I made a further small update for 7.8 because
1284 there is a change of source arrangements: the pcre_searchfuncs.c module is
1285 replaced by pcre_ucd.c.
1286
1287
1288 Version 7.7 07-May-08
1289 ---------------------
1290
1291 1. Applied Craig's patch to sort out a long long problem: "If we can't convert
1292 a string to a long long, pretend we don't even have a long long." This is
1293 done by checking for the strtoq, strtoll, and _strtoi64 functions.
1294
1295 2. Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
1296 pre-7.6 versions, which defined a global no_arg variable instead of putting
1297 it in the RE class. (See also #8 below.)
1298
1299 3. Remove a line of dead code, identified by coverity and reported by Nuno
1300 Lopes.
1301
1302 4. Fixed two related pcregrep bugs involving -r with --include or --exclude:
1303
1304 (1) The include/exclude patterns were being applied to the whole pathnames
1305 of files, instead of just to the final components.
1306
1307 (2) If there was more than one level of directory, the subdirectories were
1308 skipped unless they satisfied the include/exclude conditions. This is
1309 inconsistent with GNU grep (and could even be seen as contrary to the
1310 pcregrep specification - which I improved to make it absolutely clear).
1311 The action now is always to scan all levels of directory, and just
1312 apply the include/exclude patterns to regular files.
1313
1314 5. Added the --include_dir and --exclude_dir patterns to pcregrep, and used
1315 --exclude_dir in the tests to avoid scanning .svn directories.
1316
1317 6. Applied Craig's patch to the QuoteMeta function so that it escapes the
1318 NUL character as backslash + 0 rather than backslash + NUL, because PCRE
1319 doesn't support NULs in patterns.
1320
1321 7. Added some missing "const"s to declarations of static tables in
1322 pcre_compile.c and pcre_dfa_exec.c.
1323
1324 8. Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
1325 caused by fix #2 above. (Subsequently also a second patch to fix the
1326 first patch. And a third patch - this was a messy problem.)
1327
1328 9. Applied Craig's patch to remove the use of push_back().
1329
1330 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
1331 matching function regexec().
1332
1333 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
1334 which, however, unlike Perl's \g{...}, are subroutine calls, not back
1335 references. PCRE supports relative numbers with this syntax (I don't think
1336 Oniguruma does).
1337
1338 12. Previously, a group with a zero repeat such as (...){0} was completely
1339 omitted from the compiled regex. However, this means that if the group
1340 was called as a subroutine from elsewhere in the pattern, things went wrong
1341 (an internal error was given). Such groups are now left in the compiled
1342 pattern, with a new opcode that causes them to be skipped at execution
1343 time.
1344
1345 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
1346 to the way PCRE behaves:
1347
1348 (a) A lone ] character is dis-allowed (Perl treats it as data).
1349
1350 (b) A back reference to an unmatched subpattern matches an empty string
1351 (Perl fails the current match path).
1352
1353 (c) A data ] in a character class must be notated as \] because if the
1354 first data character in a class is ], it defines an empty class. (In
1355 Perl it is not possible to have an empty class.) The empty class []
1356 never matches; it forces failure and is equivalent to (*FAIL) or (?!).
1357 The negative empty class [^] matches any one character, independently
1358 of the DOTALL setting.
1359
1360 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
1361 non-existent subpattern following a character class starting with ']' and
1362 containing () gave an internal compiling error instead of "reference to
1363 non-existent subpattern". Fortunately, when the pattern did exist, the
1364 compiled code was correct. (When scanning forwards to check for the
1365 existencd of the subpattern, it was treating the data ']' as terminating
1366 the class, so got the count wrong. When actually compiling, the reference
1367 was subsequently set up correctly.)
1368
1369 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
1370 it was being rejected as not supported by pcre_dfa_exec(), even though
1371 other assertions are supported. I have made pcre_dfa_exec() support
1372 (*FAIL).
1373
1374 16. The implementation of 13c above involved the invention of a new opcode,
1375 OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
1376 cannot be changed at match time, I realized I could make a small
1377 improvement to matching performance by compiling OP_ALLANY instead of
1378 OP_ANY for "." when DOTALL was set, and then removing the runtime tests
1379 on the OP_ANY path.
1380
1381 17. Compiling pcretest on Windows with readline support failed without the
1382 following two fixes: (1) Make the unistd.h include conditional on
1383 HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
1384
1385 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
1386 ncurses library to be included for pcretest when ReadLine support is
1387 requested, but also to allow for it to be overridden. This patch came from
1388 Daniel Bergström.
1389
1390 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
1391 as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
1392 any errors with the current Unicode tables. Thanks to Peter Kankowski for
1393 spotting this.
1394
1395
1396 Version 7.6 28-Jan-08
1397 ---------------------
1398
1399 1. A character class containing a very large number of characters with
1400 codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
1401 overflow.
1402
1403 2. Patch to cut out the "long long" test in pcrecpp_unittest when
1404 HAVE_LONG_LONG is not defined.
1405
1406 3. Applied Christian Ehrlicher's patch to update the CMake build files to
1407 bring them up to date and include new features. This patch includes:
1408
1409 - Fixed PH's badly added libz and libbz2 support.
1410 - Fixed a problem with static linking.
1411 - Added pcredemo. [But later removed - see 7 below.]
1412 - Fixed dftables problem and added an option.
1413 - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
1414 HAVE_LONG_LONG.
1415 - Added readline support for pcretest.
1416 - Added an listing of the option settings after cmake has run.
1417
1418 4. A user submitted a patch to Makefile that makes it easy to create
1419 "pcre.dll" under mingw when using Configure/Make. I added stuff to
1420 Makefile.am that cause it to include this special target, without
1421 affecting anything else. Note that the same mingw target plus all
1422 the other distribution libraries and programs are now supported
1423 when configuring with CMake (see 6 below) instead of with
1424 Configure/Make.
1425
1426 5. Applied Craig's patch that moves no_arg into the RE class in the C++ code.
1427 This is an attempt to solve the reported problem "pcrecpp::no_arg is not
1428 exported in the Windows port". It has not yet been confirmed that the patch
1429 solves the problem, but it does no harm.
1430
1431 6. Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
1432 NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
1433 with CMake, and also correct the comment about stack recursion.
1434
1435 7. Remove the automatic building of pcredemo from the ./configure system and
1436 from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
1437 of a program that users should build themselves after PCRE is installed, so
1438 building it automatically is not really right. What is more, it gave
1439 trouble in some build environments.
1440
1441 8. Further tidies to CMakeLists.txt from Sheri and Christian.
1442
1443
1444 Version 7.5 10-Jan-08
1445 ---------------------
1446
1447 1. Applied a patch from Craig: "This patch makes it possible to 'ignore'
1448 values in parens when parsing an RE using the C++ wrapper."
1449
1450 2. Negative specials like \S did not work in character classes in UTF-8 mode.
1451 Characters greater than 255 were excluded from the class instead of being
1452 included.
1453
1454 3. The same bug as (2) above applied to negated POSIX classes such as
1455 [:^space:].
1456
1457 4. PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
1458 defined or documented. It seems to have been a typo for PCRE_STATIC, so
1459 I have changed it.
1460
1461 5. The construct (?&) was not diagnosed as a syntax error (it referenced the
1462 first named subpattern) and a construct such as (?&a) would reference the
1463 first named subpattern whose name started with "a" (in other words, the
1464 length check was missing). Both these problems are fixed. "Subpattern name
1465 expected" is now given for (?&) (a zero-length name), and this patch also
1466 makes it give the same error for \k'' (previously it complained that that
1467 was a reference to a non-existent subpattern).
1468
1469 6. The erroneous patterns (?+-a) and (?-+a) give different error messages;
1470 this is right because (?- can be followed by option settings as well as by
1471 digits. I have, however, made the messages clearer.
1472
1473 7. Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
1474 than the number used in the conditional) now cause a compile-time error.
1475 This is actually not compatible with Perl, which accepts such patterns, but
1476 treats the conditional as always being FALSE (as PCRE used to), but it
1477 seems to me that giving a diagnostic is better.
1478
1479 8. Change "alphameric" to the more common word "alphanumeric" in comments
1480 and messages.
1481
1482 9. Fix two occurrences of "backslash" in comments that should have been
1483 "backspace".
1484
1485 10. Remove two redundant lines of code that can never be obeyed (their function
1486 was moved elsewhere).
1487
1488 11. The program that makes PCRE's Unicode character property table had a bug
1489 which caused it to generate incorrect table entries for sequences of
1490 characters that have the same character type, but are in different scripts.
1491 It amalgamated them into a single range, with the script of the first of
1492 them. In other words, some characters were in the wrong script. There were
1493 thirteen such cases, affecting characters in the following ranges:
1494
1495 U+002b0 - U+002c1
1496 U+0060c - U+0060d
1497 U+0061e - U+00612
1498 U+0064b - U+0065e
1499 U+0074d - U+0076d
1500 U+01800 - U+01805
1501 U+01d00 - U+01d77
1502 U+01d9b - U+01dbf
1503 U+0200b - U+0200f
1504 U+030fc - U+030fe
1505 U+03260 - U+0327f
1506 U+0fb46 - U+0fbb1
1507 U+10450 - U+1049d
1508
1509 12. The -o option (show only the matching part of a line) for pcregrep was not
1510 compatible with GNU grep in that, if there was more than one match in a
1511 line, it showed only the first of them. It now behaves in the same way as
1512 GNU grep.
1513
1514 13. If the -o and -v options were combined for pcregrep, it printed a blank
1515 line for every non-matching line. GNU grep prints nothing, and pcregrep now
1516 does the same. The return code can be used to tell if there were any
1517 non-matching lines.
1518
1519 14. Added --file-offsets and --line-offsets to pcregrep.
1520
1521 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
1522 infinitely looping recursion. The bug was that positive lookaheads were not
1523 being skipped when checking for a possible empty match (negative lookaheads
1524 and both kinds of lookbehind were skipped).
1525
1526 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
1527 inclusion of <windows.h> to before rather than after the definition of
1528 INVALID_FILE_ATTRIBUTES (patch from David Byron).
1529
1530 17. Specifying a possessive quantifier with a specific limit for a Unicode
1531 character property caused pcre_compile() to compile bad code, which led at
1532 runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
1533 are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
1534 caused the error; without that there was no problem.
1535
1536 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
1537
1538 19. Added --enable-pcretest-libreadline.
1539
1540 20. In pcrecpp.cc, the variable 'count' was incremented twice in
1541 RE::GlobalReplace(). As a result, the number of replacements returned was
1542 double what it should be. I removed one of the increments, but Craig sent a
1543 later patch that removed the other one (the right fix) and added unit tests
1544 that check the return values (which was not done before).
1545
1546 21. Several CMake things:
1547
1548 (1) Arranged that, when cmake is used on Unix, the libraries end up with
1549 the names libpcre and libpcreposix, not just pcre and pcreposix.
1550
1551 (2) The above change means that pcretest and pcregrep are now correctly
1552 linked with the newly-built libraries, not previously installed ones.
1553
1554 (3) Added PCRE_SUPPORT_LIBREADLINE, PCRE_SUPPORT_LIBZ, PCRE_SUPPORT_LIBBZ2.
1555
1556 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
1557 crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
1558 UTF-8 newline character). The key issue is that the pattern starts .*;
1559 this means that the match must be either at the beginning, or after a
1560 newline. The bug was in the code for advancing after a failed match and
1561 checking that the new position followed a newline. It was not taking
1562 account of UTF-8 characters correctly.
1563
1564 23. PCRE was behaving differently from Perl in the way it recognized POSIX
1565 character classes. PCRE was not treating the sequence [:...:] as a
1566 character class unless the ... were all letters. Perl, however, seems to
1567 allow any characters between [: and :], though of course it rejects as
1568 unknown any "names" that contain non-letters, because all the known class
1569 names consist only of letters. Thus, Perl gives an error for [[:1234:]],
1570 for example, whereas PCRE did not - it did not recognize a POSIX character
1571 class. This seemed a bit dangerous, so the code has been changed to be
1572 closer to Perl. The behaviour is not identical to Perl, because PCRE will
1573 diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
1574 treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
1575 Perl does, and where it didn't before.
1576
1577 24. Rewrite so as to remove the single use of %n from pcregrep because in some
1578 Windows environments %n is disabled by default.
1579
1580
1581 Version 7.4 21-Sep-07
1582 ---------------------
1583
1584 1. Change 7.3/28 was implemented for classes by looking at the bitmap. This
1585 means that a class such as [\s] counted as "explicit reference to CR or
1586 LF". That isn't really right - the whole point of the change was to try to
1587 help when there was an actual mention of one of the two characters. So now
1588 the change happens only if \r or \n (or a literal CR or LF) character is
1589 encountered.
1590
1591 2. The 32-bit options word was also used for 6 internal flags, but the numbers
1592 of both had grown to the point where there were only 3 bits left.
1593 Fortunately, there was spare space in the data structure, and so I have
1594 moved the internal flags into a new 16-bit field to free up more option
1595 bits.
1596
1597 3. The appearance of (?J) at the start of a pattern set the DUPNAMES option,
1598 but did not set the internal JCHANGED flag - either of these is enough to
1599 control the way the "get" function works - but the PCRE_INFO_JCHANGED
1600 facility is supposed to tell if (?J) was ever used, so now (?J) at the
1601 start sets both bits.
1602
1603 4. Added options (at build time, compile time, exec time) to change \R from
1604 matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
1605
1606 5. doc/pcresyntax.html was missing from the distribution.
1607
1608 6. Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
1609 compatibility, even though it is no longer used.
1610
1611 7. Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
1612 strtoull to pcrecpp.cc to select the available functions in WIN32 when the
1613 windows.h file is present (where different names are used). [This was
1614 reversed later after testing - see 16 below.]
1615
1616 8. Changed all #include <config.h> to #include "config.h". There were also
1617 some further <pcre.h> cases that I changed to "pcre.h".
1618
1619 9. When pcregrep was used with the --colour option, it missed the line ending
1620 sequence off the lines that it output.
1621
1622 10. It was pointed out to me that arrays of string pointers cause lots of
1623 relocations when a shared library is dynamically loaded. A technique of
1624 using a single long string with a table of offsets can drastically reduce
1625 these. I have refactored PCRE in four places to do this. The result is
1626 dramatic:
1627
1628 Originally: 290
1629 After changing UCP table: 187
1630 After changing error message table: 43
1631 After changing table of "verbs" 36
1632 After changing table of Posix names 22
1633
1634 Thanks to the folks working on Gregex for glib for this insight.
1635
1636 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
1637 unicode-properties was also set.
1638
1639 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
1640
1641 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
1642 checked only for CRLF.
1643
1644 14. Added casts to pcretest.c to avoid compiler warnings.
1645
1646 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
1647
1648 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
1649 and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
1650 entirely. This removes changes made in 7 above.
1651
1652 17. The CMake files have been updated, and there is now more information about
1653 building with CMake in the NON-UNIX-USE document.
1654
1655
1656 Version 7.3 28-Aug-07
1657 ---------------------
1658
1659 1. In the rejigging of the build system that eventually resulted in 7.1, the
1660 line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
1661 brackets there is not right, since it causes compilers to look for an
1662 installed pcre.h, not the version that is in the source that is being
1663 compiled (which of course may be different). I have changed it back to:
1664
1665 #include "pcre.h"
1666
1667 I have a vague recollection that the change was concerned with compiling in
1668 different directories, but in the new build system, that is taken care of
1669 by the VPATH setting the Makefile.
1670
1671 2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
1672 when the subject happened to end in the byte 0x85 (e.g. if the last
1673 character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
1674 characters but of course it shouldn't be taken as a newline when it is part
1675 of another character. The bug was that, for an unlimited repeat of . in
1676 not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
1677 characters when looking for a newline.
1678
1679 3. A small performance improvement in the DOTALL UTF-8 mode .* case.
1680
1681 4. Debugging: adjusted the names of opcodes for different kinds of parentheses
1682 in debug output.
1683
1684 5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
1685 long printing in the pcrecpp unittest when running under MinGW.
1686
1687 6. ESC_K was left out of the EBCDIC table.
1688
1689 7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
1690 parentheses; I made it 1000, which seemed large enough. Unfortunately, the
1691 limit also applies to "virtual nesting" when a pattern is recursive, and in
1692 this case 1000 isn't so big. I have been able to remove this limit at the
1693 expense of backing off one optimization in certain circumstances. Normally,
1694 when pcre_exec() would call its internal match() function recursively and
1695 immediately return the result unconditionally, it uses a "tail recursion"
1696 feature to save stack. However, when a subpattern that can match an empty
1697 string has an unlimited repetition quantifier, it no longer makes this
1698 optimization. That gives it a stack frame in which to save the data for
1699 checking that an empty string has been matched. Previously this was taken
1700 from the 1000-entry workspace that had been reserved. So now there is no
1701 explicit limit, but more stack is used.
1702
1703 8. Applied Daniel's patches to solve problems with the import/export magic
1704 syntax that is required for Windows, and which was going wrong for the
1705 pcreposix and pcrecpp parts of the library. These were overlooked when this
1706 problem was solved for the main library.
1707
1708 9. There were some crude static tests to avoid integer overflow when computing
1709 the size of patterns that contain repeated groups with explicit upper
1710 limits. As the maximum quantifier is 65535, the maximum group length was
1711 set at 30,000 so that the product of these two numbers did not overflow a
1712 32-bit integer. However, it turns out that people want to use groups that
1713 are longer than 30,000 bytes (though not repeat them that many times).
1714 Change 7.0/17 (the refactoring of the way the pattern size is computed) has
1715 made it possible to implement the integer overflow checks in a much more
1716 dynamic way, which I have now done. The artificial limitation on group
1717 length has been removed - we now have only the limit on the total length of
1718 the compiled pattern, which depends on the LINK_SIZE setting.
1719
1720 10. Fixed a bug in the documentation for get/copy named substring when
1721 duplicate names are permitted. If none of the named substrings are set, the
1722 functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
1723 empty string.
1724
1725 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
1726 instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
1727 because the ] is interpreted as the first data character and the
1728 terminating ] is not found. PCRE has been made compatible with Perl in this
1729 regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
1730 cause memory overwriting.
1731
1732 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
1733 string has been matched (to stop an infinite loop). It was not recognizing
1734 a conditional subpattern that could match an empty string if that
1735 subpattern was within another subpattern. For example, it looped when
1736 trying to match (((?(1)X|))*) but it was OK with ((?(1)X|)*) where the
1737 condition was not nested. This bug has been fixed.
1738
1739 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
1740 past the start of the subject in the presence of bytes with the top bit
1741 set, for example "\x8aBCD".
1742
1743 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
1744 (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
1745
1746 14. Optimized (?!) to (*FAIL).
1747
1748 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
1749 This restricts code points to be within the range 0 to 0x10FFFF, excluding
1750 the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
1751 full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
1752 does: it's just the validity check that is more restrictive.
1753
1754 16. Inserted checks for integer overflows during escape sequence (backslash)
1755 processing, and also fixed erroneous offset values for syntax errors during
1756 backslash processing.
1757
1758 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
1759 for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
1760
1761 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
1762 caused an overrun.
1763
1764 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
1765 something other than just ASCII characters) inside a group that had an
1766 unlimited repeat caused a loop at compile time (while checking to see
1767 whether the group could match an empty string).
1768
1769 20. Debugging a pattern containing \p or \P could cause a crash. For example,
1770 [\P{Any}] did so. (Error in the code for printing property names.)
1771
1772 21. An orphan \E inside a character class could cause a crash.
1773
1774 22. A repeated capturing bracket such as (A)? could cause a wild memory
1775 reference during compilation.
1776
1777 23. There are several functions in pcre_compile() that scan along a compiled
1778 expression for various reasons (e.g. to see if it's fixed length for look
1779 behind). There were bugs in these functions when a repeated \p or \P was
1780 present in the pattern. These operators have additional parameters compared
1781 with \d, etc, and these were not being taken into account when moving along
1782 the compiled data. Specifically:
1783
1784 (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
1785 length.
1786
1787 (b) An item such as \pL+ within a repeated group could cause crashes or
1788 loops.
1789
1790 (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
1791 "reference to non-existent subpattern" error.
1792
1793 (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
1794
1795 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
1796 characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
1797
1798 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
1799
1800 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
1801 character were causing crashes (broken optimization).
1802
1803 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
1804 \p or \P) caused a compile-time loop.
1805
1806 28. More problems have arisen in unanchored patterns when CRLF is a valid line
1807 break. For example, the unstudied pattern [\r\n]A does not match the string
1808 "\r\nA" because change 7.0/46 below moves the current point on by two
1809 characters after failing to match at the start. However, the pattern \nA
1810 *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
1811 the same is true. There doesn't seem any very clean way out of this, but
1812 what I have chosen to do makes the common cases work: PCRE now takes note
1813 of whether there can be an explicit match for \r or \n anywhere in the
1814 pattern, and if so, 7.0/46 no longer applies. As part of this change,
1815 there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
1816 pattern has explicit CR or LF references.
1817
1818 29. Added (*CR) etc for changing newline setting at start of pattern.
1819
1820
1821 Version 7.2 19-Jun-07
1822 ---------------------
1823
1824 1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
1825 which is apparently normally available under Windows.
1826
1827 2. Re-jig the pcregrep tests with different newline settings in an attempt
1828 to make them independent of the local environment's newline setting.
1829
1830 3. Add code to configure.ac to remove -g from the CFLAGS default settings.
1831
1832 4. Some of the "internals" tests were previously cut out when the link size
1833 was not 2, because the output contained actual offsets. The recent new
1834 "Z" feature of pcretest means that these can be cut out, making the tests
1835 usable with all link sizes.
1836
1837 5. Implemented Stan Switzer's goto replacement for longjmp() when not using
1838 stack recursion. This gives a massive performance boost under BSD, but just
1839 a small improvement under Linux. However, it saves one field in the frame
1840 in all cases.
1841
1842 6. Added more features from the forthcoming Perl 5.10:
1843
1844 (a) (?-n) (where n is a string of digits) is a relative subroutine or
1845 recursion call. It refers to the nth most recently opened parentheses.
1846
1847 (b) (?+n) is also a relative subroutine call; it refers to the nth next
1848 to be opened parentheses.
1849
1850 (c) Conditions that refer to capturing parentheses can be specified
1851 relatively, for example, (?(-2)... or (?(+3)...
1852
1853 (d) \K resets the start of the current match so that everything before
1854 is not part of it.
1855
1856 (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
1857
1858 (f) \g{name} is another synonym - part of Perl 5.10's unification of
1859 reference syntax.
1860
1861 (g) (?| introduces a group in which the numbering of parentheses in each
1862 alternative starts with the same number.
1863
1864 (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
1865
1866 7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
1867 PCRE_INFO_JCHANGED.
1868
1869 8. A pattern such as (.*(.)?)* caused pcre_exec() to fail by either not
1870 terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
1871 for detecting groups that can match an empty string.
1872
1873 9. A pattern with a very large number of alternatives (more than several
1874 hundred) was running out of internal workspace during the pre-compile
1875 phase, where pcre_compile() figures out how much memory will be needed. A
1876 bit of new cunning has reduced the workspace needed for groups with
1877 alternatives. The 1000-alternative test pattern now uses 12 bytes of
1878 workspace instead of running out of the 4096 that are available.
1879
1880 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
1881
1882 11. Applied patch from Google to remove an optimization that didn't quite work.
1883 The report of the bug said:
1884
1885 pcrecpp::RE("a*").FullMatch("aaa") matches, while
1886 pcrecpp::RE("a*?").FullMatch("aaa") does not, and
1887 pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
1888
1889 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
1890 it matched the wrong number of bytes.
1891
1892
1893 Version 7.1 24-Apr-07
1894 ---------------------
1895
1896 1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
1897 that is more "standard", making use of automake and other Autotools. There
1898 is some re-arrangement of the files and adjustment of comments consequent
1899 on this.
1900
1901 2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
1902 for recursive directory scanning broke on some systems because the files
1903 are not scanned in any specific order and on different systems the order
1904 was different. A call to "sort" has been inserted into RunGrepTest for the
1905 approprate test as a short-term fix. In the longer term there may be an
1906 alternative.
1907
1908 3. I had an email from Eric Raymond about problems translating some of PCRE's
1909 man pages to HTML (despite the fact that I distribute HTML pages, some
1910 people do their own conversions for various reasons). The problems
1911 concerned the use of low-level troff macros .br and .in. I have therefore
1912 removed all such uses from the man pages (some were redundant, some could
1913 be replaced by .nf/.fi pairs). The 132html script that I use to generate
1914 HTML has been updated to handle .nf/.fi and to complain if it encounters
1915 .br or .in.
1916
1917 4. Updated comments in configure.ac that get placed in config.h.in and also
1918 arranged for config.h to be included in the distribution, with the name
1919 config.h.generic, for the benefit of those who have to compile without
1920 Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
1921
1922 5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
1923 Weber: (1) pcre_internal.h was missing some function renames; (2) updated
1924 makevp.bat for the current PCRE, using the additional files
1925 makevp_c.txt, makevp_l.txt, and pcregexp.pas.
1926
1927 6. A Windows user reported a minor discrepancy with test 2, which turned out
1928 to be caused by a trailing space on an input line that had got lost in his
1929 copy. The trailing space was an accident, so I've just removed it.
1930
1931 7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
1932 that is needed.
1933
1934 8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
1935 as "const" (a) because they are and (b) because it helps the PHP
1936 maintainers who have recently made a script to detect big data structures
1937 in the php code that should be moved to the .rodata section. I remembered
1938 to update Builducptable as well, so it won't revert if ucptable.h is ever
1939 re-created.
1940
1941 9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
1942 pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
1943 order to be able to cut out the UTF-8 tables in the latter when UTF-8
1944 support is not required. This saves 1.5-2K of code, which is important in
1945 some applications.
1946
1947 Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
1948 so as not to refer to the tables, even though these functions will never be
1949 called when UTF-8 support is disabled. Otherwise there are problems with a
1950 shared library.
1951
1952 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
1953
1954 (a) It was defining its arguments as char * instead of void *.
1955
1956 (b) It was assuming that all moves were upwards in memory; this was true
1957 a long time ago when I wrote it, but is no longer the case.
1958
1959 The emulated memove() is provided for those environments that have neither
1960 memmove() nor bcopy(). I didn't think anyone used it these days, but that
1961 is clearly not the case, as these two bugs were recently reported.
1962
1963 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
1964 and Detrail to create the HTML documentation, the .txt form of the man
1965 pages, and it removes trailing spaces from listed files. It also creates
1966 pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
1967 case, it wraps all the #defines with #ifndefs. This script should be run
1968 before "make dist".
1969
1970 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
1971 with Unicode property support.
1972
1973 (a) For a maximizing quantifier, if the two different cases of the
1974 character were of different lengths in their UTF-8 codings (there are
1975 some cases like this - I found 11), and the matching function had to
1976 back up over a mixture of the two cases, it incorrectly assumed they
1977 were both the same length.
1978
1979 (b) When PCRE was configured to use the heap rather than the stack for
1980 recursion during matching, it was not correctly preserving the data for
1981 the other case of a UTF-8 character when checking ahead for a match
1982 while processing a minimizing repeat. If the check also involved
1983 matching a wide character, but failed, corruption could cause an
1984 erroneous result when trying to check for a repeat of the original
1985 character.
1986
1987 13. Some tidying changes to the testing mechanism:
1988
1989 (a) The RunTest script now detects the internal link size and whether there
1990 is UTF-8 and UCP support by running ./pcretest -C instead of relying on
1991 values substituted by "configure". (The RunGrepTest script already did
1992 this for UTF-8.) The configure.ac script no longer substitutes the
1993 relevant variables.
1994
1995 (b) The debugging options /B and /D in pcretest show the compiled bytecode
1996 with length and offset values. This means that the output is different
1997 for different internal link sizes. Test 2 is skipped for link sizes
1998 other than 2 because of this, bypassing the problem. Unfortunately,
1999 there was also a test in test 3 (the locale tests) that used /B and
2000 failed for link sizes other than 2. Rather than cut the whole test out,
2001 I have added a new /Z option to pcretest that replaces the length and
2002 offset values with spaces. This is now used to make test 3 independent
2003 of link size. (Test 2 will be tidied up later.)
2004
2005 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
2006 segmentation fault instead of returning the appropriate error message.
2007
2008 15. In multiline mode when the newline sequence was set to "any", the pattern
2009 ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
2010 This doesn't seem right; it now treats the CRLF combination as the line
2011 ending, and so does not match in that case. It's only a pattern such as ^$
2012 that would hit this one: something like ^ABC$ would have failed after \r
2013 and then tried again after \r\n.
2014
2015 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
2016 in an attempt to make files that differ only in their line terminators
2017 compare equal. This works on Linux.
2018
2019 17. Under certain error circumstances pcregrep might try to free random memory
2020 as it exited. This is now fixed, thanks to valgrind.
2021
2022 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
2023 "abc\r\n\r\n", it found an unwanted second match after the second \r. This
2024 was because its rules for how to advance for /g after matching an empty
2025 string at the end of a line did not allow for this case. They now check for
2026 it specially.
2027
2028 20. pcretest is supposed to handle patterns and data of any length, by
2029 extending its buffers when necessary. It was getting this wrong when the
2030 buffer for a data line had to be extended.
2031
2032 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
2033 CRLF as a newline sequence.
2034
2035 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
2036 out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
2037 I have nevertheless tidied it up.
2038
2039 23. Added some casts to kill warnings from HP-UX ia64 compiler.
2040
2041 24. Added a man page for pcre-config.
2042
2043
2044 Version 7.0 19-Dec-06
2045 ---------------------
2046
2047 1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
2048 moving to gcc 4.1.1.
2049
2050 2. The -S option for pcretest uses setrlimit(); I had omitted to #include
2051 sys/time.h, which is documented as needed for this function. It doesn't
2052 seem to matter on Linux, but it showed up on some releases of OS X.
2053
2054 3. It seems that there are systems where bytes whose values are greater than
2055 127 match isprint() in the "C" locale. The "C" locale should be the
2056 default when a C program starts up. In most systems, only ASCII printing
2057 characters match isprint(). This difference caused the output from pcretest
2058 to vary, making some of the tests fail. I have changed pcretest so that:
2059
2060 (a) When it is outputting text in the compiled version of a pattern, bytes
2061 other than 32-126 are always shown as hex escapes.
2062
2063 (b) When it is outputting text that is a matched part of a subject string,
2064 it does the same, unless a different locale has been set for the match
2065 (using the /L modifier). In this case, it uses isprint() to decide.
2066
2067 4. Fixed a major bug that caused incorrect computation of the amount of memory
2068 required for a compiled pattern when options that changed within the
2069 pattern affected the logic of the preliminary scan that determines the
2070 length. The relevant options are -x, and -i in UTF-8 mode. The result was
2071 that the computed length was too small. The symptoms of this bug were
2072 either the PCRE error "internal error: code overflow" from pcre_compile(),
2073 or a glibc crash with a message such as "pcretest: free(): invalid next
2074 size (fast)". Examples of patterns that provoked this bug (shown in
2075 pcretest format) are:
2076
2077 /(?-x: )/x
2078 /(?x)(?-x: \s*#\s*)/
2079 /((?i)[\x{c0}])/8
2080 /(?i:[\x{c0}])/8
2081
2082 HOWEVER: Change 17 below makes this fix obsolete as the memory computation
2083 is now done differently.
2084
2085 5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
2086 wrapper classes; (b) implement a new function in the C++ scanner that is
2087 more efficient than the old way of doing things because it avoids levels of
2088 recursion in the regex matching; (c) add a paragraph to the documentation
2089 for the FullMatch() function.
2090
2091 6. The escape sequence \n was being treated as whatever was defined as
2092 "newline". Not only was this contrary to the documentation, which states
2093 that \n is character 10 (hex 0A), but it also went horribly wrong when
2094 "newline" was defined as CRLF. This has been fixed.
2095
2096 7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
2097 was being set to -1 for the "end of line" case (supposedly a value that no
2098 character can have). Though this value is never used (the check for end of
2099 line is "zero bytes in current character"), it caused compiler complaints.
2100 I've changed it to 0xffffffff.
2101
2102 8. In pcre_version.c, the version string was being built by a sequence of
2103 C macros that, in the event of PCRE_PRERELEASE being defined as an empty
2104 string (as it is for production releases) called a macro with an empty
2105 argument. The C standard says the result of this is undefined. The gcc
2106 compiler treats it as an empty string (which was what was wanted) but it is
2107 reported that Visual C gives an error. The source has been hacked around to
2108 avoid this problem.
2109
2110 9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
2111 builds of pcretest, and changed the call to _setmode() to use _O_BINARY
2112 instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
2113 of them did).
2114
2115 10. Originally, pcretest opened its input and output without "b"; then I was
2116 told that "b" was needed in some environments, so it was added for release
2117 5.0 to both the input and output. (It makes no difference on Unix-like
2118 systems.) Later I was told that it is wrong for the input on Windows. I've
2119 now abstracted the modes into two macros, to make it easier to fiddle with
2120 them, and removed "b" from the input mode under Windows.
2121
2122 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
2123
2124 12. Added -help and --help to pcretest as an official way of being reminded
2125 of the options.
2126
2127 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
2128 and pcrecpp.cc because they annoy compilers at high warning levels.
2129
2130 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
2131
2132 15. Fixed an occurrence of == in configure.ac that should have been = (shell
2133 scripts are not C programs :-) and which was not noticed because it works
2134 on Linux.
2135
2136 16. pcretest is supposed to handle any length of pattern and data line (as one
2137 line or as a continued sequence of lines) by extending its input buffer if
2138 necessary. This feature was broken for very long pattern lines, leading to
2139 a string of junk being passed to pcre_compile() if the pattern was longer
2140 than about 50K.
2141
2142 17. I have done a major re-factoring of the way pcre_compile() computes the
2143 amount of memory needed for a compiled pattern. Previously, there was code
2144 that made a preliminary scan of the pattern in order to do this. That was
2145 OK when PCRE was new, but as the facilities have expanded, it has become
2146 harder and harder to keep it in step with the real compile phase, and there
2147 have been a number of bugs (see for example, 4 above). I have now found a
2148 cunning way of running the real compile function in a "fake" mode that
2149 enables it to compute how much memory it would need, while actually only
2150 ever using a few hundred bytes of working memory and without too many
2151 tests of the mode. This should make future maintenance and development
2152 easier. A side effect of this work is that the limit of 200 on the nesting
2153 depth of parentheses has been removed (though this was never a serious
2154 limitation, I suspect). However, there is a downside: pcre_compile() now
2155 runs more slowly than before (30% or more, depending on the pattern). I
2156 hope this isn't a big issue. There is no effect on runtime performance.
2157
2158 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
2159 newline (only possible for the last line of a file) and it was a
2160 pattern that set a locale (followed by /Lsomething), pcretest crashed.
2161
2162 19. Added additional timing features to pcretest. (1) The -tm option now times
2163 matching only, not compiling. (2) Both -t and -tm can be followed, as a
2164 separate command line item, by a number that specifies the number of
2165 repeats to use when timing. The default is 50000; this gives better
2166 precision, but takes uncomfortably long for very large patterns.
2167
2168 20. Extended pcre_study() to be more clever in cases where a branch of a
2169 subpattern has no definite first character. For example, (a*|b*)[cd] would
2170 previously give no result from pcre_study(). Now it recognizes that the
2171 first character must be a, b, c, or d.
2172
2173 21. There was an incorrect error "recursive call could loop indefinitely" if
2174 a subpattern (or the entire pattern) that was being tested for matching an
2175 empty string contained only one non-empty item after a nested subpattern.
2176 For example, the pattern (?>\x{100}*)\d(?R) provoked this error
2177 incorrectly, because the \d was being skipped in the check.
2178
2179 22. The pcretest program now has a new pattern option /B and a command line
2180 option -b, which is equivalent to adding /B to every pattern. This causes
2181 it to show the compiled bytecode, without the additional information that
2182 -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
2183 is the same as /B/I).
2184
2185 23. A new optimization is now able automatically to treat some sequences such
2186 as a*b as a*+b. More specifically, if something simple (such as a character
2187 or a simple class like \d) has an unlimited quantifier, and is followed by
2188 something that cannot possibly match the quantified thing, the quantifier
2189 is automatically "possessified".
2190
2191 24. A recursive reference to a subpattern whose number was greater than 39
2192 went wrong under certain circumstances in UTF-8 mode. This bug could also
2193 have affected the operation of pcre_study().
2194
2195 25. Realized that a little bit of performance could be had by replacing
2196 (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
2197
2198 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
2199
2200 27. Possessive quantifiers such as a++ were previously implemented by turning
2201 them into atomic groups such as ($>a+). Now they have their own opcodes,
2202 which improves performance. This includes the automatically created ones
2203 from 23 above.
2204
2205 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
2206 lookahead was broken if it was not anchored. PCRE was mistakenly expecting
2207 the first matched character to be a colon. This applied both to named and
2208 numbered groups.
2209
2210 29. The ucpinternal.h header file was missing its idempotency #ifdef.
2211
2212 30. I was sent a "project" file called libpcre.a.dev which I understand makes
2213 building PCRE on Windows easier, so I have included it in the distribution.
2214
2215 31. There is now a check in pcretest against a ridiculously large number being
2216 returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
2217 loop, the loop is abandoned.
2218
2219 32. Forward references to subpatterns in conditions such as (?(2)...) where
2220 subpattern 2 is defined later cause pcre_compile() to search forwards in
2221 the pattern for the relevant set of parentheses. This search went wrong
2222 when there were unescaped parentheses in a character class, parentheses
2223 escaped with \Q...\E, or parentheses in a #-comment in /x mode.
2224
2225 33. "Subroutine" calls and backreferences were previously restricted to
2226 referencing subpatterns earlier in the regex. This restriction has now
2227 been removed.
2228
2229 34. Added a number of extra features that are going to be in Perl 5.10. On the
2230 whole, these are just syntactic alternatives for features that PCRE had
2231 previously implemented using the Python syntax or my own invention. The
2232 other formats are all retained for compatibility.
2233
2234 (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
2235 as (?P<name>...). The new forms, as well as being in Perl 5.10, are
2236 also .NET compatible.
2237
2238 (b) A recursion or subroutine call to a named group can now be defined as
2239 (?&name) as well as (?P>name).
2240
2241 (c) A backreference to a named group can now be defined as \k<name> or
2242 \k'name' as well as (?P=name). The new forms, as well as being in Perl
2243 5.10, are also .NET compatible.
2244
2245 (d) A conditional reference to a named group can now use the syntax
2246 (?(<name>) or (?('name') as well as (?(name).
2247
2248 (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
2249 groups (named and numbered) that are never evaluated inline, but can be
2250 called as "subroutines" from elsewhere. In effect, the DEFINE condition
2251 is always false. There may be only one alternative in such a group.
2252
2253 (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
2254 as the simple (?(R). The condition is true only if the most recent
2255 recursion is that of the given number or name. It does not search out
2256 through the entire recursion stack.
2257
2258 (g) The escape \gN or \g{N} has been added, where N is a positive or
2259 negative number, specifying an absolute or relative reference.
2260
2261 35. Tidied to get rid of some further signed/unsigned compiler warnings and
2262 some "unreachable code" warnings.
2263
2264 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
2265 things, this adds five new scripts.
2266
2267 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
2268 There were also incompatibilities regarding the handling of \Q..\E inside
2269 character classes, for example with patterns like [\Qa\E-\Qz\E] where the
2270 hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
2271
2272 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
2273 matches an empty string, and forcibly breaks the loop. There were bugs in
2274 this code in non-simple cases. For a pattern such as ^(a()*)* matched
2275 against aaaa the result was just "a" rather than "aaaa", for example. Two
2276 separate and independent bugs (that affected different cases) have been
2277 fixed.
2278
2279 39. Refactored the code to abolish the use of different opcodes for small
2280 capturing bracket numbers. This is a tidy that I avoided doing when I
2281 removed the limit on the number of capturing brackets for 3.5 back in 2001.
2282 The new approach is not only tidier, it makes it possible to reduce the
2283 memory needed to fix the previous bug (38).
2284
2285 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
2286 sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
2287 processing dot, circumflex, or dollar metacharacters, or #-comments in /x
2288 mode.
2289
2290 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
2291 report.
2292
2293 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
2294 copy construction and assignment in the C++ wrapper.
2295
2296 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
2297 couple of bugs that could have given wrong results in the "--newline=crlf"
2298 case.
2299
2300 44. Added a number of casts and did some reorganization of signed/unsigned int
2301 variables following suggestions from Dair Grant. Also renamed the variable
2302 "this" as "item" because it is a C++ keyword.
2303
2304 45. Arranged for dftables to add
2305
2306 #include "pcre_internal.h"
2307
2308 to pcre_chartables.c because without it, gcc 4.x may remove the array
2309 definition from the final binary if PCRE is built into a static library and
2310 dead code stripping is activated.
2311
2312 46. For an unanchored pattern, if a match attempt fails at the start of a
2313 newline sequence, and the newline setting is CRLF or ANY, and the next two
2314 characters are CRLF, advance by two characters instead of one.
2315
2316
2317 Version 6.7 04-Jul-06
2318 ---------------------
2319
2320 1. In order to handle tests when input lines are enormously long, pcretest has
2321 been re-factored so that it automatically extends its buffers when
2322 necessary. The code is crude, but this _is_ just a test program. The
2323 default size has been increased from 32K to 50K.
2324
2325 2. The code in pcre_study() was using the value of the re argument before
2326 testing it for NULL. (Of course, in any sensible call of the function, it
2327 won't be NULL.)
2328
2329 3. The memmove() emulation function in pcre_internal.h, which is used on
2330 systems that lack both memmove() and bcopy() - that is, hardly ever -
2331 was missing a "static" storage class specifier.
2332
2333 4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
2334 containing an extended class (one that cannot be represented by a bitmap
2335 because it contains high-valued characters or Unicode property items, e.g.
2336 [\pZ]). Almost always one would set UTF-8 mode when processing such a
2337 pattern, but PCRE should not loop if you do not (it no longer does).
2338 [Detail: two cases were found: (a) a repeated subpattern containing an
2339 extended class; (b) a recursive reference to a subpattern that followed a
2340 previous extended class. It wasn't skipping over the extended class
2341 correctly when UTF-8 mode was not set.]
2342
2343 5. A negated single-character class was not being recognized as fixed-length
2344 in lookbehind assertions such as (?<=[^f]), leading to an incorrect
2345 compile error "lookbehind assertion is not fixed length".
2346
2347 6. The RunPerlTest auxiliary script was showing an unexpected difference
2348 between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
2349 write a Perl script that can interpret lines of an input file either as
2350 byte characters or as UTF-8, which is what "perltest" was being required to
2351 do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
2352 can't do is switch easily at run time between having the "use utf8;" pragma
2353 or not. In the end, I fudged it by using the RunPerlTest script to insert
2354 "use utf8;" explicitly for the UTF-8 tests.
2355
2356 7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
2357 the end of the subject string, contrary to the documentation and to what
2358 Perl does. This was true of both matching functions. Now it matches only at
2359 the start of the subject and immediately after *internal* newlines.
2360
2361 8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
2362 a pointer to an int instead of a pointer to an unsigned long int. This
2363 caused problems on 64-bit systems.
2364
2365 9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
2366 instance of the 'standard' template library not being so standard".
2367
2368 10. There was no check on the number of named subpatterns nor the maximum
2369 length of a subpattern name. The product of these values is used to compute
2370 the size of the memory block for a compiled pattern. By supplying a very
2371 long subpattern name and a large number of named subpatterns, the size
2372 computation could be caused to overflow. This is now prevented by limiting
2373 the length of names to 32 characters, and the number of named subpatterns
2374 to 10,000.
2375
2376 11. Subpatterns that are repeated with specific counts have to be replicated in
2377 the compiled pattern. The size of memory for this was computed from the
2378 length of the subpattern and the repeat count. The latter is limited to
2379 65535, but there was no limit on the former, meaning that integer overflow
2380 could in principle occur. The compiled length of a repeated subpattern is
2381 now limited to 30,000 bytes in order to prevent this.
2382
2383 12. Added the optional facility to have named substrings with the same name.
2384
2385 13. Added the ability to use a named substring as a condition, using the
2386 Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
2387 are numbers (not recommended). Forward references are permitted.
2388
2389 14. Added forward references in named backreferences (if you see what I mean).
2390
2391 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
2392 pattern could run off the end of the subject. For example, the pattern
2393 "(?s)(.{1,5})"8 did this with the subject "ab".
2394
2395 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
2396 PCRE_CASELESS was set when matching characters that were quantified with ?
2397 or *.
2398
2399 17. A character class other than a single negated character that had a minimum
2400 but no maximum quantifier - for example [ab]{6,} - was not handled
2401 correctly by pce_dfa_exec(). It would match only one character.
2402
2403 18. A valid (though odd) pattern that looked like a POSIX character
2404 class but used an invalid character after [ (for example [[,abc,]]) caused
2405 pcre_compile() to give the error "Failed: internal error: code overflow" or
2406 in some cases to crash with a glibc free() error. This could even happen if
2407 the pattern terminated after [[ but there just happened to be a sequence of
2408 letters, a binary zero, and a closing ] in the memory that followed.
2409
2410 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
2411 over the years. Originally (before any Unicode support), just the bottom 8
2412 bits were taken. Thus, for example, \500 really meant \100. Nowadays the
2413 output from "man perlunicode" includes this:
2414
2415 The regular expression compiler produces polymorphic opcodes. That
2416 is, the pattern adapts to the data and automatically switches to
2417 the Unicode character scheme when presented with Unicode data--or
2418 instead uses a traditional byte scheme when presented with byte
2419 data.
2420
2421 Sadly, a wide octal escape does not cause a switch, and in a string with
2422 no other multibyte characters, these octal escapes are treated as before.
2423 Thus, in Perl, the pattern /\500/ actually matches \100 but the pattern
2424 /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
2425 Unicode string.
2426
2427 I have not perpetrated such confusion in PCRE. Up till now, it took just
2428 the bottom 8 bits, as in old Perl. I have now made octal escapes with
2429 values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
2430 translate to the appropriate multibyte character.
2431
2432 29. Applied some refactoring to reduce the number of warnings from Microsoft
2433 and Borland compilers. This has included removing the fudge introduced
2434 seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
2435 a warning about an unused variable.
2436
2437 21. PCRE has not included VT (character 0x0b) in the set of whitespace
2438 characters since release 4.0, because Perl (from release 5.004) does not.
2439 [Or at least, is documented not to: some releases seem to be in conflict
2440 with the documentation.] However, when a pattern was studied with
2441 pcre_study() and all its branches started with \s, PCRE still included VT
2442 as a possible starting character. Of course, this did no harm; it just
2443 caused an unnecessary match attempt.
2444
2445 22. Removed a now-redundant internal flag bit that recorded the fact that case
2446 dependency changed within the pattern. This was once needed for "required
2447 byte" processing, but is no longer used. This recovers a now-scarce options
2448 bit. Also moved the least significant internal flag bit to the most-
2449 significant bit of the word, which was not previously used (hangover from
2450 the days when it was an int rather than a uint) to free up another bit for
2451 the future.
2452
2453 23. Added support for CRLF line endings as well as CR and LF. As well as the
2454 default being selectable at build time, it can now be changed at runtime
2455 via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
2456 specify that it is scanning data with non-default line endings.
2457
2458 24. Changed the definition of CXXLINK to make it agree with the definition of
2459 LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
2460
2461 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
2462 recursions. This makes a big different to stack usage for some patterns.
2463
2464 26. If a subpattern containing a named recursion or subroutine reference such
2465 as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
2466 the space required for the compiled pattern went wrong and gave too small a
2467 value. Depending on the environment, this could lead to "Failed: internal
2468 error: code overflow at offset 49" or "glibc detected double free or
2469 corruption" errors.
2470
2471 27. Applied patches from Google (a) to support the new newline modes and (b) to
2472 advance over multibyte UTF-8 characters in GlobalReplace.
2473
2474 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
2475 difference for some implementation of PCRE in some Windows version.
2476
2477 29. Added some extra testing facilities to pcretest:
2478
2479 \q<number> in a data line sets the "match limit" value
2480 \Q<number> in a data line sets the "match recursion limt" value
2481 -S <number> sets the stack size, where <number> is in megabytes
2482
2483 The -S option isn't available for Windows.
2484
2485
2486 Version 6.6 06-Feb-06
2487 ---------------------
2488
2489 1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
2490 in pcreposix.h. I have copied the definition from pcre.h.
2491
2492 2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
2493 because pcre.h is no longer a built file.
2494
2495 3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
2496 not normally included in the compiled code.
2497
2498
2499 Version 6.5 01-Feb-06
2500 ---------------------
2501
2502 1. When using the partial match feature with pcre_dfa_exec(), it was not
2503 anchoring the second and subsequent partial matches at the new starting
2504 point. This could lead to incorrect results. For example, with the pattern
2505 /1234/, partially matching against "123" and then "a4" gave a match.
2506
2507 2. Changes to pcregrep:
2508
2509 (a) All non-match returns from pcre_exec() were being treated as failures
2510 to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
2511 error message is output. Some extra information is given for the
2512 PCRE_ERROR_MATCHLIMIT and PCRE_ERROR_RECURSIONLIMIT errors, which are
2513 probably the only errors that are likely to be caused by users (by
2514 specifying a regex that has nested indefinite repeats, for instance).
2515 If there are more than 20 of these errors, pcregrep is abandoned.
2516
2517 (b) A binary zero was treated as data while matching, but terminated the
2518 output line if it was written out. This has been fixed: binary zeroes
2519 are now no different to any other data bytes.
2520
2521 (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
2522 used to set a locale for matching. The --locale=xxxx long option has
2523 been added (no short equivalent) to specify a locale explicitly on the
2524 pcregrep command, overriding the environment variables.
2525
2526 (d) When -B was used with -n, some line numbers in the output were one less
2527 than they should have been.
2528
2529 (e) Added the -o (--only-matching) option.
2530
2531 (f) If -A or -C was used with -c (count only), some lines of context were
2532 accidentally printed for the final match.
2533
2534 (g) Added the -H (--with-filename) option.
2535
2536 (h) The combination of options -rh failed to suppress file names for files
2537 that were found from directory arguments.
2538
2539 (i) Added the -D (--devices) and -d (--directories) options.
2540
2541 (j) Added the -F (--fixed-strings) option.
2542
2543 (k) Allow "-" to be used as a file name for -f as well as for a data file.
2544
2545 (l) Added the --colo(u)r option.
2546
2547 (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
2548 is not present by default.
2549
2550 3. A nasty bug was discovered in the handling of recursive patterns, that is,
2551 items such as (?R) or (?1), when the recursion could match a number of
2552 alternatives. If it matched one of the alternatives, but subsequently,
2553 outside the recursion, there was a failure, the code tried to back up into
2554 the recursion. However, because of the way PCRE is implemented, this is not
2555 possible, and the result was an incorrect result from the match.
2556
2557 In order to prevent this happening, the specification of recursion has
2558 been changed so that all such subpatterns are automatically treated as
2559 atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
2560
2561 4. I had overlooked the fact that, in some locales, there are characters for
2562 which isalpha() is true but neither isupper() nor islower() are true. In
2563 the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
2564 and ordfeminine) are like this. This affected the treatment of \w and \W
2565 when they appeared in character classes, but not when they appeared outside
2566 a character class. The bit map for "word" characters is now created
2567 separately from the results of isalnum() instead of just taking it from the
2568 upper, lower, and digit maps. (Plus the underscore character, of course.)
2569
2570 5. The above bug also affected the handling of POSIX character classes such as
2571 [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
2572 permanent tables. Instead, the bit maps for such a class were previously
2573 created as the appropriate unions of the upper, lower, and digit bitmaps.
2574 Now they are created by subtraction from the [[:word:]] class, which has
2575 its own bitmap.
2576
2577 6. The [[:blank:]] character class matches horizontal, but not vertical space.
2578 It is created by subtracting the vertical space characters (\x09, \x0a,
2579 \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
2580 subtraction was done in the overall bitmap for a character class, meaning
2581 that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
2582 be recognized. This bug has been fixed.
2583
2584 7. Patches from the folks at Google:
2585
2586 (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
2587 real life, but is still worth protecting against".
2588
2589 (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
2590 regular expressions".
2591
2592 (c) pcre_scanner.cc: avoid use of std::count() because not all systems
2593 have it.
2594
2595 (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
2596 "configure" and the latter not, in order to fix a problem somebody had
2597 with compiling the Arg class on HP-UX.
2598
2599 (e) Improve the error-handling of the C++ wrapper a little bit.
2600
2601 (f) New tests for checking recursion limiting.
2602
2603 8. The pcre_memmove() function, which is used only if the environment does not
2604 have a standard memmove() function (and is therefore rarely compiled),
2605 contained two bugs: (a) use of int instead of size_t, and (b) it was not
2606 returning a result (though PCRE never actually uses the result).
2607
2608 9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
2609 large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
2610 returned instead of calling malloc() with an overflowing number that would
2611 most likely cause subsequent chaos.
2612
2613 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
2614
2615 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
2616 with this option is matched, the nmatch and pmatch options of regexec() are
2617 ignored.
2618
2619 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
2620 provided in case anyone wants to the the POSIX interface with UTF-8
2621 strings.
2622
2623 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
2624 C++ linking (needed for some HP-UX environments).
2625
2626 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
2627 (unused parameter) and in the pcre_printint() function (omitted "default"
2628 switch label when the default is to do nothing).
2629
2630 15. Added some code to make it possible, when PCRE is compiled as a C++
2631 library, to replace subject pointers for pcre_exec() with a smart pointer
2632 class, thus making it possible to process discontinuous strings.
2633
2634 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
2635 much the same function. They were added by different people who were trying
2636 to make PCRE easy to compile on non-Unix systems. It has been suggested
2637 that PCRE_EXPORT be abolished now that there is more automatic apparatus
2638 for compiling on Windows systems. I have therefore replaced it with
2639 PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
2640 defaults to "extern" for C or "extern C" for C++, which works fine on
2641 Unix-like systems. It is now possible to override the value of PCRE_DATA_
2642 SCOPE with something explicit in config.h. In addition:
2643
2644 (a) pcreposix.h still had just "extern" instead of either of these macros;
2645 I have replaced it with PCRE_DATA_SCOPE.
2646
2647 (b) Functions such as _pcre_xclass(), which are internal to the library,
2648 but external in the C sense, all had PCRE_EXPORT in their definitions.
2649 This is apparently wrong for the Windows case, so I have removed it.
2650 (It makes no difference on Unix-like systems.)
2651
2652 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
2653 of recursive calls to match(). This is different to MATCH_LIMIT because
2654 that limits the total number of calls to match(), not all of which increase
2655 the depth of recursion. Limiting the recursion depth limits the amount of
2656 stack (or heap if NO_RECURSE is set) that is used. The default can be set
2657 when PCRE is compiled, and changed at run time. A patch from Google adds
2658 this functionality to the C++ interface.
2659
2660 18. Changes to the handling of Unicode character properties:
2661
2662 (a) Updated the table to Unicode 4.1.0.
2663
2664 (b) Recognize characters that are not in the table as "Cn" (undefined).
2665
2666 (c) I revised the way the table is implemented to a much improved format
2667 which includes recognition of ranges. It now supports the ranges that
2668 are defined in UnicodeData.txt, and it also amalgamates other
2669 characters into ranges. This has reduced the number of entries in the
2670 table from around 16,000 to around 3,000, thus reducing its size
2671 considerably. I realized I did not need to use a tree structure after
2672 all - a binary chop search is just as efficient. Having reduced the
2673 number of entries, I extended their size from 6 bytes to 8 bytes to
2674 allow for more data.
2675
2676 (d) Added support for Unicode script names via properties such as \p{Han}.
2677
2678 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
2679 matching that character.
2680
2681 20. When matching a repeated Unicode property with a minimum greater than zero,
2682 (for example \pL{2,}), PCRE could look past the end of the subject if it
2683 reached it while seeking the minimum number of characters. This could
2684 happen only if some of the characters were more than one byte long, because
2685 there is a check for at least the minimum number of bytes.
2686
2687 21. Refactored the implementation of \p and \P so as to be more general, to
2688 allow for more different types of property in future. This has changed the
2689 compiled form incompatibly. Anybody with saved compiled patterns that use
2690 \p or \P will have to recompile them.
2691
2692 22. Added "Any" and "L&" to the supported property types.
2693
2694 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
2695 but give a compile time error if the value is greater than 0xff.
2696
2697 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
2698 accidentally not being installed or uninstalled.
2699
2700 25. The pcre.h file was built from pcre.h.in, but the only changes that were
2701 made were to insert the current release number. This seemed silly, because
2702 it made things harder for people building PCRE on systems that don't run
2703 "configure". I have turned pcre.h into a distributed file, no longer built
2704 by "configure", with the version identification directly included. There is
2705 no longer a pcre.h.in file.
2706
2707 However, this change necessitated a change to the pcre-config script as
2708 well. It is built from pcre-config.in, and one of the substitutions was the
2709 release number. I have updated configure.ac so that ./configure now finds
2710 the release number by grepping pcre.h.
2711
2712 26. Added the ability to run the tests under valgrind.
2713
2714
2715 Version 6.4 05-Sep-05
2716 ---------------------
2717
2718 1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
2719 "--" to be printed when multiple files were scanned, even when none of the
2720 -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
2721 consider it to be a bug, and have restored the previous behaviour.
2722
2723 2. A couple of code tidies to get rid of compiler warnings.
2724
2725 3. The pcretest program used to cheat by referring to symbols in the library
2726 whose names begin with _pcre_. These are internal symbols that are not
2727 really supposed to be visible externally, and in some environments it is
2728 possible to suppress them. The cheating is now confined to including
2729 certain files from the library's source, which is a bit cleaner.
2730
2731 4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
2732 file's purpose clearer.
2733
2734 5. Reorganized pcre_ucp_findchar().
2735
2736
2737 Version 6.3 15-Aug-05
2738 ---------------------
2739
2740 1. The file libpcre.pc.in did not have general read permission in the tarball.
2741
2742 2. There were some problems when building without C++ support:
2743
2744 (a) If C++ support was not built, "make install" and "make test" still
2745 tried to test it.
2746
2747 (b) There were problems when the value of CXX was explicitly set. Some
2748 changes have been made to try to fix these, and ...
2749
2750 (c) --disable-cpp can now be used to explicitly disable C++ support.
2751
2752 (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
2753 backslash in a target when C++ was disabled. This confuses some
2754 versions of "make", apparently. Using an intermediate variable solves
2755 this. (Same for CPP_LOBJ.)
2756
2757 3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
2758 (non-Windows) now includes $(CFLAGS) because these flags are sometimes
2759 necessary on certain architectures.
2760
2761 4. Added a setting of -export-symbols-regex to the link command to remove
2762 those symbols that are exported in the C sense, but actually are local
2763 within the library, and not documented. Their names all begin with
2764 "_pcre_". This is not a perfect job, because (a) we have to except some
2765 symbols that pcretest ("illegally") uses, and (b) the facility isn't always
2766 available (and never for static libraries). I have made a note to try to
2767 find a way round (a) in the future.
2768
2769
2770 Version 6.2 01-Aug-05
2771 ---------------------
2772
2773 1. There was no test for integer overflow of quantifier values. A construction
2774 such as {1111111111111111} would give undefined results. What is worse, if
2775 a minimum quantifier for a parenthesized subpattern overflowed and became
2776 negative, the calculation of the memory size went wrong. This could have
2777 led to memory overwriting.
2778
2779 2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
2780
2781 3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
2782 operating environments where this matters.
2783
2784 4. Applied Giuseppe Maxia's patch to add additional features for controlling
2785 PCRE options from within the C++ wrapper.
2786
2787 5. Named capturing subpatterns were not being correctly counted when a pattern
2788 was compiled. This caused two problems: (a) If there were more than 100
2789 such subpatterns, the calculation of the memory needed for the whole
2790 compiled pattern went wrong, leading to an overflow error. (b) Numerical
2791 back references of the form \12, where the number was greater than 9, were
2792 not recognized as back references, even though there were sufficient
2793 previous subpatterns.
2794
2795 6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
2796 versions of gcc, e.g. 2.95.4.
2797
2798
2799 Version 6.1 21-Jun-05
2800 ---------------------
2801
2802 1. There was one reference to the variable "posix" in pcretest.c that was not
2803 surrounded by "#if !defined NOPOSIX".
2804
2805 2. Make it possible to compile pcretest without DFA support, UTF8 support, or
2806 the cross-check on the old pcre_info() function, for the benefit of the
2807 cut-down version of PCRE that is currently imported into Exim.
2808
2809 3. A (silly) pattern starting with (?i)(?-i) caused an internal space
2810 allocation error. I've done the easy fix, which wastes 2 bytes for sensible
2811 patterns that start (?i) but I don't think that matters. The use of (?i) is
2812 just an example; this all applies to the other options as well.
2813
2814 4. Since libtool seems to echo the compile commands it is issuing, the output
2815 from "make" can be reduced a bit by putting "@" in front of each libtool
2816 compile command.
2817
2818 5. Patch from the folks at Google for configure.in to be a bit more thorough
2819 in checking for a suitable C++ installation before trying to compile the
2820 C++ stuff. This should fix a reported problem when a compiler was present,
2821 but no suitable headers.
2822
2823 6. The man pages all had just "PCRE" as their title. I have changed them to
2824 be the relevant file name. I have also arranged that these names are
2825 retained in the file doc/pcre.txt, which is a concatenation in text format
2826 of all the man pages except the little individual ones for each function.
2827
2828 7. The NON-UNIX-USE file had not been updated for the different set of source
2829 files that come with release 6. I also added a few comments about the C++
2830 wrapper.
2831
2832
2833 Version 6.0 07-Jun-05
2834 ---------------------
2835
2836 1. Some minor internal re-organization to help with my DFA experiments.
2837
2838 2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
2839 didn't matter for the library itself when fully configured, but did matter
2840 when compiling without UCP support, or within Exim, where the ucp files are
2841 not imported.
2842
2843 3. Refactoring of the library code to split up the various functions into
2844 different source modules. The addition of the new DFA matching code (see
2845 below) to a single monolithic source would have made it really too
2846 unwieldy, quite apart from causing all the code to be include in a
2847 statically linked application, when only some functions are used. This is
2848 relevant even without the DFA addition now that patterns can be compiled in
2849 one application and matched in another.
2850
2851 The downside of splitting up is that there have to be some external
2852 functions and data tables that are used internally in different modules of
2853 the library but which are not part of the API. These have all had their
2854 names changed to start with "_pcre_" so that they are unlikely to clash
2855 with other external names.
2856
2857 4. Added an alternate matching function, pcre_dfa_exec(), which matches using
2858 a different (DFA) algorithm. Although it is slower than the original
2859 function, it does have some advantages for certain types of matching
2860 problem.
2861
2862 5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
2863 including restarting after a partial match.
2864
2865 6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
2866 defined when compiling for Windows was sent to me. I have put it into the
2867 code, though I have no means of testing or verifying it.
2868
2869 7. Added the pcre_refcount() auxiliary function.
2870
2871 8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
2872 match before or at the first newline in the subject string. In pcretest,
2873 the /f option on a pattern can be used to set this.
2874
2875 9. A repeated \w when used in UTF-8 mode with characters greater than 256
2876 would behave wrongly. This has been present in PCRE since release 4.0.
2877
2878 10. A number of changes to the pcregrep command:
2879
2880 (a) Refactored how -x works; insert ^(...)$ instead of setting
2881 PCRE_ANCHORED and checking the length, in preparation for adding
2882 something similar for -w.
2883
2884 (b) Added the -w (match as a word) option.
2885
2886 (c) Refactored the way lines are read and buffered so as to have more
2887 than one at a time available.
2888
2889 (d) Implemented a pcregrep test script.
2890
2891 (e) Added the -M (multiline match) option. This allows patterns to match
2892 over several lines of the subject. The buffering ensures that at least
2893 8K, or the rest of the document (whichever is the shorter) is available
2894 for matching (and similarly the previous 8K for lookbehind assertions).
2895
2896 (f) Changed the --help output so that it now says
2897
2898 -w, --word-regex(p)
2899
2900 instead of two lines, one with "regex" and the other with "regexp"
2901 because that confused at least one person since the short forms are the
2902 same. (This required a bit of code, as the output is generated
2903 automatically from a table. It wasn't just a text change.)
2904
2905 (g) -- can be used to terminate pcregrep options if the next thing isn't an
2906 option but starts with a hyphen. Could be a pattern or a path name
2907 starting with a hyphen, for instance.
2908
2909 (h) "-" can be given as a file name to represent stdin.
2910
2911 (i) When file names are being printed, "(standard input)" is used for
2912 the standard input, for compatibility with GNU grep. Previously
2913 "<stdin>" was used.
2914
2915 (j) The option --label=xxx can be used to supply a name to be used for
2916 stdin when file names are being printed. There is no short form.
2917
2918 (k) Re-factored the options decoding logic because we are going to add
2919 two more options that take data. Such options can now be given in four
2920 different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
2921
2922 (l) Added the -A, -B, and -C options for requesting that lines of context
2923 around matches be printed.
2924
2925 (m) Added the -L option to print the names of files that do not contain
2926 any matching lines, that is, the complement of -l.
2927
2928 (n) The return code is 2 if any file cannot be opened, but pcregrep does
2929 continue to scan other files.
2930
2931 (o) The -s option was incorrectly implemented. For compatibility with other
2932 greps, it now suppresses the error message for a non-existent or non-
2933 accessible file (but not the return code). There is a new option called
2934 -q that suppresses the output of matching lines, which was what -s was
2935 previously doing.
2936
2937 (p) Added --include and --exclude options to specify files for inclusion
2938 and exclusion when recursing.
2939
2940 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
2941 Hopefully, it now does.
2942
2943 12. Missing cast in pcre_study().
2944
2945 13. Added an "uninstall" target to the makefile.
2946
2947 14. Replaced "extern" in the function prototypes in Makefile.in with
2948 "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
2949 world, but is set differently for Windows.
2950
2951 15. Added a second compiling function called pcre_compile2(). The only
2952 difference is that it has an extra argument, which is a pointer to an
2953 integer error code. When there is a compile-time failure, this is set
2954 non-zero, in addition to the error test pointer being set to point to an
2955 error message. The new argument may be NULL if no error number is required
2956 (but then you may as well call pcre_compile(), which is now just a
2957 wrapper). This facility is provided because some applications need a
2958 numeric error indication, but it has also enabled me to tidy up the way
2959 compile-time errors are handled in the POSIX wrapper.
2960
2961 16. Added VPATH=.libs to the makefile; this should help when building with one
2962 prefix path and installing with another. (Or so I'm told by someone who
2963 knows more about this stuff than I do.)
2964
2965 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
2966 passes PCRE_DOTALL to the pcre_compile() function, making the "." character
2967 match everything, including newlines. This is not POSIX-compatible, but
2968 somebody wanted the feature. From pcretest it can be activated by using
2969 both the P and the s flags.
2970
2971 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
2972
2973 19. libpcre.pc was being incorrectly installed as executable.
2974
2975 20. A couple of places in pcretest check for end-of-line by looking for '\n';
2976 it now also looks for '\r' so that it will work unmodified on Windows.
2977
2978 21. Added Google's contributed C++ wrapper to the distribution.
2979
2980 22. Added some untidy missing memory free() calls in pcretest, to keep
2981 Electric Fence happy when testing.
2982
2983
2984
2985 Version 5.0 13-Sep-04
2986 ---------------------
2987
2988 1. Internal change: literal characters are no longer packed up into items
2989 containing multiple characters in a single byte-string. Each character
2990 is now matched using a separate opcode. However, there may be more than one
2991 byte in the character in UTF-8 mode.
2992
2993 2. The pcre_callout_block structure has two new fields: pattern_position and
2994 next_item_length. These contain the offset in the pattern to the next match
2995 item, and its length, respectively.
2996
2997 3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
2998 insertion of callouts before each pattern item. Added the /C option to
2999 pcretest to make use of this.
3000
3001 4. On the advice of a Windows user, the lines
3002
3003 #if defined(_WIN32) || defined(WIN32)
3004 _setmode( _fileno( stdout ), 0x8000 );
3005 #endif /* defined(_WIN32) || defined(WIN32) */
3006
3007 have been added to the source of pcretest. This apparently does useful
3008 magic in relation to line terminators.
3009
3010 5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
3011 for the benefit of those environments where the "b" makes a difference.
3012
3013 6. The icc compiler has the same options as gcc, but "configure" doesn't seem
3014 to know about it. I have put a hack into configure.in that adds in code
3015 to set GCC=yes if CC=icc. This seems to end up at a point in the
3016 generated configure script that is early enough to affect the setting of
3017 compiler options, which is what is needed, but I have no means of testing
3018 whether it really works. (The user who reported this had patched the
3019 generated configure script, which of course I cannot do.)
3020
3021 LATER: After change 22 below (new libtool files), the configure script
3022 seems to know about icc (and also ecc). Therefore, I have commented out
3023 this hack in configure.in.
3024
3025 7. Added support for pkg-config (2 patches were sent in).
3026
3027 8. Negated POSIX character classes that used a combination of internal tables
3028 were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
3029 [[:^ascii]]. Typically, they would match almost any characters. The other
3030 POSIX classes were not broken in this way.
3031
3032 9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
3033 to find the match, as PCRE was deluded into thinking that the match had to
3034 start at the start point or following a newline. The same bug applied to
3035 patterns with negative forward assertions or any backward assertions
3036 preceding ".*" at the start, unless the pattern required a fixed first
3037 character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
3038
3039 10. In UTF-8 mode, when moving forwards in the subject after a failed match
3040 starting at the last subject character, bytes beyond the end of the subject
3041 string were read.
3042
3043 11. Renamed the variable "class" as "classbits" to make life easier for C++
3044 users. (Previously there was a macro definition, but it apparently wasn't
3045 enough.)
3046
3047 12. Added the new field "tables" to the extra data so that tables can be passed
3048 in at exec time, or the internal tables can be re-selected. This allows
3049 a compiled regex to be saved and re-used at a later time by a different
3050 program that might have everything at different addresses.
3051
3052 13. Modified the pcre-config script so that, when run on Solaris, it shows a
3053 -R library as well as a -L library.
3054
3055 14. The debugging options of pcretest (-d on the command line or D on a
3056 pattern) showed incorrect output for anything following an extended class
3057 that contained multibyte characters and which was followed by a quantifier.
3058
3059 15. Added optional support for general category Unicode character properties
3060 via the \p, \P, and \X escapes. Unicode property support implies UTF-8
3061 support. It adds about 90K to the size of the library. The meanings of the
3062 inbuilt class escapes such as \d and \s have NOT been changed.
3063
3064 16. Updated pcredemo.c to include calls to free() to release the memory for the
3065 compiled pattern.
3066
3067 17. The generated file chartables.c was being created in the source directory
3068 instead of in the building directory. This caused the build to fail if the
3069 source directory was different from the building directory, and was
3070 read-only.
3071
3072 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
3073 file. No doubt somebody will tell me if they don't make sense... Also added
3074 Dan Mooney's comments about building on OpenVMS.
3075
3076 19. Added support for partial matching via the PCRE_PARTIAL option for
3077 pcre_exec() and the \P data escape in pcretest.
3078
3079 20. Extended pcretest with 3 new pattern features:
3080
3081 (i) A pattern option of the form ">rest-of-line" causes pcretest to
3082 write the compiled pattern to the file whose name is "rest-of-line".
3083 This is a straight binary dump of the data, with the saved pointer to
3084 the character tables forced to be NULL. The study data, if any, is
3085 written too. After writing, pcretest reads a new pattern.
3086
3087 (ii) If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
3088 compiled pattern from the given file. There must not be any
3089 occurrences of "<" in the file name (pretty unlikely); if there are,
3090 pcretest will instead treat the initial "<" as a pattern delimiter.
3091 After reading in the pattern, pcretest goes on to read data lines as
3092 usual.
3093
3094 (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
3095 and 16-bit fields in a compiled pattern, to simulate a pattern that
3096 was compiled on a host of opposite endianness.
3097
3098 21. The pcre-exec() function can now cope with patterns that were compiled on
3099 hosts of opposite endianness, with this restriction:
3100
3101 As for any compiled expression that is saved and used later, the tables
3102 pointer field cannot be preserved; the extra_data field in the arguments
3103 to pcre_exec() should be used to pass in a tables address if a value
3104 other than the default internal tables were used at compile time.
3105
3106 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
3107 now diagnosed as an error. Previously, most of the time, a negative number
3108 would have been treated as zero, but if in addition "ovector" was passed as
3109 NULL, a crash could occur.
3110
3111 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
3112 new versions from the libtool 1.5 distribution (the last one is a copy of
3113 a file called libtool.m4). This seems to have fixed the need to patch
3114 "configure" to support Darwin 1.3 (which I used to do). However, I still
3115 had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
3116 workstation).
3117
3118 24. Changed the PCRE licence to be the more standard "BSD" licence.
3119
3120
3121 Version 4.5 01-Dec-03
3122 ---------------------
3123
3124 1. There has been some re-arrangement of the code for the match() function so
3125 that it can be compiled in a version that does not call itself recursively.
3126 Instead, it keeps those local variables that need separate instances for
3127 each "recursion" in a frame on the heap, and gets/frees frames whenever it
3128 needs to "recurse". Keeping track of where control must go is done by means
3129 of setjmp/longjmp. The whole thing is implemented by a set of macros that
3130 hide most of the details from the main code, and operates only if
3131 NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
3132 "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
3133 operating.
3134
3135 To make it easier for callers to provide specially tailored get/free
3136 functions for this usage, two new functions, pcre_stack_malloc, and
3137 pcre_stack_free, are used. They are always called in strict stacking order,
3138 and the size of block requested is always the same.
3139
3140 The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
3141 PCRE has been compiled to use the stack or the heap for recursion. The
3142 -C option of pcretest uses this to show which version is compiled.
3143
3144 A new data escape \S, is added to pcretest; it causes the amounts of store
3145 obtained and freed by both kinds of malloc/free at match time to be added
3146 to the output.
3147
3148 2. Changed the locale test to use "fr_FR" instead of "fr" because that's
3149 what's available on my current Linux desktop machine.
3150
3151 3. When matching a UTF-8 string, the test for a valid string at the start has
3152 been extended. If start_offset is not zero, PCRE now checks that it points
3153 to a byte that is the start of a UTF-8 character. If not, it returns
3154 PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
3155 this is necessary because there may be backward assertions in the pattern.
3156 When matching the same subject several times, it may save resources to use
3157 PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
3158
3159 4. The code for checking the validity of UTF-8 strings has been tightened so
3160 that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
3161 containing "overlong sequences".
3162
3163 5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
3164 I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
3165 should have been "&", but it just so happened that all the cases this let
3166 through by mistake were picked up later in the function.
3167
3168 6. I had used a variable called "isblank" - this is a C99 function, causing
3169 some compilers to warn. To avoid this, I renamed it (as "blankclass").
3170
3171 7. Cosmetic: (a) only output another newline at the end of pcretest if it is
3172 prompting; (b) run "./pcretest /dev/null" at the start of the test script
3173 so the version is shown; (c) stop "make test" echoing "./RunTest".
3174
3175 8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
3176
3177 9. The prototype for memmove() for systems that don't have it was using
3178 size_t, but the inclusion of the header that defines size_t was later. I've
3179 moved the #includes for the C headers earlier to avoid this.
3180
3181 10. Added some adjustments to the code to make it easier to compiler on certain
3182 special systems:
3183
3184 (a) Some "const" qualifiers were missing.
3185 (b) Added the macro EXPORT before all exported functions; by default this
3186 is defined to be empty.
3187 (c) Changed the dftables auxiliary program (that builds chartables.c) so
3188 that it reads its output file name as an argument instead of writing
3189 to the standard output and assuming this can be redirected.
3190
3191 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
3192 class containing characters with values greater than 255, PCRE compilation
3193 went into a loop.
3194
3195 12. A recursive reference to a subpattern that was within another subpattern
3196 that had a minimum quantifier of zero caused PCRE to crash. For example,
3197 (x(y(?2))z)? provoked this bug with a subject that got as far as the
3198 recursion. If the recursively-called subpattern itself had a zero repeat,
3199 that was OK.
3200
3201 13. In pcretest, the buffer for reading a data line was set at 30K, but the
3202 buffer into which it was copied (for escape processing) was still set at
3203 1024, so long lines caused crashes.
3204
3205 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
3206 "internal error: code overflow...". This applied to any character class
3207 that was followed by a possessive quantifier.
3208
3209 15. Modified the Makefile to add libpcre.la as a prerequisite for
3210 libpcreposix.la because I was told this is needed for a parallel build to
3211 work.
3212
3213 16. If a pattern that contained .* following optional items at the start was
3214 studied, the wrong optimizing data was generated, leading to matching
3215 errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
3216 matching string must start with a or b or c. The correct conclusion for
3217 this pattern is that a match can start with any character.
3218
3219
3220 Version 4.4 13-Aug-03
3221 ---------------------
3222
3223 1. In UTF-8 mode, a character class containing characters with values between
3224 127 and 255 was not handled correctly if the compiled pattern was studied.
3225 In fixing this, I have also improved the studying algorithm for such
3226 classes (slightly).
3227
3228 2. Three internal functions had redundant arguments passed to them. Removal
3229 might give a very teeny performance improvement.
3230
3231 3. Documentation bug: the value of the capture_top field in a callout is *one
3232 more than* the number of the hightest numbered captured substring.
3233
3234 4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
3235 in incorrectly linking with a previously installed version. They now link
3236 explicitly with libpcre.la.
3237
3238 5. configure.in no longer needs to recognize Cygwin specially.
3239
3240 6. A problem in pcre.in for Windows platforms is fixed.
3241
3242 7. If a pattern was successfully studied, and the -d (or /D) flag was given to
3243 pcretest, it used to include the size of the study block as part of its
3244 output. Unfortunately, the structure contains a field that has a different
3245 size on different hardware architectures. This meant that the tests that
3246 showed this size failed. As the block is currently always of a fixed size,
3247 this information isn't actually particularly useful in pcretest output, so
3248 I have just removed it.
3249
3250 8. Three pre-processor statements accidentally did not start in column 1.
3251 Sadly, there are *still* compilers around that complain, even though
3252 standard C has not required this for well over a decade. Sigh.
3253
3254 9. In pcretest, the code for checking callouts passed small integers in the
3255 callout_data field, which is a void * field. However, some picky compilers
3256 complained about the casts involved for this on 64-bit systems. Now
3257 pcretest passes the address of the small integer instead, which should get
3258 rid of the warnings.
3259
3260 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
3261 both compile and run time, and gives an error if an invalid UTF-8 sequence
3262 is found. There is a option for disabling this check in cases where the
3263 string is known to be correct and/or the maximum performance is wanted.
3264
3265 11. In response to a bug report, I changed one line in Makefile.in from
3266
3267 -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
3268 to
3269 -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
3270
3271 to look similar to other lines, but I have no way of telling whether this
3272 is the right thing to do, as I do not use Windows. No doubt I'll get told
3273 if it's wrong...
3274
3275
3276 Version 4.3 21-May-03
3277 ---------------------
3278
3279 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
3280 Makefile.
3281
3282 2. Some refactoring to improve the quality of the code:
3283
3284 (i) The utf8_table... variables are now declared "const".
3285
3286 (ii) The code for \cx, which used the "case flipping" table to upper case
3287 lower case letters, now just substracts 32. This is ASCII-specific,
3288 but the whole concept of \cx is ASCII-specific, so it seems
3289 reasonable.
3290
3291 (iii) PCRE was using its character types table to recognize decimal and
3292 hexadecimal digits in the pattern. This is silly, because it handles
3293 only 0-9, a-f, and A-F, but the character types table is locale-
3294 specific, which means strange things might happen. A private
3295 table is now used for this - though it costs 256 bytes, a table is
3296 much faster than multiple explicit tests. Of course, the standard
3297 character types table is still used for matching digits in subject
3298 strings against \d.
3299
3300 (iv) Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
3301 ending in _t are). So I've renamed it as ESC_tee.
3302
3303 3. The first argument for regexec() in the POSIX wrapper should have been
3304 defined as "const".
3305
3306 4. Changed pcretest to use malloc() for its buffers so that they can be
3307 Electric Fenced for debugging.
3308
3309 5. There were several places in the code where, in UTF-8 mode, PCRE would try
3310 to read one or more bytes before the start of the subject string. Often this
3311 had no effect on PCRE's behaviour, but in some circumstances it could
3312 provoke a segmentation fault.
3313
3314 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
3315 to try to read one or more bytes before the start of the subject string.
3316
3317 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
3318 UTF-8 support could misbehave in various ways if the subject string
3319 contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
3320 area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
3321 back over UTF-8 characters.)
3322
3323
3324 Version 4.2 14-Apr-03
3325 ---------------------
3326
3327 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
3328
3329 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
3330 [ON_WINDOWS]: new variable, "#" on non-Windows platforms
3331 [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
3332 [WIN_PREFIX]: new variable, "cyg" for Cygwin
3333 * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
3334 and BUILD_EXEEXT
3335 Note: automatic setting of the BUILD variables is not yet working
3336 set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
3337 compile-time but not at link-time
3338 [LINK]: use for linking executables only
3339 make different versions for Windows and non-Windows
3340 [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
3341 libraries
3342 [LINK_FOR_BUILD]: new variable
3343 [OBJEXT]: use throughout
3344 [EXEEXT]: use throughout
3345 <winshared>: new target
3346 <wininstall>: new target
3347 <dftables.o>: use native compiler
3348 <dftables>: use native linker
3349 <install>: handle Windows platform correctly
3350 <clean>: ditto
3351 <check>: ditto
3352 copy DLL to top builddir before testing
3353
3354 As part of these changes, -no-undefined was removed again. This was reported
3355 to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
3356 in any case.
3357
3358 3. Some tidies to get rid of compiler warnings:
3359
3360 . In the match_data structure, match_limit was an unsigned long int, whereas
3361 match_call_count was an int. I've made them both unsigned long ints.
3362
3363 . In pcretest the fact that a const uschar * doesn't automatically cast to
3364 a void * provoked a warning.
3365
3366 . Turning on some more compiler warnings threw up some "shadow" variables
3367 and a few more missing casts.
3368
3369 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
3370 option, a class that contained a single character with a value between 128
3371 and 255 (e.g. /[\xFF]/) caused PCRE to crash.
3372
3373 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
3374 option, a class that contained several characters, but with at least one
3375 whose value was between 128 and 255 caused PCRE to crash.
3376
3377
3378 Version 4.1 12-Mar-03
3379 ---------------------
3380
3381 1. Compiling with gcc -pedantic found a couple of places where casts were
3382 needed, and a string in dftables.c that was longer than standard compilers are
3383 required to support.
3384
3385 2. Compiling with Sun's compiler found a few more places where the code could
3386 be tidied up in order to avoid warnings.
3387
3388 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
3389 first of these names is deprecated in the latest Autoconf in favour of the name
3390 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
3391 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
3392 analogy I have changed it to CFLAGS_FOR_BUILD.
3393
3394 4. Added -no-undefined to the linking command in the Makefile, because this is
3395 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
3396 linking step for the pcreposix library.
3397
3398 5. PCRE was failing to diagnose the case of two named groups with the same
3399 name.
3400
3401 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
3402 literal character that is needed in the subject for a match, and scans along to
3403 ensure that it is present before embarking on the full matching process. This
3404 saves time in cases of nested unlimited repeats that are never going to match.
3405 Problem: the scan can take a lot of time if the subject is very long (e.g.
3406 megabytes), thus penalizing straightforward matches. It is now done only if the
3407 amount of subject to be scanned is less than 1000 bytes.
3408
3409 7. A lesser problem with the same optimization is that it was recording the
3410 first character of an anchored pattern as "needed", thus provoking a search
3411 right along the subject, even when the first match of the pattern was going to
3412 fail. The "needed" character is now not set for anchored patterns, unless it
3413 follows something in the pattern that is of non-fixed length. Thus, it still
3414 fulfils its original purpose of finding quick non-matches in cases of nested
3415 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
3416
3417
3418 Version 4.0 17-Feb-03
3419 ---------------------
3420
3421 1. If a comment in an extended regex that started immediately after a meta-item
3422 extended to the end of string, PCRE compiled incorrect data. This could lead to
3423 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
3424
3425 2. Moved to autoconf 2.53 and libtool 1.4.2.
3426
3427 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
3428 the special perltest8 script is no longer needed - all the tests can be run
3429 from a single perltest script.
3430
3431 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
3432 by \s. It has now been removed in PCRE. This means it isn't recognized as
3433 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
3434 class [:space:] *does* include VT, thereby creating a mess.
3435
3436 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
3437 space and tab.
3438
3439 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
3440 its new features into the main test script, reducing the number of scripts.
3441
3442 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
3443 were backward compatible, and made the (?i) apply to the whole pattern, as if
3444 /i were given. Now it behaves more logically, and applies the option setting
3445 only to what follows. PCRE has been changed to follow suit. However, if it
3446 finds options settings right at the start of the pattern, it extracts them into
3447 the global options, as before. Thus, they show up in the info data.
3448
3449 8. Added support for the \Q...\E escape sequence. Characters in between are
3450 treated as literals. This is slightly different from Perl in that $ and @ are
3451 also handled as literals inside the quotes. In Perl, they will cause variable
3452 interpolation. Note the following examples:
3453
3454 Pattern PCRE matches Perl matches
3455
3456 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
3457 \Qabc\$xyz\E abc\$xyz abc\$xyz
3458 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
3459
3460 For compatibility with Perl, \Q...\E sequences are recognized inside character
3461 classes as well as outside them.
3462
3463 9. Re-organized 3 code statements in pcretest to avoid "overflow in
3464 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
3465 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
3466 signed/unsigned warnings.
3467
3468 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
3469 option for pcretest, so I've replaced it by a simple function that does just
3470 that job.
3471
3472 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
3473 "pcregrep -".
3474
3475 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
3476 Java package. This provides some syntactic sugar for simple cases of what my
3477 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
3478 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
3479 item, you can use this simplified notation. Note that only makes sense with
3480 greedy quantifiers. Consequently, the use of the possessive quantifier forces
3481 greediness, whatever the setting of the PCRE_UNGREEDY option.
3482
3483 13. A change of greediness default within a pattern was not taking effect at
3484 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
3485 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
3486 was abstracted outside.
3487
3488 14. PCRE now supports the \G assertion. It is true when the current matching
3489 position is at the start point of the match. This differs from \A when the
3490 starting offset is non-zero. Used with the /g option of pcretest (or similar
3491 code), it works in the same way as it does for Perl's /g option. If all
3492 alternatives of a regex begin with \G, the expression is anchored to the start
3493 match position, and the "anchored" flag is set in the compiled expression.
3494
3495 15. Some bugs concerning the handling of certain option changes within patterns
3496 have been fixed. These applied to options other than (?ims). For example,
3497 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
3498 been the other way round. Some of this was related to change 7 above.
3499
3500 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
3501 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
3502 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
3503 POSIX classes only within a class (e.g. /[[:alpha:]]/).
3504
3505 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
3506 mode. Unlike ".", it always matches newline, whatever the setting of
3507 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
3508 assertions. Perl allows it, but it doesn't (in general) work because it can't
3509 calculate the length of the lookbehind. At least, that's the case for Perl
3510 5.8.0 - I've been told they are going to document that it doesn't work in
3511 future.
3512
3513 18. Added an error diagnosis for escapes that PCRE does not support: these are
3514 \L, \l, \N, \P, \p, \U, \u, and \X.
3515
3516 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
3517 reading past the end of the pattern in cases such as /[abcd/.
3518
3519 20. PCRE was getting more memory than necessary for patterns with classes that
3520 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
3521
3522 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
3523 compiling PCRE for use with Virtual Pascal.
3524
3525 22. Small fix to the Makefile to make it work properly if the build is done
3526 outside the source tree.
3527
3528 23. Added a new extension: a condition to go with recursion. If a conditional
3529 subpattern starts with (?(R) the "true" branch is used if recursion has
3530 happened, whereas the "false" branch is used only at the top level.
3531
3532 24. When there was a very long string of literal characters (over 255 bytes
3533 without UTF support, over 250 bytes with UTF support), the computation of how
3534 much memory was required could be incorrect, leading to segfaults or other
3535 strange effects.
3536
3537 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
3538 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
3539 there was a subsequent back reference to those brackets. This meant that, for
3540 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
3541 possible to check for precisely this case. All we can do is abandon the
3542 optimization if .* occurs inside capturing brackets when there are any back
3543 references whatsoever. (See below for a better fix that came later.)
3544
3545 26. The handling of the optimization for finding the first character of a
3546 non-anchored pattern, and for finding a character that is required later in the
3547 match were failing in some cases. This didn't break the matching; it just
3548 failed to optimize when it could. The way this is done has been re-implemented.
3549
3550 27. Fixed typo in error message for invalid (?R item (it said "(?p").
3551
3552 28. Added a new feature that provides some of the functionality that Perl
3553 provides with (?{...}). The facility is termed a "callout". The way it is done
3554 in PCRE is for the caller to provide an optional function, by setting
3555 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
3556 global variable. By default it is unset, which disables all calling out. To get
3557 the function called, the regex must include (?C) at appropriate points. This
3558 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
3559 This provides a means of identifying different callout points. When PCRE
3560 reaches such a point in the regex, if pcre_callout has been set, the external
3561 function is called. It is provided with data in a structure called
3562 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
3563 matching continues; if it returns a non-zero value, the match at the current
3564 point fails. However, backtracking will occur if possible. [This was changed
3565 later and other features added - see item 49 below.]
3566
3567 29. pcretest is upgraded to test the callout functionality. It provides a
3568 callout function that displays information. By default, it shows the start of
3569 the match and the current position in the text. There are some new data escapes
3570 to vary what happens:
3571
3572 \C+ in addition, show current contents of captured substrings
3573 \C- do not supply a callout function
3574 \C!n return 1 when callout number n is reached
3575 \C!n!m return 1 when callout number n is reached for the mth time
3576
3577 30. If pcregrep was called with the -l option and just a single file name, it
3578 output "<stdin>" if a match was found, instead of the file name.
3579
3580 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
3581 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
3582 pcre_exec(). This saves a malloc/free per call. The default value of
3583 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
3584 when configuring.
3585
3586 32. The default maximum size of a compiled pattern is 64K. There have been a
3587 few cases of people hitting this limit. The code now uses macros to handle the
3588 storing of links as offsets within the compiled pattern. It defaults to 2-byte
3589 links, but this can be changed to 3 or 4 bytes by --with-link-size when
3590 configuring. Tests 2 and 5 work only with 2-byte links because they output
3591 debugging information about compiled patterns.
3592
3593 33. Internal code re-arrangements:
3594
3595 (a) Moved the debugging function for printing out a compiled regex into
3596 its own source file (printint.c) and used #include to pull it into
3597 pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
3598 separate copies.
3599
3600 (b) Defined the list of op-code names for debugging as a macro in
3601 internal.h so that it is next to the definition of the opcodes.
3602
3603 (c) Defined a table of op-code lengths for simpler skipping along compiled
3604 code. This is again a macro in internal.h so that it is next to the
3605 definition of the opcodes.
3606
3607 34. Added support for recursive calls to individual subpatterns, along the
3608 lines of Robin Houston's patch (but implemented somewhat differently).
3609
3610 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
3611 allow it to read and process whole directories in Win32. This code was
3612 contributed by Lionel Fourquaux; it has not been tested by me.
3613
3614 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
3615 used to name a group. Names consist of alphanumerics and underscores, and must
3616 be unique. Back references use the syntax (?P=name) and recursive calls use
3617 (?P>name) which is a PCRE extension to the Python extension. Groups still have
3618 numbers. The function pcre_fullinfo() can be used after compilation to extract
3619 a name/number map. There are three relevant calls:
3620
3621 PCRE_INFO_NAMEENTRYSIZE yields the size of each entry in the map
3622 PCRE_INFO_NAMECOUNT yields the number of entries
3623 PCRE_INFO_NAMETABLE yields a pointer to the map.
3624
3625 The map is a vector of fixed-size entries. The size of each entry depends on
3626 the length of the longest name used. The first two bytes of each entry are the
3627 group number, most significant byte first. There follows the corresponding
3628 name, zero terminated. The names are in alphabetical order.
3629
3630 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
3631 case instead of 255. Making it the same both with and without UTF-8 support
3632 means that the same test output works with both.
3633
3634 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
3635 calling malloc() with a zero argument.
3636
3637 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
3638 optimization. I've improved things by keeping a bitmap of backreferences with
3639 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
3640 fact referenced, the optimization can be applied. It is unlikely that a
3641 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
3642 the match to follow \n) will appear inside brackets with a number greater than
3643 31, but if it does, any back reference > 31 suppresses the optimization.
3644
3645 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
3646 of disabling numbered capturing parentheses. Any opening parenthesis that is
3647 not followed by ? behaves as if it were followed by ?: but named parentheses
3648 can still be used for capturing (and they will acquire numbers in the usual
3649 way).
3650
3651 41. Redesigned the return codes from the match() function into yes/no/error so
3652 that errors can be passed back from deep inside the nested calls. A malloc
3653 failure while inside a recursive subpattern call now causes the
3654 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
3655
3656 42. It is now possible to set a limit on the number of times the match()
3657 function is called in a call to pcre_exec(). This facility makes it possible to
3658 limit the amount of recursion and backtracking, though not in a directly
3659 obvious way, because the match() function is used in a number of different
3660 circumstances. The count starts from zero for each position in the subject
3661 string (for non-anchored patterns). The default limit is, for compatibility, a
3662 large number, namely 10 000 000. You can change this in two ways:
3663
3664 (a) When configuring PCRE before making, you can use --with-match-limit=n
3665 to set a default value for the compiled library.
3666
3667 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
3668 a different value is set. See 45 below.
3669
3670 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
3671
3672 43. Added a new function pcre_config(int, void *) to enable run-time extraction
3673 of things that can be changed at compile time. The first argument specifies
3674 what is wanted and the second points to where the information is to be placed.
3675 The current list of available information is:
3676
3677 PCRE_CONFIG_UTF8
3678
3679 The output is an integer that is set to one if UTF-8 support is available;
3680 otherwise it is set to zero.
3681
3682 PCRE_CONFIG_NEWLINE
3683
3684 The output is an integer that it set to the value of the code that is used for
3685 newline. It is either LF (10) or CR (13).
3686
3687 PCRE_CONFIG_LINK_SIZE
3688
3689 The output is an integer that contains the number of bytes used for internal
3690 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
3691
3692 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
3693
3694 The output is an integer that contains the threshold above which the POSIX
3695 interface uses malloc() for output vectors. See item 31 above.
3696
3697 PCRE_CONFIG_MATCH_LIMIT
3698
3699 The output is an unsigned integer that contains the default limit of the number
3700 of match() calls in a pcre_exec() execution. See 42 above.
3701
3702 44. pcretest has been upgraded by the addition of the -C option. This causes it
3703 to extract all the available output from the new pcre_config() function, and to
3704 output it. The program then exits immediately.
3705
3706 45. A need has arisen to pass over additional data with calls to pcre_exec() in
3707 order to support additional features. One way would have been to define
3708 pcre_exec2() (for example) with extra arguments, but this would not have been
3709 extensible, and would also have required all calls to the original function to
3710 be mapped to the new one. Instead, I have chosen to extend the mechanism that
3711 is used for passing in "extra" data from pcre_study().
3712
3713 The pcre_extra structure is now exposed and defined in pcre.h. It currently
3714 contains the following fields:
3715
3716 flags a bitmap indicating which of the following fields are set
3717 study_data opaque data from pcre_study()
3718 match_limit a way of specifying a limit on match() calls for a specific
3719 call to pcre_exec()
3720 callout_data data for callouts (see 49 below)
3721
3722 The flag bits are also defined in pcre.h, and are
3723
3724 PCRE_EXTRA_STUDY_DATA
3725 PCRE_EXTRA_MATCH_LIMIT
3726 PCRE_EXTRA_CALLOUT_DATA
3727
3728 The pcre_study() function now returns one of these new pcre_extra blocks, with
3729 the actual study data pointed to by the study_data field, and the
3730 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
3731 before. That is, this change is entirely upwards-compatible and requires no
3732 change to existing code.
3733
3734 If you want to pass in additional data to pcre_exec(), you can either place it
3735 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
3736 block.
3737
3738 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
3739 data string contains the escape sequence \M, pcretest calls pcre_exec() several
3740 times with different match limits, until it finds the minimum value needed for
3741 pcre_exec() to complete. The value is then output. This can be instructive; for
3742 most simple matches the number is quite small, but for pathological cases it
3743 gets very large very quickly.
3744
3745 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
3746 returns the size of the data block pointed to by the study_data field in a
3747 pcre_extra block, that is, the value that was passed as the argument to
3748 pcre_malloc() when PCRE was getting memory in which to place the information
3749 created by pcre_study(). The fourth argument should point to a size_t variable.
3750 pcretest has been extended so that this information is shown after a successful
3751 pcre_study() call when information about the compiled regex is being displayed.
3752
3753 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
3754 because what follows is always an absolute path. (Later: it turns out that this
3755 is more than cosmetic for MinGW, because it doesn't like empty path
3756 components.)
3757
3758 49. Some changes have been made to the callout feature (see 28 above):
3759
3760 (i) A callout function now has three choices for what it returns:
3761
3762 0 => success, carry on matching
3763 > 0 => failure at this point, but backtrack if possible
3764 < 0 => serious error, return this value from pcre_exec()
3765
3766 Negative values should normally be chosen from the set of PCRE_ERROR_xxx
3767 values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
3768 "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
3769 use by callout functions. It will never be used by PCRE itself.
3770
3771 (ii) The pcre_extra structure (see 45 above) has a void * field called
3772 callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
3773 pcre_callout_block structure has a field of the same name. The contents of
3774 the field passed in the pcre_extra structure are passed to the callout
3775 function in the corresponding field in the callout block. This makes it
3776 easier to use the same callout-containing regex from multiple threads. For
3777 testing, the pcretest program has a new data escape
3778
3779 \C*n pass the number n (may be negative) as callout_data
3780
3781 If the callout function in pcretest receives a non-zero value as
3782 callout_data, it returns that value.
3783
3784 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
3785 there were some redundant $(CFLAGS) in commands that are now specified as
3786 $(LINK), which already includes $(CFLAGS).
3787
3788 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
3789 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
3790 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
3791 one-byte characters throughout. Note that case-insensitive matching applies
3792 only to characters whose values are less than 256. PCRE doesn't support the
3793 notion of cases for higher-valued characters.
3794
3795 (i) A character class whose characters are all within 0-255 is handled as
3796 a bit map, and the map is inverted for negative classes. Previously, a
3797 character > 255 always failed to match such a class; however it should
3798 match if the class was a negative one (e.g. [^ab]). This has been fixed.
3799
3800 (ii) A negated character class with a single character < 255 is coded as
3801 "not this character" (OP_NOT). This wasn't working properly when the test
3802 character was multibyte, either singly or repeated.
3803
3804 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
3805 mode, for example: \x{100}{2,3}.
3806
3807 (iv) The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
3808 singly or repeated) now correctly test multibyte characters. However,
3809 PCRE doesn't recognize any characters with values greater than 255 as
3810 digits, spaces, or word characters. Such characters always match \D, \S,
3811 and \W, and never match \d, \s, or \w.
3812
3813 (v) Classes may now contain characters and character ranges with values
3814 greater than 255. For example: [ab\x{100}-\x{400}].
3815
3816 (vi) pcregrep now has a --utf-8 option (synonym -u) which makes it call
3817 PCRE in UTF-8 mode.
3818
3819 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
3820 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
3821 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
3822 value.)
3823
3824 53. The single man page has become too large. I have therefore split it up into
3825 a number of separate man pages. These also give rise to individual HTML pages;
3826 these are now put in a separate directory, and there is an index.html page that
3827 lists them all. Some hyperlinking between the pages has been installed.
3828
3829 54. Added convenience functions for handling named capturing parentheses.
3830
3831 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
3832 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
3833 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
3834 are faulted.
3835
3836 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
3837 calling configure. These values are used when compiling the dftables.c program
3838 which is run to generate the source of the default character tables. They
3839 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
3840 you will need to set these values.
3841
3842 57. Updated the building process for Windows DLL, as provided by Fred Cox.
3843
3844
3845 Version 3.9 02-Jan-02
3846 ---------------------
3847
3848 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
3849
3850 2. If --disable-static was given, the building process failed when trying to
3851 build pcretest and pcregrep. (For some reason it was using libtool to compile
3852 them, which is not right, as they aren't part of the library.)
3853
3854
3855 Version 3.8 18-Dec-01
3856 ---------------------
3857
3858 1. The experimental UTF-8 code was completely screwed up. It was packing the
3859 bytes in the wrong order. How dumb can you get?
3860
3861
3862 Version 3.7 29-Oct-01
3863 ---------------------
3864
3865 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
3866 This caused pcretest, when used on the test data, to segfault. Unfortunately,
3867 this didn't happen under Solaris 8, where I normally test things.
3868
3869 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
3870 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
3871 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
3872 this fix an hour or so after the initial 3.7 release.)
3873
3874
3875 Version 3.6 23-Oct-01
3876 ---------------------
3877
3878 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
3879 offsets passed as NULL with zero offset count.
3880
3881 2. The config.guess and config.sub files had not been updated when I moved to
3882 the latest autoconf.
3883
3884
3885 Version 3.5 15-Aug-01
3886 ---------------------
3887
3888 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
3889 had been forgotten.
3890
3891 2. By using declared but undefined structures, we can avoid using "void"
3892 definitions in pcre.h while keeping the internal definitions of the structures
3893 private.
3894
3895 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
3896 user point of view, this means that both static and shared libraries are built
3897 by default, but this can be individually controlled. More of the work of
3898 handling this static/shared cases is now inside libtool instead of PCRE's make
3899 file.
3900
3901 4. The pcretest utility is now installed along with pcregrep because it is
3902 useful for users (to test regexs) and by doing this, it automatically gets
3903 relinked by libtool. The documentation has been turned into a man page, so
3904 there are now .1, .txt, and .html versions in /doc.
3905
3906 5. Upgrades to pcregrep:
3907 (i) Added long-form option names like gnu grep.
3908 (ii) Added --help to list all options with an explanatory phrase.
3909 (iii) Added -r, --recursive to recurse into sub-directories.
3910 (iv) Added -f, --file to read patterns from a file.
3911
3912 6. pcre_exec() was referring to its "code" argument before testing that
3913 argument for NULL (and giving an error if it was NULL).
3914
3915 7. Upgraded Makefile.in to allow for compiling in a different directory from
3916 the source directory.
3917
3918 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
3919 options bits, the pointer it was passed was to an int instead of to an unsigned
3920 long int. This mattered only on 64-bit systems.
3921
3922 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
3923 generated) instead of pcre.in, which it its source. Also made the same change
3924 in several of the .c files.
3925
3926 10. A new release of gcc defines printf() as a macro, which broke pcretest
3927 because it had an ifdef in the middle of a string argument for printf(). Fixed
3928 by using separate calls to printf().
3929
3930 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
3931 script, to force use of CR or LF instead of \n in the source. On non-Unix
3932 systems, the value can be set in config.h.
3933
3934 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
3935 absolute limit. Changed the text of the error message to make this clear, and
3936 likewise updated the man page.
3937
3938 13. The limit of 99 on the number of capturing subpatterns has been removed.
3939 The new limit is 65535, which I hope will not be a "real" limit.
3940
3941
3942 Version 3.4 22-Aug-00
3943 ---------------------
3944
3945 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
3946
3947 2. Diagnose condition (?(0) as an error instead of crashing on matching.
3948
3949
3950 Version 3.3 01-Aug-00
3951 ---------------------
3952
3953 1. If an octal character was given, but the value was greater than \377, it
3954 was not getting masked to the least significant bits, as documented. This could
3955 lead to crashes in some systems.
3956
3957 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
3958 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
3959
3960 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
3961 These just pass their arguments on to (pcre_free)(), but they are provided
3962 because some uses of PCRE bind it to non-C systems that can call its functions,
3963 but cannot call free() or pcre_free() directly.
3964
3965 4. Add "make test" as a synonym for "make check". Corrected some comments in
3966 the Makefile.
3967
3968 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
3969 Makefile.
3970
3971 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
3972 command called pgrep for grepping around the active processes.
3973
3974 7. Added the beginnings of support for UTF-8 character strings.
3975
3976 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
3977 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
3978 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
3979 out for the ar command.)
3980
3981
3982 Version 3.2 12-May-00
3983 ---------------------
3984
3985 This is purely a bug fixing release.
3986
3987 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
3988 of ZA. This was just one example of several cases that could provoke this bug,
3989 which was introduced by change 9 of version 2.00. The code for breaking
3990 infinite loops after an iteration that matches an empty string was't working
3991 correctly.
3992
3993 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
3994 when matched against abbab (for example). After matching an empty string, it
3995 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
3996 caused it to match further down the string than it should.
3997
3998 3. The code contained an inclusion of sys/types.h. It isn't clear why this
3999 was there because it doesn't seem to be needed, and it causes trouble on some
4000 systems, as it is not a Standard C header. It has been removed.
4001
4002 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
4003 were reported on the Macintosh. The changes were from
4004
4005 while ((c = *(++ptr)) != 0 && c != '\n');
4006 to
4007 while ((c = *(++ptr)) != 0 && c != '\n') ;
4008
4009 Totally extraordinary, but if that's what it takes...
4010
4011 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
4012 available. Added HAVE_BCOPY and an autoconf test for it; if neither
4013 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
4014 assumes the way PCRE uses memmove() (always moving upwards).
4015
4016 6. PCRE is being used in one environment where strchr() is not available. There
4017 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
4018 faster code anyway.
4019
4020
4021 Version 3.1 09-Feb-00
4022 ---------------------
4023
4024 The only change in this release is the fixing of some bugs in Makefile.in for
4025 the "install" target:
4026
4027 (1) It was failing to install pcreposix.h.
4028
4029 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
4030
4031
4032 Version 3.0 01-Feb-00
4033 ---------------------
4034
4035 1. Add support for the /+ modifier to perltest (to output $` like it does in
4036 pcretest).
4037
4038 2. Add support for the /g modifier to perltest.
4039
4040 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
4041 matches null strings.
4042
4043 4. Fix perltest so that it doesn't do unwanted things when fed an empty
4044 pattern. Perl treats empty patterns specially - it reuses the most recent
4045 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
4046 effect.
4047
4048 5. The POSIX interface was broken in that it was just handing over the POSIX
4049 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
4050 required a bigger vector, with some working space on the end. This means that
4051 the POSIX wrapper now has to get and free some memory, and copy the results.
4052
4053 6. Added some simple autoconf support, placing the test data and the
4054 documentation in separate directories, re-organizing some of the
4055 information files, and making it build pcre-config (a GNU standard). Also added
4056 libtool support for building PCRE as a shared library, which is now the
4057 default.
4058
4059 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
4060 09 are not valid octal constants. Single digits will be used for minor values
4061 less than 10.
4062
4063 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
4064 existing programs that set these in the POSIX interface can use PCRE without
4065 modification.
4066
4067 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
4068 return all that pcre_info() returns, plus additional data. The pcre_info()
4069 function is retained for compatibility, but is considered to be obsolete.
4070
4071 10. Added experimental recursion feature (?R) to handle one common case that
4072 Perl 5.6 will be able to do with (?p{...}).
4073
4074 11. Added support for POSIX character classes like [:alpha:], which Perl is
4075 adopting.
4076
4077
4078 Version 2.08 31-Aug-99
4079 ----------------------
4080
4081 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
4082 trying to match at the startoffset position, but instead was moving forward to
4083 the next newline as if a previous match had failed.
4084
4085 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
4086 and could get into a loop if a null string was matched other than at the start
4087 of the subject.
4088
4089 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
4090 be distinguished at compile time, and for completeness also added PCRE_DATE.
4091
4092 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
4093 in GnuWin32 environments.
4094
4095
4096 Version 2.07 29-Jul-99
4097 ----------------------
4098
4099 1. The documentation is now supplied in plain text form and HTML as well as in
4100 the form of man page sources.
4101
4102 2. C++ compilers don't like assigning (void *) values to other pointer types.
4103 In particular this affects malloc(). Although there is no problem in Standard
4104 C, I've put in casts to keep C++ compilers happy.
4105
4106 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
4107 should be (const char *).
4108
4109 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
4110 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
4111 However, I haven't made this a standard facility. The documentation doesn't
4112 mention it, and the Makefile doesn't support it.
4113
4114 5. The Makefile now contains an "install" target, with editable destinations at
4115 the top of the file. The pcretest program is not installed.
4116
4117 6. pgrep -V now gives the PCRE version number and date.
4118
4119 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
4120 causing the entire string to be ignored, instead of just the last character.
4121
4122 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
4123 non-matching string, it can take a very, very long time, even for strings of
4124 quite modest length, because of the nested recursion. PCRE now does better in
4125 some of these cases. It does this by remembering the last required literal
4126 character in the pattern, and pre-searching the subject to ensure it is present
4127 before running the real match. In other words, it applies a heuristic to detect
4128 some types of certain failure quickly, and in the above example, if presented
4129 with a string that has no trailing " it gives "no match" very quickly.
4130
4131 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
4132 other alternatives are tried instead.
4133
4134
4135 Version 2.06 09-Jun-99
4136 ----------------------
4137
4138 1. Change pcretest's output for amount of store used to show just the code
4139 space, because the remainder (the data block) varies in size between 32-bit and
4140 64-bit systems.
4141
4142 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
4143 start matching at. This allows lookbehinds to work when searching for multiple
4144 occurrences in a string.
4145
4146 3. Added additional options to pcretest for testing multiple occurrences:
4147
4148 /+ outputs the rest of the string that follows a match
4149 /g loops for multiple occurrences, using the new startoffset argument
4150 /G loops for multiple occurrences by passing an incremented pointer
4151
4152 4. PCRE wasn't doing the "first character" optimization for patterns starting
4153 with \b or \B, though it was doing it for other lookbehind assertions. That is,
4154 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
4155 the letter 'x'. On long subject strings, this gives a significant speed-up.
4156
4157
4158 Version 2.05 21-Apr-99
4159 ----------------------
4160
4161 1. Changed the type of magic_number from int to long int so that it works
4162 properly on 16-bit systems.
4163
4164 2. Fixed a bug which caused patterns starting with .* not to work correctly
4165 when the subject string contained newline characters. PCRE was assuming
4166 anchoring for such patterns in all cases, which is not correct because .* will
4167 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
4168 DOTALL is set at top level; otherwise it knows that patterns starting with .*
4169 must be retried after every newline in the subject.
4170
4171
4172 Version 2.04 18-Feb-99
4173 ----------------------
4174
4175 1. For parenthesized subpatterns with repeats whose minimum was zero, the
4176 computation of the store needed to hold the pattern was incorrect (too large).
4177 If such patterns were nested a few deep, this could multiply and become a real
4178 problem.
4179
4180 2. Added /M option to pcretest to show the memory requirement of a specific
4181 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
4182
4183 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
4184 compiled in such a way that the backtracking after subsequent failure was
4185 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
4186 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
4187
4188
4189 Version 2.03 02-Feb-99
4190 ----------------------
4191
4192 1. Fixed typo and small mistake in man page.
4193
4194 2. Added 4th condition (GPL supersedes if conflict) and created separate
4195 LICENCE file containing the conditions.
4196
4197 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
4198 Perl, that is the internal \ allows the delimiter to be included in the
4199 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
4200 the final delimiter, add \ to the end of the pattern (to test the error).
4201
4202 4. Added the convenience functions for extracting substrings after a successful
4203 match. Updated pcretest to make it able to test these functions.
4204
4205
4206 Version 2.02 14-Jan-99
4207 ----------------------
4208
4209 1. Initialized the working variables associated with each extraction so that
4210 their saving and restoring doesn't refer to uninitialized store.
4211
4212 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
4213 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
4214 fix the problem.
4215
4216 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
4217 calls, and wasn't printing the correct value for compiling calls. Increased the
4218 default value of LOOPREPEAT, and the number of significant figures in the
4219 times.
4220
4221 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
4222
4223 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
4224 a building problem on Windows NT with a FAT file system.
4225
4226
4227 Version 2.01 21-Oct-98
4228 ----------------------
4229
4230 1. Changed the API for pcre_compile() to allow for the provision of a pointer
4231 to character tables built by pcre_maketables() in the current locale. If NULL
4232 is passed, the default tables are used.
4233
4234
4235 Version 2.00 24-Sep-98
4236 ----------------------
4237
4238 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
4239 it any more.
4240
4241 2. Allow quantification of (?>) groups, and make it work correctly.
4242
4243 3. The first character computation wasn't working for (?>) groups.
4244
4245 4. Correct the implementation of \Z (it is permitted to match on the \n at the
4246 end of the subject) and add 5.005's \z, which really does match only at the
4247 very end of the subject.
4248
4249 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
4250
4251 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
4252 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
4253 localized options. All options to pcre_study() were also removed.
4254
4255 7. Add other new features from 5.005:
4256
4257 $(?<= positive lookbehind
4258 $(?<! negative lookbehind
4259 (?imsx-imsx) added the unsetting capability
4260 such a setting is global if at outer level; local otherwise
4261 (?imsx-imsx:) non-capturing groups with option setting
4262 (?(cond)re|re) conditional pattern matching
4263
4264 A backreference to itself in a repeated group matches the previous
4265 captured string.
4266
4267 8. General tidying up of studying (both automatic and via "study")
4268 consequential on the addition of new assertions.
4269
4270 9. As in 5.005, unlimited repeated groups that could match an empty substring
4271 are no longer faulted at compile time. Instead, the loop is forcibly broken at
4272 runtime if any iteration does actually match an empty substring.
4273
4274 10. Include the RunTest script in the distribution.
4275
4276 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
4277 discrepancies, some of which were old and were also with respect to 5.004. They
4278 have now been fixed.
4279
4280
4281 Version 1.09 28-Apr-98
4282 ----------------------
4283
4284 1. A negated single character class followed by a quantifier with a minimum
4285 value of one (e.g. [^x]{1,6} ) was not compiled correctly. This could lead to
4286 program crashes, or just wrong answers. This did not apply to negated classes
4287 containing more than one character, or to minima other than one.
4288
4289
4290 Version 1.08 27-Mar-98
4291 ----------------------
4292
4293 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
4294
4295 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
4296 latter must appear before anything that relies on it in the pattern.
4297
4298
4299 Version 1.07 16-Feb-98
4300 ----------------------
4301
4302 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
4303 repeat of a potentially empty string).
4304
4305
4306 Version 1.06 23-Jan-98
4307 ----------------------
4308
4309 1. Added Markus Oberhumer's little patches for C++.
4310
4311 2. Literal strings longer than 255 characters were broken.
4312
4313
4314 Version 1.05 23-Dec-97
4315 ----------------------
4316
4317 1. Negated character classes containing more than one character were failing if
4318 PCRE_CASELESS was set at run time.
4319
4320
4321 Version 1.04 19-Dec-97
4322 ----------------------
4323
4324 1. Corrected the man page, where some "const" qualifiers had been omitted.
4325
4326 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
4327 input syntax.
4328
4329 3. Fixed memory leak which occurred when a regex with back references was
4330 matched with an offsets vector that wasn't big enough. The temporary memory
4331 that is used in this case wasn't being freed if the match failed.
4332
4333 4. Tidied pcretest to ensure it frees memory that it gets.
4334
4335 5. Temporary memory was being obtained in the case where the passed offsets
4336 vector was exactly big enough.
4337
4338 6. Corrected definition of offsetof() from change 5 below.
4339
4340 7. I had screwed up change 6 below and broken the rules for the use of
4341 setjmp(). Now fixed.
4342
4343
4344 Version 1.03 18-Dec-97
4345 ----------------------
4346
4347 1. A erroneous regex with a missing opening parenthesis was correctly
4348 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
4349 on some systems.
4350
4351 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
4352 it was reported that one broken compiler failed on the former because "code" is
4353 also an independent variable.
4354
4355 3. The erroneous regex a[]b caused an array overrun reference.
4356
4357 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
4358 fail on data ending with that character. (It was going on too far, and checking
4359 the next character, typically a binary zero.) This was specific to the
4360 optimized code for single-character negative classes.
4361
4362 5. Added a contributed patch from the TIN world which does the following:
4363
4364 + Add an undef for memmove, in case the the system defines a macro for it.
4365
4366 + Add a definition of offsetof(), in case there isn't one. (I don't know
4367 the reason behind this - offsetof() is part of the ANSI standard - but
4368 it does no harm).
4369
4370 + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
4371 most of the places where whitespace preceded '#'. I have given up and
4372 allowed the remaining 2 cases to be at the margin.
4373
4374 + Rename some variables in pcre to eliminate shadowing. This seems very
4375 pedantic, but does no harm, of course.
4376
4377 6. Moved the call to setjmp() into its own function, to get rid of warnings
4378 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
4379
4380 7. Constructs such as \d{8,} were compiling into the equivalent of
4381 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
4382 outcome, but in this particular case used more store than had been allocated,
4383 which caused the bug to be discovered because it threw up an internal error.
4384
4385 8. The debugging code in both pcre and pcretest for outputting the compiled
4386 form of a regex was going wrong in the case of back references followed by
4387 curly-bracketed repeats.
4388
4389
4390 Version 1.02 12-Dec-97
4391 ----------------------
4392
4393 1. Typos in pcre.3 and comments in the source fixed.
4394
4395 2. Applied a contributed patch to get rid of places where it used to remove
4396 'const' from variables, and fixed some signed/unsigned and uninitialized
4397 variable warnings.
4398
4399 3. Added the "runtest" target to Makefile.
4400
4401 4. Set default compiler flag to -O2 rather than just -O.
4402
4403
4404 Version 1.01 19-Nov-97
4405 ----------------------
4406
4407 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
4408 like /([ab]*)*/, that is, for classes with more than one character in them.
4409
4410 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
4411 as /((?>a*))*/ (a PCRE_EXTRA facility).
4412
4413
4414 Version 1.00 18-Nov-97
4415 ----------------------
4416
4417 1. Added compile-time macros to support systems such as SunOS4 which don't have
4418 memmove() or strerror() but have other things that can be used instead.
4419
4420 2. Arranged that "make clean" removes the executables.
4421
4422
4423 Version 0.99 27-Oct-97
4424 ----------------------
4425
4426 1. Fixed bug in code for optimizing classes with only one character. It was
4427 initializing a 32-byte map regardless, which could cause it to run off the end
4428 of the memory it had got.
4429
4430 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
4431
4432
4433 Version 0.98 22-Oct-97
4434 ----------------------
4435
4436 1. Fixed bug in code for handling temporary memory usage when there are more
4437 back references than supplied space in the ovector. This could cause segfaults.
4438
4439
4440 Version 0.97 21-Oct-97
4441 ----------------------
4442
4443 1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
4444
4445 2. Optimized negated single characters not to use a bit map.
4446
4447 3. Brought error texts together as macro definitions; clarified some of them;
4448 fixed one that was wrong - it said "range out of order" when it meant "invalid
4449 escape sequence".
4450
4451 4. Changed some char * arguments to const char *.
4452
4453 5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
4454
4455 6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
4456 pcretest.
4457
4458
4459 Version 0.96 16-Oct-97
4460 ----------------------
4461
4462 1. Added a simple "pgrep" utility to the distribution.
4463
4464 2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
4465 unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
4466 where "ddd" means "one or more decimal digits".
4467
4468 3. Fixed serious bug. If a pattern had a back reference, but the call to
4469 pcre_exec() didn't supply a large enough ovector to record the related
4470 identifying subpattern, the match always failed. PCRE now remembers the number
4471 of the largest back reference, and gets some temporary memory in which to save
4472 the offsets during matching if necessary, in order to ensure that
4473 backreferences always work.
4474
4475 4. Increased the compatibility with Perl in a number of ways:
4476
4477 (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
4478 to request this handling. The option can be set at compile or exec time.
4479
4480 (b) $ matches before a terminating newline by default; an option
4481 PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
4482 mode). The option can be set at compile or exec time.
4483
4484 (c) The handling of \ followed by a digit other than 0 is now supposed to be
4485 the same as Perl's. If the decimal number it represents is less than 10
4486 or there aren't that many previous left capturing parentheses, an octal
4487 escape is read. Inside a character class, it's always an octal escape,
4488 even if it is a single digit.
4489
4490 (d) An escaped but undefined alphabetic character is taken as a literal,
4491 unless PCRE_EXTRA is set. Currently this just reserves the remaining
4492 escapes.
4493
4494 (e) {0} is now permitted. (The previous item is removed from the compiled
4495 pattern).
4496
4497 5. Changed all the names of code files so that the basic parts are no longer
4498 than 10 characters, and abolished the teeny "globals.c" file.
4499
4500 6. Changed the handling of character classes; they are now done with a 32-byte
4501 bit map always.
4502
4503 7. Added the -d and /D options to pcretest to make it possible to look at the
4504 internals of compilation without having to recompile pcre.
4505
4506
4507 Version 0.95 23-Sep-97
4508 ----------------------
4509
4510 1. Fixed bug in pre-pass concerning escaped "normal" characters such as \x5c or
4511 \x20 at the start of a run of normal characters. These were being treated as
4512 real characters, instead of the source characters being re-checked.
4513
4514
4515 Version 0.94 18-Sep-97
4516 ----------------------
4517
4518 1. The functions are now thread-safe, with the caveat that the global variables
4519 containing pointers to malloc() and free() or alternative functions are the
4520 same for all threads.
4521
4522 2. Get pcre_study() to generate a bitmap of initial characters for non-
4523 anchored patterns when this is possible, and use it if passed to pcre_exec().
4524
4525
4526 Version 0.93 15-Sep-97
4527 ----------------------
4528
4529 1. /(b)|(:+)/ was computing an incorrect first character.
4530
4531 2. Add pcre_study() to the API and the passing of pcre_extra to pcre_exec(),
4532 but not actually doing anything yet.
4533
4534 3. Treat "-" characters in classes that cannot be part of ranges as literals,
4535 as Perl does (e.g. [-az] or [az-]).
4536
4537 4. Set the anchored flag if a branch starts with .* or .*? because that tests
4538 all possible positions.
4539
4540 5. Split up into different modules to avoid including unneeded functions in a
4541 compiled binary. However, compile and exec are still in one module. The "study"
4542 function is split off.
4543
4544 6. The character tables are now in a separate module whose source is generated
4545 by an auxiliary program - but can then be edited by hand if required. There are
4546 now no calls to isalnum(), isspace(), isdigit(), isxdigit(), tolower() or
4547 toupper() in the code.
4548
4549 7. Turn the malloc/free funtions variables into pcre_malloc and pcre_free and
4550 make them global. Abolish the function for setting them, as the caller can now
4551 set them directly.
4552
4553
4554 Version 0.92 11-Sep-97
4555 ----------------------
4556
4557 1. A repeat with a fixed maximum and a minimum of 1 for an ordinary character
4558 (e.g. /a{1,3}/) was broken (I mis-optimized it).
4559
4560 2. Caseless matching was not working in character classes if the characters in
4561 the pattern were in upper case.
4562
4563 3. Make ranges like [W-c] work in the same way as Perl for caseless matching.
4564
4565 4. Make PCRE_ANCHORED public and accept as a compile option.
4566
4567 5. Add an options word to pcre_exec() and accept PCRE_ANCHORED and
4568 PCRE_CASELESS at run time. Add escapes \A and \I to pcretest to cause it to
4569 pass them.
4570
4571 6. Give an error if bad option bits passed at compile or run time.
4572
4573 7. Add PCRE_MULTILINE at compile and exec time, and (?m) as well. Add \M to
4574 pcretest to cause it to pass that flag.
4575
4576 8. Add pcre_info(), to get the number of identifying subpatterns, the stored
4577 options, and the first character, if set.
4578
4579 9. Recognize C+ or C{n,m} where n >= 1 as providing a fixed starting character.
4580
4581
4582 Version 0.91 10-Sep-97
4583 ----------------------
4584
4585 1. PCRE was failing to diagnose unlimited repeats of subpatterns that could
4586 match the empty string as in /(a*)*/. It was looping and ultimately crashing.
4587
4588 2. PCRE was looping on encountering an indefinitely repeated back reference to
4589 a subpattern that had matched an empty string, e.g. /(a|)\1*/. It now does what
4590 Perl does - treats the match as successful.
4591
4592 ****

Properties

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

  ViewVC Help
Powered by ViewVC 1.1.5