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

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1510 - (show annotations)
Wed Nov 5 15:08:03 2014 UTC (5 years, 1 month ago) by ph10
File size: 268978 byte(s)
Fix bug when there are unset groups prior to (*ACCEPT) within a capturing 
group.
1 ChangeLog for PCRE
2 ------------------
3
4 Version 8.37 xx-xxx-201x
5 ------------------------
6
7 1. When an (*ACCEPT) is triggered inside capturing parentheses, it arranges
8 for those parentheses to be closed with whatever has been captured so far.
9 However, it was failing to mark any other groups between the hightest
10 capture so far and the currrent group as "unset". Thus, the ovector for
11 those groups contained whatever was previously there. An example is the
12 pattern /(x)|((*ACCEPT))/ when matched against "abcd".
13
14
15
16 Version 8.36 26-September-2014
17 ------------------------------
18
19 1. Got rid of some compiler warnings in the C++ modules that were shown up by
20 -Wmissing-field-initializers and -Wunused-parameter.
21
22 2. The tests for quantifiers being too big (greater than 65535) were being
23 applied after reading the number, and stupidly assuming that integer
24 overflow would give a negative number. The tests are now applied as the
25 numbers are read.
26
27 3. Tidy code in pcre_exec.c where two branches that used to be different are
28 now the same.
29
30 4. The JIT compiler did not generate match limit checks for certain
31 bracketed expressions with quantifiers. This may lead to exponential
32 backtracking, instead of returning with PCRE_ERROR_MATCHLIMIT. This
33 issue should be resolved now.
34
35 5. Fixed an issue, which occures when nested alternatives are optimized
36 with table jumps.
37
38 6. Inserted two casts and changed some ints to size_t in the light of some
39 reported 64-bit compiler warnings (Bugzilla 1477).
40
41 7. Fixed a bug concerned with zero-minimum possessive groups that could match
42 an empty string, which sometimes were behaving incorrectly in the
43 interpreter (though correctly in the JIT matcher). This pcretest input is
44 an example:
45
46 '\A(?:[^"]++|"(?:[^"]*+|"")*+")++'
47 NON QUOTED "QUOT""ED" AFTER "NOT MATCHED
48
49 the interpreter was reporting a match of 'NON QUOTED ' only, whereas the
50 JIT matcher and Perl both matched 'NON QUOTED "QUOT""ED" AFTER '. The test
51 for an empty string was breaking the inner loop and carrying on at a lower
52 level, when possessive repeated groups should always return to a higher
53 level as they have no backtrack points in them. The empty string test now
54 occurs at the outer level.
55
56 8. Fixed a bug that was incorrectly auto-possessifying \w+ in the pattern
57 ^\w+(?>\s*)(?<=\w) which caused it not to match "test test".
58
59 9. Give a compile-time error for \o{} (as Perl does) and for \x{} (which Perl
60 doesn't).
61
62 10. Change 8.34/15 introduced a bug that caused the amount of memory needed
63 to hold a pattern to be incorrectly computed (too small) when there were
64 named back references to duplicated names. This could cause "internal
65 error: code overflow" or "double free or corruption" or other memory
66 handling errors.
67
68 11. When named subpatterns had the same prefixes, back references could be
69 confused. For example, in this pattern:
70
71 /(?P<Name>a)?(?P<Name2>b)?(?(<Name>)c|d)*l/
72
73 the reference to 'Name' was incorrectly treated as a reference to a
74 duplicate name.
75
76 12. A pattern such as /^s?c/mi8 where the optional character has more than
77 one "other case" was incorrectly compiled such that it would only try to
78 match starting at "c".
79
80 13. When a pattern starting with \s was studied, VT was not included in the
81 list of possible starting characters; this should have been part of the
82 8.34/18 patch.
83
84 14. If a character class started [\Qx]... where x is any character, the class
85 was incorrectly terminated at the ].
86
87 15. If a pattern that started with a caseless match for a character with more
88 than one "other case" was studied, PCRE did not set up the starting code
89 unit bit map for the list of possible characters. Now it does. This is an
90 optimization improvement, not a bug fix.
91
92 16. The Unicode data tables have been updated to Unicode 7.0.0.
93
94 17. Fixed a number of memory leaks in pcregrep.
95
96 18. Avoid a compiler warning (from some compilers) for a function call with
97 a cast that removes "const" from an lvalue by using an intermediate
98 variable (to which the compiler does not object).
99
100 19. Incorrect code was compiled if a group that contained an internal recursive
101 back reference was optional (had quantifier with a minimum of zero). This
102 example compiled incorrect code: /(((a\2)|(a*)\g<-1>))*/ and other examples
103 caused segmentation faults because of stack overflows at compile time.
104
105 20. A pattern such as /((?(R)a|(?1)))+/, which contains a recursion within a
106 group that is quantified with an indefinite repeat, caused a compile-time
107 loop which used up all the system stack and provoked a segmentation fault.
108 This was not the same bug as 19 above.
109
110 21. Add PCRECPP_EXP_DECL declaration to operator<< in pcre_stringpiece.h.
111 Patch by Mike Frysinger.
112
113
114 Version 8.35 04-April-2014
115 --------------------------
116
117 1. A new flag is set, when property checks are present in an XCLASS.
118 When this flag is not set, PCRE can perform certain optimizations
119 such as studying these XCLASS-es.
120
121 2. The auto-possessification of character sets were improved: a normal
122 and an extended character set can be compared now. Furthermore
123 the JIT compiler optimizes more character set checks.
124
125 3. Got rid of some compiler warnings for potentially uninitialized variables
126 that show up only when compiled with -O2.
127
128 4. A pattern such as (?=ab\K) that uses \K in an assertion can set the start
129 of a match later then the end of the match. The pcretest program was not
130 handling the case sensibly - it was outputting from the start to the next
131 binary zero. It now reports this situation in a message, and outputs the
132 text from the end to the start.
133
134 5. Fast forward search is improved in JIT. Instead of the first three
135 characters, any three characters with fixed position can be searched.
136 Search order: first, last, middle.
137
138 6. Improve character range checks in JIT. Characters are read by an inprecise
139 function now, which returns with an unknown value if the character code is
140 above a certain threshold (e.g: 256). The only limitation is that the value
141 must be bigger than the threshold as well. This function is useful when
142 the characters above the threshold are handled in the same way.
143
144 7. The macros whose names start with RAWUCHAR are placeholders for a future
145 mode in which only the bottom 21 bits of 32-bit data items are used. To
146 make this more memorable for those maintaining the code, the names have
147 been changed to start with UCHAR21, and an extensive comment has been added
148 to their definition.
149
150 8. Add missing (new) files sljitNativeTILEGX.c and sljitNativeTILEGX-encoder.c
151 to the export list in Makefile.am (they were accidentally omitted from the
152 8.34 tarball).
153
154 9. The informational output from pcretest used the phrase "starting byte set"
155 which is inappropriate for the 16-bit and 32-bit libraries. As the output
156 for "first char" and "need char" really means "non-UTF-char", I've changed
157 "byte" to "char", and slightly reworded the output. The documentation about
158 these values has also been (I hope) clarified.
159
160 10. Another JIT related optimization: use table jumps for selecting the correct
161 backtracking path, when more than four alternatives are present inside a
162 bracket.
163
164 11. Empty match is not possible, when the minimum length is greater than zero,
165 and there is no \K in the pattern. JIT should avoid empty match checks in
166 such cases.
167
168 12. In a caseless character class with UCP support, when a character with more
169 than one alternative case was not the first character of a range, not all
170 the alternative cases were added to the class. For example, s and \x{17f}
171 are both alternative cases for S: the class [RST] was handled correctly,
172 but [R-T] was not.
173
174 13. The configure.ac file always checked for pthread support when JIT was
175 enabled. This is not used in Windows, so I have put this test inside a
176 check for the presence of windows.h (which was already tested for).
177
178 14. Improve pattern prefix search by a simplified Boyer-Moore algorithm in JIT.
179 The algorithm provides a way to skip certain starting offsets, and usually
180 faster than linear prefix searches.
181
182 15. Change 13 for 8.20 updated RunTest to check for the 'fr' locale as well
183 as for 'fr_FR' and 'french'. For some reason, however, it then used the
184 Windows-specific input and output files, which have 'french' screwed in.
185 So this could never have worked. One of the problems with locales is that
186 they aren't always the same. I have now updated RunTest so that it checks
187 the output of the locale test (test 3) against three different output
188 files, and it allows the test to pass if any one of them matches. With luck
189 this should make the test pass on some versions of Solaris where it was
190 failing. Because of the uncertainty, the script did not used to stop if
191 test 3 failed; it now does. If further versions of a French locale ever
192 come to light, they can now easily be added.
193
194 16. If --with-pcregrep-bufsize was given a non-integer value such as "50K",
195 there was a message during ./configure, but it did not stop. This now
196 provokes an error. The invalid example in README has been corrected.
197 If a value less than the minimum is given, the minimum value has always
198 been used, but now a warning is given.
199
200 17. If --enable-bsr-anycrlf was set, the special 16/32-bit test failed. This
201 was a bug in the test system, which is now fixed. Also, the list of various
202 configurations that are tested for each release did not have one with both
203 16/32 bits and --enable-bar-anycrlf. It now does.
204
205 18. pcretest was missing "-C bsr" for displaying the \R default setting.
206
207 19. Little endian PowerPC systems are supported now by the JIT compiler.
208
209 20. The fast forward newline mechanism could enter to an infinite loop on
210 certain invalid UTF-8 input. Although we don't support these cases
211 this issue can be fixed by a performance optimization.
212
213 21. Change 33 of 8.34 is not sufficient to ensure stack safety because it does
214 not take account if existing stack usage. There is now a new global
215 variable called pcre_stack_guard that can be set to point to an external
216 function to check stack availability. It is called at the start of
217 processing every parenthesized group.
218
219 22. A typo in the code meant that in ungreedy mode the max/min qualifier
220 behaved like a min-possessive qualifier, and, for example, /a{1,3}b/U did
221 not match "ab".
222
223 23. When UTF was disabled, the JIT program reported some incorrect compile
224 errors. These messages are silenced now.
225
226 24. Experimental support for ARM-64 and MIPS-64 has been added to the JIT
227 compiler.
228
229 25. Change all the temporary files used in RunGrepTest to be different to those
230 used by RunTest so that the tests can be run simultaneously, for example by
231 "make -j check".
232
233
234 Version 8.34 15-December-2013
235 -----------------------------
236
237 1. Add pcre[16|32]_jit_free_unused_memory to forcibly free unused JIT
238 executable memory. Patch inspired by Carsten Klein.
239
240 2. ./configure --enable-coverage defined SUPPORT_GCOV in config.h, although
241 this macro is never tested and has no effect, because the work to support
242 coverage involves only compiling and linking options and special targets in
243 the Makefile. The comment in config.h implied that defining the macro would
244 enable coverage support, which is totally false. There was also support for
245 setting this macro in the CMake files (my fault, I just copied it from
246 configure). SUPPORT_GCOV has now been removed.
247
248 3. Make a small performance improvement in strlen16() and strlen32() in
249 pcretest.
250
251 4. Change 36 for 8.33 left some unreachable statements in pcre_exec.c,
252 detected by the Solaris compiler (gcc doesn't seem to be able to diagnose
253 these cases). There was also one in pcretest.c.
254
255 5. Cleaned up a "may be uninitialized" compiler warning in pcre_exec.c.
256
257 6. In UTF mode, the code for checking whether a group could match an empty
258 string (which is used for indefinitely repeated groups to allow for
259 breaking an infinite loop) was broken when the group contained a repeated
260 negated single-character class with a character that occupied more than one
261 data item and had a minimum repetition of zero (for example, [^\x{100}]* in
262 UTF-8 mode). The effect was undefined: the group might or might not be
263 deemed as matching an empty string, or the program might have crashed.
264
265 7. The code for checking whether a group could match an empty string was not
266 recognizing that \h, \H, \v, \V, and \R must match a character.
267
268 8. Implemented PCRE_INFO_MATCH_EMPTY, which yields 1 if the pattern can match
269 an empty string. If it can, pcretest shows this in its information output.
270
271 9. Fixed two related bugs that applied to Unicode extended grapheme clusters
272 that were repeated with a maximizing qualifier (e.g. \X* or \X{2,5}) when
273 matched by pcre_exec() without using JIT:
274
275 (a) If the rest of the pattern did not match after a maximal run of
276 grapheme clusters, the code for backing up to try with fewer of them
277 did not always back up over a full grapheme when characters that do not
278 have the modifier quality were involved, e.g. Hangul syllables.
279
280 (b) If the match point in a subject started with modifier character, and
281 there was no match, the code could incorrectly back up beyond the match
282 point, and potentially beyond the first character in the subject,
283 leading to a segfault or an incorrect match result.
284
285 10. A conditional group with an assertion condition could lead to PCRE
286 recording an incorrect first data item for a match if no other first data
287 item was recorded. For example, the pattern (?(?=ab)ab) recorded "a" as a
288 first data item, and therefore matched "ca" after "c" instead of at the
289 start.
290
291 11. Change 40 for 8.33 (allowing pcregrep to find empty strings) showed up a
292 bug that caused the command "echo a | ./pcregrep -M '|a'" to loop.
293
294 12. The source of pcregrep now includes z/OS-specific code so that it can be
295 compiled for z/OS as part of the special z/OS distribution.
296
297 13. Added the -T and -TM options to pcretest.
298
299 14. The code in pcre_compile.c for creating the table of named capturing groups
300 has been refactored. Instead of creating the table dynamically during the
301 actual compiling pass, the information is remembered during the pre-compile
302 pass (on the stack unless there are more than 20 named groups, in which
303 case malloc() is used) and the whole table is created before the actual
304 compile happens. This has simplified the code (it is now nearly 150 lines
305 shorter) and prepared the way for better handling of references to groups
306 with duplicate names.
307
308 15. A back reference to a named subpattern when there is more than one of the
309 same name now checks them in the order in which they appear in the pattern.
310 The first one that is set is used for the reference. Previously only the
311 first one was inspected. This change makes PCRE more compatible with Perl.
312
313 16. Unicode character properties were updated from Unicode 6.3.0.
314
315 17. The compile-time code for auto-possessification has been refactored, based
316 on a patch by Zoltan Herczeg. It now happens after instead of during
317 compilation. The code is cleaner, and more cases are handled. The option
318 PCRE_NO_AUTO_POSSESS is added for testing purposes, and the -O and /O
319 options in pcretest are provided to set it. It can also be set by
320 (*NO_AUTO_POSSESS) at the start of a pattern.
321
322 18. The character VT has been added to the default ("C" locale) set of
323 characters that match \s and are generally treated as white space,
324 following this same change in Perl 5.18. There is now no difference between
325 "Perl space" and "POSIX space". Whether VT is treated as white space in
326 other locales depends on the locale.
327
328 19. The code for checking named groups as conditions, either for being set or
329 for being recursed, has been refactored (this is related to 14 and 15
330 above). Processing unduplicated named groups should now be as fast at
331 numerical groups, and processing duplicated groups should be faster than
332 before.
333
334 20. Two patches to the CMake build system, by Alexander Barkov:
335
336 (1) Replace the "source" command by "." in CMakeLists.txt because
337 "source" is a bash-ism.
338
339 (2) Add missing HAVE_STDINT_H and HAVE_INTTYPES_H to config-cmake.h.in;
340 without these the CMake build does not work on Solaris.
341
342 21. Perl has changed its handling of \8 and \9. If there is no previously
343 encountered capturing group of those numbers, they are treated as the
344 literal characters 8 and 9 instead of a binary zero followed by the
345 literals. PCRE now does the same.
346
347 22. Following Perl, added \o{} to specify codepoints in octal, making it
348 possible to specify values greater than 0777 and also making them
349 unambiguous.
350
351 23. Perl now gives an error for missing closing braces after \x{... instead of
352 treating the string as literal. PCRE now does the same.
353
354 24. RunTest used to grumble if an inappropriate test was selected explicitly,
355 but just skip it when running all tests. This make it awkward to run ranges
356 of tests when one of them was inappropriate. Now it just skips any
357 inappropriate tests, as it always did when running all tests.
358
359 25. If PCRE_AUTO_CALLOUT and PCRE_UCP were set for a pattern that contained
360 character types such as \d or \w, too many callouts were inserted, and the
361 data that they returned was rubbish.
362
363 26. In UCP mode, \s was not matching two of the characters that Perl matches,
364 namely NEL (U+0085) and MONGOLIAN VOWEL SEPARATOR (U+180E), though they
365 were matched by \h. The code has now been refactored so that the lists of
366 the horizontal and vertical whitespace characters used for \h and \v (which
367 are defined only in one place) are now also used for \s.
368
369 27. Add JIT support for the 64 bit TileGX architecture.
370 Patch by Jiong Wang (Tilera Corporation).
371
372 28. Possessive quantifiers for classes (both explicit and automatically
373 generated) now use special opcodes instead of wrapping in ONCE brackets.
374
375 29. Whereas an item such as A{4}+ ignored the possessivenes of the quantifier
376 (because it's meaningless), this was not happening when PCRE_CASELESS was
377 set. Not wrong, but inefficient.
378
379 30. Updated perltest.pl to add /u (force Unicode mode) when /W (use Unicode
380 properties for \w, \d, etc) is present in a test regex. Otherwise if the
381 test contains no characters greater than 255, Perl doesn't realise it
382 should be using Unicode semantics.
383
384 31. Upgraded the handling of the POSIX classes [:graph:], [:print:], and
385 [:punct:] when PCRE_UCP is set so as to include the same characters as Perl
386 does in Unicode mode.
387
388 32. Added the "forbid" facility to pcretest so that putting tests into the
389 wrong test files can sometimes be quickly detected.
390
391 33. There is now a limit (default 250) on the depth of nesting of parentheses.
392 This limit is imposed to control the amount of system stack used at compile
393 time. It can be changed at build time by --with-parens-nest-limit=xxx or
394 the equivalent in CMake.
395
396 34. Character classes such as [A-\d] or [a-[:digit:]] now cause compile-time
397 errors. Perl warns for these when in warning mode, but PCRE has no facility
398 for giving warnings.
399
400 35. Change 34 for 8.13 allowed quantifiers on assertions, because Perl does.
401 However, this was not working for (?!) because it is optimized to (*FAIL),
402 for which PCRE does not allow quantifiers. The optimization is now disabled
403 when a quantifier follows (?!). I can't see any use for this, but it makes
404 things uniform.
405
406 36. Perl no longer allows group names to start with digits, so I have made this
407 change also in PCRE. It simplifies the code a bit.
408
409 37. In extended mode, Perl ignores spaces before a + that indicates a
410 possessive quantifier. PCRE allowed a space before the quantifier, but not
411 before the possessive +. It now does.
412
413 38. The use of \K (reset reported match start) within a repeated possessive
414 group such as (a\Kb)*+ was not working.
415
416 40. Document that the same character tables must be used at compile time and
417 run time, and that the facility to pass tables to pcre_exec() and
418 pcre_dfa_exec() is for use only with saved/restored patterns.
419
420 41. Applied Jeff Trawick's patch CMakeLists.txt, which "provides two new
421 features for Builds with MSVC:
422
423 1. Support pcre.rc and/or pcreposix.rc (as is already done for MinGW
424 builds). The .rc files can be used to set FileDescription and many other
425 attributes.
426
427 2. Add an option (-DINSTALL_MSVC_PDB) to enable installation of .pdb files.
428 This allows higher-level build scripts which want .pdb files to avoid
429 hard-coding the exact files needed."
430
431 42. Added support for [[:<:]] and [[:>:]] as used in the BSD POSIX library to
432 mean "start of word" and "end of word", respectively, as a transition aid.
433
434 43. A minimizing repeat of a class containing codepoints greater than 255 in
435 non-UTF 16-bit or 32-bit modes caused an internal error when PCRE was
436 compiled to use the heap for recursion.
437
438 44. Got rid of some compiler warnings for unused variables when UTF but not UCP
439 is configured.
440
441
442 Version 8.33 28-May-2013
443 ------------------------
444
445 1. Added 'U' to some constants that are compared to unsigned integers, to
446 avoid compiler signed/unsigned warnings. Added (int) casts to unsigned
447 variables that are added to signed variables, to ensure the result is
448 signed and can be negated.
449
450 2. Applied patch by Daniel Richard G for quashing MSVC warnings to the
451 CMake config files.
452
453 3. Revise the creation of config.h.generic so that all boolean macros are
454 #undefined, whereas non-boolean macros are #ifndef/#endif-ed. This makes
455 overriding via -D on the command line possible.
456
457 4. Changing the definition of the variable "op" in pcre_exec.c from pcre_uchar
458 to unsigned int is reported to make a quite noticeable speed difference in
459 a specific Windows environment. Testing on Linux did also appear to show
460 some benefit (and it is clearly not harmful). Also fixed the definition of
461 Xop which should be unsigned.
462
463 5. Related to (4), changing the definition of the intermediate variable cc
464 in repeated character loops from pcre_uchar to pcre_uint32 also gave speed
465 improvements.
466
467 6. Fix forward search in JIT when link size is 3 or greater. Also removed some
468 unnecessary spaces.
469
470 7. Adjust autogen.sh and configure.ac to lose warnings given by automake 1.12
471 and later.
472
473 8. Fix two buffer over read issues in 16 and 32 bit modes. Affects JIT only.
474
475 9. Optimizing fast_forward_start_bits in JIT.
476
477 10. Adding support for callouts in JIT, and fixing some issues revealed
478 during this work. Namely:
479
480 (a) Unoptimized capturing brackets incorrectly reset on backtrack.
481
482 (b) Minimum length was not checked before the matching is started.
483
484 11. The value of capture_last that is passed to callouts was incorrect in some
485 cases when there was a capture on one path that was subsequently abandoned
486 after a backtrack. Also, the capture_last value is now reset after a
487 recursion, since all captures are also reset in this case.
488
489 12. The interpreter no longer returns the "too many substrings" error in the
490 case when an overflowing capture is in a branch that is subsequently
491 abandoned after a backtrack.
492
493 13. In the pathological case when an offset vector of size 2 is used, pcretest
494 now prints out the matched string after a yield of 0 or 1.
495
496 14. Inlining subpatterns in recursions, when certain conditions are fulfilled.
497 Only supported by the JIT compiler at the moment.
498
499 15. JIT compiler now supports 32 bit Macs thanks to Lawrence Velazquez.
500
501 16. Partial matches now set offsets[2] to the "bumpalong" value, that is, the
502 offset of the starting point of the matching process, provided the offsets
503 vector is large enough.
504
505 17. The \A escape now records a lookbehind value of 1, though its execution
506 does not actually inspect the previous character. This is to ensure that,
507 in partial multi-segment matching, at least one character from the old
508 segment is retained when a new segment is processed. Otherwise, if there
509 are no lookbehinds in the pattern, \A might match incorrectly at the start
510 of a new segment.
511
512 18. Added some #ifdef __VMS code into pcretest.c to help VMS implementations.
513
514 19. Redefined some pcre_uchar variables in pcre_exec.c as pcre_uint32; this
515 gives some modest performance improvement in 8-bit mode.
516
517 20. Added the PCRE-specific property \p{Xuc} for matching characters that can
518 be expressed in certain programming languages using Universal Character
519 Names.
520
521 21. Unicode validation has been updated in the light of Unicode Corrigendum #9,
522 which points out that "non characters" are not "characters that may not
523 appear in Unicode strings" but rather "characters that are reserved for
524 internal use and have only local meaning".
525
526 22. When a pattern was compiled with automatic callouts (PCRE_AUTO_CALLOUT) and
527 there was a conditional group that depended on an assertion, if the
528 assertion was false, the callout that immediately followed the alternation
529 in the condition was skipped when pcre_exec() was used for matching.
530
531 23. Allow an explicit callout to be inserted before an assertion that is the
532 condition for a conditional group, for compatibility with automatic
533 callouts, which always insert a callout at this point.
534
535 24. In 8.31, (*COMMIT) was confined to within a recursive subpattern. Perl also
536 confines (*SKIP) and (*PRUNE) in the same way, and this has now been done.
537
538 25. (*PRUNE) is now supported by the JIT compiler.
539
540 26. Fix infinite loop when /(?<=(*SKIP)ac)a/ is matched against aa.
541
542 27. Fix the case where there are two or more SKIPs with arguments that may be
543 ignored.
544
545 28. (*SKIP) is now supported by the JIT compiler.
546
547 29. (*THEN) is now supported by the JIT compiler.
548
549 30. Update RunTest with additional test selector options.
550
551 31. The way PCRE handles backtracking verbs has been changed in two ways.
552
553 (1) Previously, in something like (*COMMIT)(*SKIP), COMMIT would override
554 SKIP. Now, PCRE acts on whichever backtracking verb is reached first by
555 backtracking. In some cases this makes it more Perl-compatible, but Perl's
556 rather obscure rules do not always do the same thing.
557
558 (2) Previously, backtracking verbs were confined within assertions. This is
559 no longer the case for positive assertions, except for (*ACCEPT). Again,
560 this sometimes improves Perl compatibility, and sometimes does not.
561
562 32. A number of tests that were in test 2 because Perl did things differently
563 have been moved to test 1, because either Perl or PCRE has changed, and
564 these tests are now compatible.
565
566 32. Backtracking control verbs are now handled in the same way in JIT and
567 interpreter.
568
569 33. An opening parenthesis in a MARK/PRUNE/SKIP/THEN name in a pattern that
570 contained a forward subroutine reference caused a compile error.
571
572 34. Auto-detect and optimize limited repetitions in JIT.
573
574 35. Implement PCRE_NEVER_UTF to lock out the use of UTF, in particular,
575 blocking (*UTF) etc.
576
577 36. In the interpreter, maximizing pattern repetitions for characters and
578 character types now use tail recursion, which reduces stack usage.
579
580 37. The value of the max lookbehind was not correctly preserved if a compiled
581 and saved regex was reloaded on a host of different endianness.
582
583 38. Implemented (*LIMIT_MATCH) and (*LIMIT_RECURSION). As part of the extension
584 of the compiled pattern block, expand the flags field from 16 to 32 bits
585 because it was almost full.
586
587 39. Try madvise first before posix_madvise.
588
589 40. Change 7 for PCRE 7.9 made it impossible for pcregrep to find empty lines
590 with a pattern such as ^$. It has taken 4 years for anybody to notice! The
591 original change locked out all matches of empty strings. This has been
592 changed so that one match of an empty string per line is recognized.
593 Subsequent searches on the same line (for colouring or for --only-matching,
594 for example) do not recognize empty strings.
595
596 41. Applied a user patch to fix a number of spelling mistakes in comments.
597
598 42. Data lines longer than 65536 caused pcretest to crash.
599
600 43. Clarified the data type for length and startoffset arguments for pcre_exec
601 and pcre_dfa_exec in the function-specific man pages, where they were
602 explicitly stated to be in bytes, never having been updated. I also added
603 some clarification to the pcreapi man page.
604
605 44. A call to pcre_dfa_exec() with an output vector size less than 2 caused
606 a segmentation fault.
607
608
609 Version 8.32 30-November-2012
610 -----------------------------
611
612 1. Improved JIT compiler optimizations for first character search and single
613 character iterators.
614
615 2. Supporting IBM XL C compilers for PPC architectures in the JIT compiler.
616 Patch by Daniel Richard G.
617
618 3. Single character iterator optimizations in the JIT compiler.
619
620 4. Improved JIT compiler optimizations for character ranges.
621
622 5. Rename the "leave" variable names to "quit" to improve WinCE compatibility.
623 Reported by Giuseppe D'Angelo.
624
625 6. The PCRE_STARTLINE bit, indicating that a match can occur only at the start
626 of a line, was being set incorrectly in cases where .* appeared inside
627 atomic brackets at the start of a pattern, or where there was a subsequent
628 *PRUNE or *SKIP.
629
630 7. Improved instruction cache flush for POWER/PowerPC.
631 Patch by Daniel Richard G.
632
633 8. Fixed a number of issues in pcregrep, making it more compatible with GNU
634 grep:
635
636 (a) There is now no limit to the number of patterns to be matched.
637
638 (b) An error is given if a pattern is too long.
639
640 (c) Multiple uses of --exclude, --exclude-dir, --include, and --include-dir
641 are now supported.
642
643 (d) --exclude-from and --include-from (multiple use) have been added.
644
645 (e) Exclusions and inclusions now apply to all files and directories, not
646 just to those obtained from scanning a directory recursively.
647
648 (f) Multiple uses of -f and --file-list are now supported.
649
650 (g) In a Windows environment, the default for -d has been changed from
651 "read" (the GNU grep default) to "skip", because otherwise the presence
652 of a directory in the file list provokes an error.
653
654 (h) The documentation has been revised and clarified in places.
655
656 9. Improve the matching speed of capturing brackets.
657
658 10. Changed the meaning of \X so that it now matches a Unicode extended
659 grapheme cluster.
660
661 11. Patch by Daniel Richard G to the autoconf files to add a macro for sorting
662 out POSIX threads when JIT support is configured.
663
664 12. Added support for PCRE_STUDY_EXTRA_NEEDED.
665
666 13. In the POSIX wrapper regcomp() function, setting re_nsub field in the preg
667 structure could go wrong in environments where size_t is not the same size
668 as int.
669
670 14. Applied user-supplied patch to pcrecpp.cc to allow PCRE_NO_UTF8_CHECK to be
671 set.
672
673 15. The EBCDIC support had decayed; later updates to the code had included
674 explicit references to (e.g.) \x0a instead of CHAR_LF. There has been a
675 general tidy up of EBCDIC-related issues, and the documentation was also
676 not quite right. There is now a test that can be run on ASCII systems to
677 check some of the EBCDIC-related things (but is it not a full test).
678
679 16. The new PCRE_STUDY_EXTRA_NEEDED option is now used by pcregrep, resulting
680 in a small tidy to the code.
681
682 17. Fix JIT tests when UTF is disabled and both 8 and 16 bit mode are enabled.
683
684 18. If the --only-matching (-o) option in pcregrep is specified multiple
685 times, each one causes appropriate output. For example, -o1 -o2 outputs the
686 substrings matched by the 1st and 2nd capturing parentheses. A separating
687 string can be specified by --om-separator (default empty).
688
689 19. Improving the first n character searches.
690
691 20. Turn case lists for horizontal and vertical white space into macros so that
692 they are defined only once.
693
694 21. This set of changes together give more compatible Unicode case-folding
695 behaviour for characters that have more than one other case when UCP
696 support is available.
697
698 (a) The Unicode property table now has offsets into a new table of sets of
699 three or more characters that are case-equivalent. The MultiStage2.py
700 script that generates these tables (the pcre_ucd.c file) now scans
701 CaseFolding.txt instead of UnicodeData.txt for character case
702 information.
703
704 (b) The code for adding characters or ranges of characters to a character
705 class has been abstracted into a generalized function that also handles
706 case-independence. In UTF-mode with UCP support, this uses the new data
707 to handle characters with more than one other case.
708
709 (c) A bug that is fixed as a result of (b) is that codepoints less than 256
710 whose other case is greater than 256 are now correctly matched
711 caselessly. Previously, the high codepoint matched the low one, but not
712 vice versa.
713
714 (d) The processing of \h, \H, \v, and \ in character classes now makes use
715 of the new class addition function, using character lists defined as
716 macros alongside the case definitions of 20 above.
717
718 (e) Caseless back references now work with characters that have more than
719 one other case.
720
721 (f) General caseless matching of characters with more than one other case
722 is supported.
723
724 22. Unicode character properties were updated from Unicode 6.2.0
725
726 23. Improved CMake support under Windows. Patch by Daniel Richard G.
727
728 24. Add support for 32-bit character strings, and UTF-32
729
730 25. Major JIT compiler update (code refactoring and bugfixing).
731 Experimental Sparc 32 support is added.
732
733 26. Applied a modified version of Daniel Richard G's patch to create
734 pcre.h.generic and config.h.generic by "make" instead of in the
735 PrepareRelease script.
736
737 27. Added a definition for CHAR_NULL (helpful for the z/OS port), and use it in
738 pcre_compile.c when checking for a zero character.
739
740 28. Introducing a native interface for JIT. Through this interface, the compiled
741 machine code can be directly executed. The purpose of this interface is to
742 provide fast pattern matching, so several sanity checks are not performed.
743 However, feature tests are still performed. The new interface provides
744 1.4x speedup compared to the old one.
745
746 29. If pcre_exec() or pcre_dfa_exec() was called with a negative value for
747 the subject string length, the error given was PCRE_ERROR_BADOFFSET, which
748 was confusing. There is now a new error PCRE_ERROR_BADLENGTH for this case.
749
750 30. In 8-bit UTF-8 mode, pcretest failed to give an error for data codepoints
751 greater than 0x7fffffff (which cannot be represented in UTF-8, even under
752 the "old" RFC 2279). Instead, it ended up passing a negative length to
753 pcre_exec().
754
755 31. Add support for GCC's visibility feature to hide internal functions.
756
757 32. Running "pcretest -C pcre8" or "pcretest -C pcre16" gave a spurious error
758 "unknown -C option" after outputting 0 or 1.
759
760 33. There is now support for generating a code coverage report for the test
761 suite in environments where gcc is the compiler and lcov is installed. This
762 is mainly for the benefit of the developers.
763
764 34. If PCRE is built with --enable-valgrind, certain memory regions are marked
765 unaddressable using valgrind annotations, allowing valgrind to detect
766 invalid memory accesses. This is mainly for the benefit of the developers.
767
768 25. (*UTF) can now be used to start a pattern in any of the three libraries.
769
770 26. Give configure error if --enable-cpp but no C++ compiler found.
771
772
773 Version 8.31 06-July-2012
774 -------------------------
775
776 1. Fixing a wrong JIT test case and some compiler warnings.
777
778 2. Removed a bashism from the RunTest script.
779
780 3. Add a cast to pcre_exec.c to fix the warning "unary minus operator applied
781 to unsigned type, result still unsigned" that was given by an MS compiler
782 on encountering the code "-sizeof(xxx)".
783
784 4. Partial matching support is added to the JIT compiler.
785
786 5. Fixed several bugs concerned with partial matching of items that consist
787 of more than one character:
788
789 (a) /^(..)\1/ did not partially match "aba" because checking references was
790 done on an "all or nothing" basis. This also applied to repeated
791 references.
792
793 (b) \R did not give a hard partial match if \r was found at the end of the
794 subject.
795
796 (c) \X did not give a hard partial match after matching one or more
797 characters at the end of the subject.
798
799 (d) When newline was set to CRLF, a pattern such as /a$/ did not recognize
800 a partial match for the string "\r".
801
802 (e) When newline was set to CRLF, the metacharacter "." did not recognize
803 a partial match for a CR character at the end of the subject string.
804
805 6. If JIT is requested using /S++ or -s++ (instead of just /S+ or -s+) when
806 running pcretest, the text "(JIT)" added to the output whenever JIT is
807 actually used to run the match.
808
809 7. Individual JIT compile options can be set in pcretest by following -s+[+]
810 or /S+[+] with a digit between 1 and 7.
811
812 8. OP_NOT now supports any UTF character not just single-byte ones.
813
814 9. (*MARK) control verb is now supported by the JIT compiler.
815
816 10. The command "./RunTest list" lists the available tests without actually
817 running any of them. (Because I keep forgetting what they all are.)
818
819 11. Add PCRE_INFO_MAXLOOKBEHIND.
820
821 12. Applied a (slightly modified) user-supplied patch that improves performance
822 when the heap is used for recursion (compiled with --disable-stack-for-
823 recursion). Instead of malloc and free for each heap frame each time a
824 logical recursion happens, frames are retained on a chain and re-used where
825 possible. This sometimes gives as much as 30% improvement.
826
827 13. As documented, (*COMMIT) is now confined to within a recursive subpattern
828 call.
829
830 14. As documented, (*COMMIT) is now confined to within a positive assertion.
831
832 15. It is now possible to link pcretest with libedit as an alternative to
833 libreadline.
834
835 16. (*COMMIT) control verb is now supported by the JIT compiler.
836
837 17. The Unicode data tables have been updated to Unicode 6.1.0.
838
839 18. Added --file-list option to pcregrep.
840
841 19. Added binary file support to pcregrep, including the -a, --binary-files,
842 -I, and --text options.
843
844 20. The madvise function is renamed for posix_madvise for QNX compatibility
845 reasons. Fixed by Giuseppe D'Angelo.
846
847 21. Fixed a bug for backward assertions with REVERSE 0 in the JIT compiler.
848
849 22. Changed the option for creating symbolic links for 16-bit man pages from
850 -s to -sf so that re-installing does not cause issues.
851
852 23. Support PCRE_NO_START_OPTIMIZE in JIT as (*MARK) support requires it.
853
854 24. Fixed a very old bug in pcretest that caused errors with restarted DFA
855 matches in certain environments (the workspace was not being correctly
856 retained). Also added to pcre_dfa_exec() a simple plausibility check on
857 some of the workspace data at the beginning of a restart.
858
859 25. \s*\R was auto-possessifying the \s* when it should not, whereas \S*\R
860 was not doing so when it should - probably a typo introduced by SVN 528
861 (change 8.10/14).
862
863 26. When PCRE_UCP was not set, \w+\x{c4} was incorrectly auto-possessifying the
864 \w+ when the character tables indicated that \x{c4} was a word character.
865 There were several related cases, all because the tests for doing a table
866 lookup were testing for characters less than 127 instead of 255.
867
868 27. If a pattern contains capturing parentheses that are not used in a match,
869 their slots in the ovector are set to -1. For those that are higher than
870 any matched groups, this happens at the end of processing. In the case when
871 there were back references that the ovector was too small to contain
872 (causing temporary malloc'd memory to be used during matching), and the
873 highest capturing number was not used, memory off the end of the ovector
874 was incorrectly being set to -1. (It was using the size of the temporary
875 memory instead of the true size.)
876
877 28. To catch bugs like 27 using valgrind, when pcretest is asked to specify an
878 ovector size, it uses memory at the end of the block that it has got.
879
880 29. Check for an overlong MARK name and give an error at compile time. The
881 limit is 255 for the 8-bit library and 65535 for the 16-bit library.
882
883 30. JIT compiler update.
884
885 31. JIT is now supported on jailbroken iOS devices. Thanks for Ruiger
886 Rill for the patch.
887
888 32. Put spaces around SLJIT_PRINT_D in the JIT compiler. Required by CXX11.
889
890 33. Variable renamings in the PCRE-JIT compiler. No functionality change.
891
892 34. Fixed typos in pcregrep: in two places there was SUPPORT_LIBZ2 instead of
893 SUPPORT_LIBBZ2. This caused a build problem when bzip2 but not gzip (zlib)
894 was enabled.
895
896 35. Improve JIT code generation for greedy plus quantifier.
897
898 36. When /((?:a?)*)*c/ or /((?>a?)*)*c/ was matched against "aac", it set group
899 1 to "aa" instead of to an empty string. The bug affected repeated groups
900 that could potentially match an empty string.
901
902 37. Optimizing single character iterators in JIT.
903
904 38. Wide characters specified with \uxxxx in JavaScript mode are now subject to
905 the same checks as \x{...} characters in non-JavaScript mode. Specifically,
906 codepoints that are too big for the mode are faulted, and in a UTF mode,
907 disallowed codepoints are also faulted.
908
909 39. If PCRE was compiled with UTF support, in three places in the DFA
910 matcher there was code that should only have been obeyed in UTF mode, but
911 was being obeyed unconditionally. In 8-bit mode this could cause incorrect
912 processing when bytes with values greater than 127 were present. In 16-bit
913 mode the bug would be provoked by values in the range 0xfc00 to 0xdc00. In
914 both cases the values are those that cannot be the first data item in a UTF
915 character. The three items that might have provoked this were recursions,
916 possessively repeated groups, and atomic groups.
917
918 40. Ensure that libpcre is explicitly listed in the link commands for pcretest
919 and pcregrep, because some OS require shared objects to be explicitly
920 passed to ld, causing the link step to fail if they are not.
921
922 41. There were two incorrect #ifdefs in pcre_study.c, meaning that, in 16-bit
923 mode, patterns that started with \h* or \R* might be incorrectly matched.
924
925
926 Version 8.30 04-February-2012
927 -----------------------------
928
929 1. Renamed "isnumber" as "is_a_number" because in some Mac environments this
930 name is defined in ctype.h.
931
932 2. Fixed a bug in fixed-length calculation for lookbehinds that would show up
933 only in quite long subpatterns.
934
935 3. Removed the function pcre_info(), which has been obsolete and deprecated
936 since it was replaced by pcre_fullinfo() in February 2000.
937
938 4. For a non-anchored pattern, if (*SKIP) was given with a name that did not
939 match a (*MARK), and the match failed at the start of the subject, a
940 reference to memory before the start of the subject could occur. This bug
941 was introduced by fix 17 of release 8.21.
942
943 5. A reference to an unset group with zero minimum repetition was giving
944 totally wrong answers (in non-JavaScript-compatibility mode). For example,
945 /(another)?(\1?)test/ matched against "hello world test". This bug was
946 introduced in release 8.13.
947
948 6. Add support for 16-bit character strings (a large amount of work involving
949 many changes and refactorings).
950
951 7. RunGrepTest failed on msys because \r\n was replaced by whitespace when the
952 command "pattern=`printf 'xxx\r\njkl'`" was run. The pattern is now taken
953 from a file.
954
955 8. Ovector size of 2 is also supported by JIT based pcre_exec (the ovector size
956 rounding is not applied in this particular case).
957
958 9. The invalid Unicode surrogate codepoints U+D800 to U+DFFF are now rejected
959 if they appear, or are escaped, in patterns.
960
961 10. Get rid of a number of -Wunused-but-set-variable warnings.
962
963 11. The pattern /(?=(*:x))(q|)/ matches an empty string, and returns the mark
964 "x". The similar pattern /(?=(*:x))((*:y)q|)/ did not return a mark at all.
965 Oddly, Perl behaves the same way. PCRE has been fixed so that this pattern
966 also returns the mark "x". This bug applied to capturing parentheses,
967 non-capturing parentheses, and atomic parentheses. It also applied to some
968 assertions.
969
970 12. Stephen Kelly's patch to CMakeLists.txt allows it to parse the version
971 information out of configure.ac instead of relying on pcre.h.generic, which
972 is not stored in the repository.
973
974 13. Applied Dmitry V. Levin's patch for a more portable method for linking with
975 -lreadline.
976
977 14. ZH added PCRE_CONFIG_JITTARGET; added its output to pcretest -C.
978
979 15. Applied Graycode's patch to put the top-level frame on the stack rather
980 than the heap when not using the stack for recursion. This gives a
981 performance improvement in many cases when recursion is not deep.
982
983 16. Experimental code added to "pcretest -C" to output the stack frame size.
984
985
986 Version 8.21 12-Dec-2011
987 ------------------------
988
989 1. Updating the JIT compiler.
990
991 2. JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
992 are added as well.
993
994 3. Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
995 PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
996 calling _pcre_jit_exec. Some extra comments are added.
997
998 4. (*MARK) settings inside atomic groups that do not contain any capturing
999 parentheses, for example, (?>a(*:m)), were not being passed out. This bug
1000 was introduced by change 18 for 8.20.
1001
1002 5. Supporting of \x, \U and \u in JavaScript compatibility mode based on the
1003 ECMA-262 standard.
1004
1005 6. Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
1006 erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
1007 This bug was probably introduced by change 9 of 8.13.
1008
1009 7. While fixing 6 above, I noticed that a number of other items were being
1010 incorrectly rejected as "not fixed length". This arose partly because newer
1011 opcodes had not been added to the fixed-length checking code. I have (a)
1012 corrected the bug and added tests for these items, and (b) arranged for an
1013 error to occur if an unknown opcode is encountered while checking for fixed
1014 length instead of just assuming "not fixed length". The items that were
1015 rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
1016 (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
1017 repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
1018
1019 8. A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
1020 being incorrectly compiled and would have given unpredicatble results.
1021
1022 9. A possessively repeated subpattern with minimum repeat count greater than
1023 one behaved incorrectly. For example, (A){2,}+ behaved as if it was
1024 (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
1025 the first (A) could occur when it should not.
1026
1027 10. Add a cast and remove a redundant test from the code.
1028
1029 11. JIT should use pcre_malloc/pcre_free for allocation.
1030
1031 12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
1032 best practice nowadays, and helps with cross-compiling. (If the exec_prefix
1033 is anything other than /usr, -L is still shown).
1034
1035 13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
1036
1037 14. Perl does not support \N without a following name in a [] class; PCRE now
1038 also gives an error.
1039
1040 15. If a forward reference was repeated with an upper limit of around 2000,
1041 it caused the error "internal error: overran compiling workspace". The
1042 maximum number of forward references (including repeats) was limited by the
1043 internal workspace, and dependent on the LINK_SIZE. The code has been
1044 rewritten so that the workspace expands (via pcre_malloc) if necessary, and
1045 the default depends on LINK_SIZE. There is a new upper limit (for safety)
1046 of around 200,000 forward references. While doing this, I also speeded up
1047 the filling in of repeated forward references.
1048
1049 16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
1050 incorrectly expecting the subject to contain another "a" after the start.
1051
1052 17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
1053 in the match, the SKIP should be ignored. This was not happening; instead
1054 the SKIP was being treated as NOMATCH. For patterns such as
1055 /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
1056 tested.
1057
1058 18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
1059 now much more compatible with Perl, in particular in cases where the result
1060 is a non-match for a non-anchored pattern. For example, if
1061 /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
1062 "m", where previously it did not return a name. A side effect of this
1063 change is that for partial matches, the last encountered mark name is
1064 returned, as for non matches. A number of tests that were previously not
1065 Perl-compatible have been moved into the Perl-compatible test files. The
1066 refactoring has had the pleasing side effect of removing one argument from
1067 the match() function, thus reducing its stack requirements.
1068
1069 19. If the /S+ option was used in pcretest to study a pattern using JIT,
1070 subsequent uses of /S (without +) incorrectly behaved like /S+.
1071
1072 21. Retrieve executable code size support for the JIT compiler and fixing
1073 some warnings.
1074
1075 22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
1076 not work when the shorter character appeared right at the end of the
1077 subject string.
1078
1079 23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
1080 systems.
1081
1082 24. Added PCRE_INFO_JITSIZE to pass on the value from (21) above, and also
1083 output it when the /M option is used in pcretest.
1084
1085 25. The CheckMan script was not being included in the distribution. Also, added
1086 an explicit "perl" to run Perl scripts from the PrepareRelease script
1087 because this is reportedly needed in Windows.
1088
1089 26. If study data was being save in a file and studying had not found a set of
1090 "starts with" bytes for the pattern, the data written to the file (though
1091 never used) was taken from uninitialized memory and so caused valgrind to
1092 complain.
1093
1094 27. Updated RunTest.bat as provided by Sheri Pierce.
1095
1096 28. Fixed a possible uninitialized memory bug in pcre_jit_compile.c.
1097
1098 29. Computation of memory usage for the table of capturing group names was
1099 giving an unnecessarily large value.
1100
1101
1102 Version 8.20 21-Oct-2011
1103 ------------------------
1104
1105 1. Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
1106 a POSIX class. After further experiments with Perl, which convinced me that
1107 Perl has bugs and confusions, a closing square bracket is no longer allowed
1108 in a POSIX name. This bug also affected patterns with classes that started
1109 with full stops.
1110
1111 2. If a pattern such as /(a)b|ac/ is matched against "ac", there is no
1112 captured substring, but while checking the failing first alternative,
1113 substring 1 is temporarily captured. If the output vector supplied to
1114 pcre_exec() was not big enough for this capture, the yield of the function
1115 was still zero ("insufficient space for captured substrings"). This cannot
1116 be totally fixed without adding another stack variable, which seems a lot
1117 of expense for a edge case. However, I have improved the situation in cases
1118 such as /(a)(b)x|abc/ matched against "abc", where the return code
1119 indicates that fewer than the maximum number of slots in the ovector have
1120 been set.
1121
1122 3. Related to (2) above: when there are more back references in a pattern than
1123 slots in the output vector, pcre_exec() uses temporary memory during
1124 matching, and copies in the captures as far as possible afterwards. It was
1125 using the entire output vector, but this conflicts with the specification
1126 that only 2/3 is used for passing back captured substrings. Now it uses
1127 only the first 2/3, for compatibility. This is, of course, another edge
1128 case.
1129
1130 4. Zoltan Herczeg's just-in-time compiler support has been integrated into the
1131 main code base, and can be used by building with --enable-jit. When this is
1132 done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
1133 runtime --no-jit option is given.
1134
1135 5. When the number of matches in a pcre_dfa_exec() run exactly filled the
1136 ovector, the return from the function was zero, implying that there were
1137 other matches that did not fit. The correct "exactly full" value is now
1138 returned.
1139
1140 6. If a subpattern that was called recursively or as a subroutine contained
1141 (*PRUNE) or any other control that caused it to give a non-standard return,
1142 invalid errors such as "Error -26 (nested recursion at the same subject
1143 position)" or even infinite loops could occur.
1144
1145 7. If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
1146 computing the minimum length on reaching *ACCEPT, and so ended up with the
1147 wrong value of 1 rather than 0. Further investigation indicates that
1148 computing a minimum subject length in the presence of *ACCEPT is difficult
1149 (think back references, subroutine calls), and so I have changed the code
1150 so that no minimum is registered for a pattern that contains *ACCEPT.
1151
1152 8. If (*THEN) was present in the first (true) branch of a conditional group,
1153 it was not handled as intended. [But see 16 below.]
1154
1155 9. Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
1156 Sheri Pierce.
1157
1158 10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
1159 the first byte in a match must be "a".
1160
1161 11. Change 17 for 8.13 increased the recursion depth for patterns like
1162 /a(?:.)*?a/ drastically. I've improved things by remembering whether a
1163 pattern contains any instances of (*THEN). If it does not, the old
1164 optimizations are restored. It would be nice to do this on a per-group
1165 basis, but at the moment that is not feasible.
1166
1167 12. In some environments, the output of pcretest -C is CRLF terminated. This
1168 broke RunTest's code that checks for the link size. A single white space
1169 character after the value is now allowed for.
1170
1171 13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
1172 For "fr", it uses the Windows-specific input and output files.
1173
1174 14. If (*THEN) appeared in a group that was called recursively or as a
1175 subroutine, it did not work as intended. [But see next item.]
1176
1177 15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
1178 pattern fragments (but not containing any | characters). If A and B are
1179 matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
1180 was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
1181 D. In other words, Perl considers parentheses that do not contain any |
1182 characters to be part of a surrounding alternative, whereas PCRE was
1183 treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
1184 differently. PCRE now behaves in the same way as Perl, except in the case
1185 of subroutine/recursion calls such as (?1) which have in any case always
1186 been different (but PCRE had them first :-).
1187
1188 16. Related to 15 above: Perl does not treat the | in a conditional group as
1189 creating alternatives. Such a group is treated in the same way as an
1190 ordinary group without any | characters when processing (*THEN). PCRE has
1191 been changed to match Perl's behaviour.
1192
1193 17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
1194 RunGrepTest script failed.
1195
1196 18. Change 22 for version 13 caused atomic groups to use more stack. This is
1197 inevitable for groups that contain captures, but it can lead to a lot of
1198 stack use in large patterns. The old behaviour has been restored for atomic
1199 groups that do not contain any capturing parentheses.
1200
1201 19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
1202 suppress the check for a minimum subject length at run time. (If it was
1203 given to pcre_exec() or pcre_dfa_exec() it did work.)
1204
1205 20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
1206 fail to work when decoding hex characters in data strings in EBCDIC
1207 environments.
1208
1209 21. It appears that in at least one Mac OS environment, the isxdigit() function
1210 is implemented as a macro that evaluates to its argument more than once,
1211 contravening the C 90 Standard (I haven't checked a later standard). There
1212 was an instance in pcretest which caused it to go wrong when processing
1213 \x{...} escapes in subject strings. The has been rewritten to avoid using
1214 things like p++ in the argument of isxdigit().
1215
1216
1217 Version 8.13 16-Aug-2011
1218 ------------------------
1219
1220 1. The Unicode data tables have been updated to Unicode 6.0.0.
1221
1222 2. Two minor typos in pcre_internal.h have been fixed.
1223
1224 3. Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
1225 pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
1226 in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
1227
1228 4. There were a number of related bugs in the code for matching backrefences
1229 caselessly in UTF-8 mode when codes for the characters concerned were
1230 different numbers of bytes. For example, U+023A and U+2C65 are an upper
1231 and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
1232 (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
1233 code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
1234 2-byte code at the end of the subject (it thought there wasn't enough data
1235 left).
1236
1237 5. Comprehensive information about what went wrong is now returned by
1238 pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
1239 as the output vector has at least 2 elements. The offset of the start of
1240 the failing character and a reason code are placed in the vector.
1241
1242 6. When the UTF-8 string check fails for pcre_compile(), the offset that is
1243 now returned is for the first byte of the failing character, instead of the
1244 last byte inspected. This is an incompatible change, but I hope it is small
1245 enough not to be a problem. It makes the returned offset consistent with
1246 pcre_exec() and pcre_dfa_exec().
1247
1248 7. pcretest now gives a text phrase as well as the error number when
1249 pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
1250 failure, the offset and reason code are output.
1251
1252 8. When \R was used with a maximizing quantifier it failed to skip backwards
1253 over a \r\n pair if the subsequent match failed. Instead, it just skipped
1254 back over a single character (\n). This seems wrong (because it treated the
1255 two characters as a single entity when going forwards), conflicts with the
1256 documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
1257 behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
1258 has been changed.
1259
1260 9. Some internal refactoring has changed the processing so that the handling
1261 of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
1262 time (the PCRE_DOTALL option was changed this way some time ago: version
1263 7.7 change 16). This has made it possible to abolish the OP_OPT op code,
1264 which was always a bit of a fudge. It also means that there is one less
1265 argument for the match() function, which reduces its stack requirements
1266 slightly. This change also fixes an incompatibility with Perl: the pattern
1267 (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
1268
1269 10. More internal refactoring has drastically reduced the number of recursive
1270 calls to match() for possessively repeated groups such as (abc)++ when
1271 using pcre_exec().
1272
1273 11. While implementing 10, a number of bugs in the handling of groups were
1274 discovered and fixed:
1275
1276 (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
1277 (a|)*(?1) gave a compile-time internal error.
1278 ((a|)+)+ did not notice that the outer group could match an empty string.
1279 (^a|^)+ was not marked as anchored.
1280 (.*a|.*)+ was not marked as matching at start or after a newline.
1281
1282 12. Yet more internal refactoring has removed another argument from the match()
1283 function. Special calls to this function are now indicated by setting a
1284 value in a variable in the "match data" data block.
1285
1286 13. Be more explicit in pcre_study() instead of relying on "default" for
1287 opcodes that mean there is no starting character; this means that when new
1288 ones are added and accidentally left out of pcre_study(), testing should
1289 pick them up.
1290
1291 14. The -s option of pcretest has been documented for ages as being an old
1292 synonym of -m (show memory usage). I have changed it to mean "force study
1293 for every regex", that is, assume /S for every regex. This is similar to -i
1294 and -d etc. It's slightly incompatible, but I'm hoping nobody is still
1295 using it. It makes it easier to run collections of tests with and without
1296 study enabled, and thereby test pcre_study() more easily. All the standard
1297 tests are now run with and without -s (but some patterns can be marked as
1298 "never study" - see 20 below).
1299
1300 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
1301 restoration of the capturing data to the outer values was not happening
1302 correctly.
1303
1304 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
1305 empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
1306 pattern had matched an empty string, and so incorrectly returned a no
1307 match.
1308
1309 17. There was optimizing code for the last branch of non-capturing parentheses,
1310 and also for the obeyed branch of a conditional subexpression, which used
1311 tail recursion to cut down on stack usage. Unfortunately, now that there is
1312 the possibility of (*THEN) occurring in these branches, tail recursion is
1313 no longer possible because the return has to be checked for (*THEN). These
1314 two optimizations have therefore been removed. [But see 8.20/11 above.]
1315
1316 18. If a pattern containing \R was studied, it was assumed that \R always
1317 matched two bytes, thus causing the minimum subject length to be
1318 incorrectly computed because \R can also match just one byte.
1319
1320 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
1321 was incorrectly computed.
1322
1323 20. If /S is present twice on a test pattern in pcretest input, it now
1324 *disables* studying, thereby overriding the use of -s on the command line
1325 (see 14 above). This is necessary for one or two tests to keep the output
1326 identical in both cases.
1327
1328 21. When (*ACCEPT) was used in an assertion that matched an empty string and
1329 PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
1330
1331 22. When an atomic group that contained a capturing parenthesis was
1332 successfully matched, but the branch in which it appeared failed, the
1333 capturing was not being forgotten if a higher numbered group was later
1334 captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
1335 group 1 to "a", when in fact it should be unset. This applied to multi-
1336 branched capturing and non-capturing groups, repeated or not, and also to
1337 positive assertions (capturing in negative assertions does not happen
1338 in PCRE) and also to nested atomic groups.
1339
1340 23. Add the ++ qualifier feature to pcretest, to show the remainder of the
1341 subject after a captured substring, to make it easier to tell which of a
1342 number of identical substrings has been captured.
1343
1344 24. The way atomic groups are processed by pcre_exec() has been changed so that
1345 if they are repeated, backtracking one repetition now resets captured
1346 values correctly. For example, if ((?>(a+)b)+aabab) is matched against
1347 "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
1348 "aaa". Previously, it would have been "a". As part of this code
1349 refactoring, the way recursive calls are handled has also been changed.
1350
1351 25. If an assertion condition captured any substrings, they were not passed
1352 back unless some other capturing happened later. For example, if
1353 (?(?=(a))a) was matched against "a", no capturing was returned.
1354
1355 26. When studying a pattern that contained subroutine calls or assertions,
1356 the code for finding the minimum length of a possible match was handling
1357 direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
1358 group 1 called group 2 while simultaneously a separate group 2 called group
1359 1). A stack overflow occurred in this case. I have fixed this by limiting
1360 the recursion depth to 10.
1361
1362 27. Updated RunTest.bat in the distribution to the version supplied by Tom
1363 Fortmann. This supports explicit test numbers on the command line, and has
1364 argument validation and error reporting.
1365
1366 28. An instance of \X with an unlimited repeat could fail if at any point the
1367 first character it looked at was a mark character.
1368
1369 29. Some minor code refactoring concerning Unicode properties and scripts
1370 should reduce the stack requirement of match() slightly.
1371
1372 30. Added the '=' option to pcretest to check the setting of unused capturing
1373 slots at the end of the pattern, which are documented as being -1, but are
1374 not included in the return count.
1375
1376 31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
1377 compiled something random. Now it gives a compile-time error (as does
1378 Perl).
1379
1380 32. A *MARK encountered during the processing of a positive assertion is now
1381 recorded and passed back (compatible with Perl).
1382
1383 33. If --only-matching or --colour was set on a pcregrep call whose pattern
1384 had alternative anchored branches, the search for a second match in a line
1385 was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
1386 matched the line "0102" twice. The same bug affected patterns that started
1387 with a backwards assertion. For example /\b01|\b02/ also matched "0102"
1388 twice.
1389
1390 34. Previously, PCRE did not allow quantification of assertions. However, Perl
1391 does, and because of capturing effects, quantifying parenthesized
1392 assertions may at times be useful. Quantifiers are now allowed for
1393 parenthesized assertions.
1394
1395 35. A minor code tidy in pcre_compile() when checking options for \R usage.
1396
1397 36. \g was being checked for fancy things in a character class, when it should
1398 just be a literal "g".
1399
1400 37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
1401 appearance of a nested POSIX class supersedes an apparent external class.
1402 For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
1403 unescaped square brackets may also appear as part of class names. For
1404 example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
1405 more like Perl. (But see 8.20/1 above.)
1406
1407 38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
1408 was because it thought it was \N{name}, which is not supported).
1409
1410 39. Add minix to OS list not supporting the -S option in pcretest.
1411
1412 40. PCRE tries to detect cases of infinite recursion at compile time, but it
1413 cannot analyze patterns in sufficient detail to catch mutual recursions
1414 such as ((?1))((?2)). There is now a runtime test that gives an error if a
1415 subgroup is called recursively as a subpattern for a second time at the
1416 same position in the subject string. In previous releases this might have
1417 been caught by the recursion limit, or it might have run out of stack.
1418
1419 41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
1420 happen only once. PCRE was, however incorrectly giving a compile time error
1421 "recursive call could loop indefinitely" because it cannot analyze the
1422 pattern in sufficient detail. The compile time test no longer happens when
1423 PCRE is compiling a conditional subpattern, but actual runaway loops are
1424 now caught at runtime (see 40 above).
1425
1426 42. It seems that Perl allows any characters other than a closing parenthesis
1427 to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
1428 has been changed to be the same.
1429
1430 43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
1431 as not to get warnings when autogen.sh is called. Also changed
1432 AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
1433
1434 44. To help people who use pcregrep to scan files containing exceedingly long
1435 lines, the following changes have been made:
1436
1437 (a) The default value of the buffer size parameter has been increased from
1438 8K to 20K. (The actual buffer used is three times this size.)
1439
1440 (b) The default can be changed by ./configure --with-pcregrep-bufsize when
1441 PCRE is built.
1442
1443 (c) A --buffer-size=n option has been added to pcregrep, to allow the size
1444 to be set at run time.
1445
1446 (d) Numerical values in pcregrep options can be followed by K or M, for
1447 example --buffer-size=50K.
1448
1449 (e) If a line being scanned overflows pcregrep's buffer, an error is now
1450 given and the return code is set to 2.
1451
1452 45. Add a pointer to the latest mark to the callout data block.
1453
1454 46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
1455 partial match of an empty string instead of no match. This was specific to
1456 the use of ".".
1457
1458 47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
1459 complete match instead of a partial match. This bug was dependent on both
1460 the PCRE_UTF8 and PCRE_DOTALL options being set.
1461
1462 48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
1463 starting byte set, because \b was not being ignored.
1464
1465
1466 Version 8.12 15-Jan-2011
1467 ------------------------
1468
1469 1. Fixed some typos in the markup of the man pages, and wrote a script that
1470 checks for such things as part of the documentation building process.
1471
1472 2. On a big-endian 64-bit system, pcregrep did not correctly process the
1473 --match-limit and --recursion-limit options (added for 8.11). In
1474 particular, this made one of the standard tests fail. (The integer value
1475 went into the wrong half of a long int.)
1476
1477 3. If the --colour option was given to pcregrep with -v (invert match), it
1478 did strange things, either producing crazy output, or crashing. It should,
1479 of course, ignore a request for colour when reporting lines that do not
1480 match.
1481
1482 4. Another pcregrep bug caused similar problems if --colour was specified with
1483 -M (multiline) and the pattern match finished with a line ending.
1484
1485 5. In pcregrep, when a pattern that ended with a literal newline sequence was
1486 matched in multiline mode, the following line was shown as part of the
1487 match. This seems wrong, so I have changed it.
1488
1489 6. Another pcregrep bug in multiline mode, when --colour was specified, caused
1490 the check for further matches in the same line (so they could be coloured)
1491 to overrun the end of the current line. If another match was found, it was
1492 incorrectly shown (and then shown again when found in the next line).
1493
1494 7. If pcregrep was compiled under Windows, there was a reference to the
1495 function pcregrep_exit() before it was defined. I am assuming this was
1496 the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
1497 reported by a user. I've moved the definition above the reference.
1498
1499
1500 Version 8.11 10-Dec-2010
1501 ------------------------
1502
1503 1. (*THEN) was not working properly if there were untried alternatives prior
1504 to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
1505 backtracked to try for "b" instead of moving to the next alternative branch
1506 at the same level (in this case, to look for "c"). The Perl documentation
1507 is clear that when (*THEN) is backtracked onto, it goes to the "next
1508 alternative in the innermost enclosing group".
1509
1510 2. (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
1511 such as (A(*COMMIT)B(*THEN)C|D) any failure after matching A should
1512 result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
1513 (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
1514 (*THEN).
1515
1516 3. If \s appeared in a character class, it removed the VT character from
1517 the class, even if it had been included by some previous item, for example
1518 in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
1519 of \s, but is part of the POSIX "space" class.)
1520
1521 4. A partial match never returns an empty string (because you can always
1522 match an empty string at the end of the subject); however the checking for
1523 an empty string was starting at the "start of match" point. This has been
1524 changed to the "earliest inspected character" point, because the returned
1525 data for a partial match starts at this character. This means that, for
1526 example, /(?<=abc)def/ gives a partial match for the subject "abc"
1527 (previously it gave "no match").
1528
1529 5. Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
1530 of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
1531 previously a full match would be given. However, setting PCRE_PARTIAL_HARD
1532 has an implication that the given string is incomplete (because a partial
1533 match is preferred over a full match). For this reason, these items now
1534 give a partial match in this situation. [Aside: previously, the one case
1535 /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
1536 match rather than a full match, which was wrong by the old rules, but is
1537 now correct.]
1538
1539 6. There was a bug in the handling of #-introduced comments, recognized when
1540 PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
1541 If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
1542 UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
1543 scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
1544 but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
1545 places in pcre_compile().
1546
1547 7. Related to (6) above, when pcre_compile() was skipping #-introduced
1548 comments when looking ahead for named forward references to subpatterns,
1549 the only newline sequence it recognized was NL. It now handles newlines
1550 according to the set newline convention.
1551
1552 8. SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
1553 former, but used strtoul(), whereas pcretest avoided strtoul() but did not
1554 cater for a lack of strerror(). These oversights have been fixed.
1555
1556 9. Added --match-limit and --recursion-limit to pcregrep.
1557
1558 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
1559
1560 11. When the -o option was used, pcregrep was setting a return code of 1, even
1561 when matches were found, and --line-buffered was not being honoured.
1562
1563 12. Added an optional parentheses number to the -o and --only-matching options
1564 of pcregrep.
1565
1566 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
1567 can match an empty string. The code to do it in pcretest and pcredemo
1568 needed fixing:
1569
1570 (a) When the newline convention was "crlf", pcretest got it wrong, skipping
1571 only one byte after an empty string match just before CRLF (this case
1572 just got forgotten; "any" and "anycrlf" were OK).
1573
1574 (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
1575 mode when an empty string match preceded an ASCII character followed by
1576 a non-ASCII character. (The code for advancing by one character rather
1577 than one byte was nonsense.)
1578
1579 (c) The pcredemo.c sample program did not have any code at all to handle
1580 the cases when CRLF is a valid newline sequence.
1581
1582 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
1583 as a starting offset was within the subject string. There is now a new
1584 error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
1585 negative or greater than the length of the string. In order to test this,
1586 pcretest is extended to allow the setting of negative starting offsets.
1587
1588 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
1589 starting offset points to the beginning of a UTF-8 character was
1590 unnecessarily clumsy. I tidied it up.
1591
1592 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
1593 bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
1594
1595 17. Nobody had reported that the --include_dir option, which was added in
1596 release 7.7 should have been called --include-dir (hyphen, not underscore)
1597 for compatibility with GNU grep. I have changed it to --include-dir, but
1598 left --include_dir as an undocumented synonym, and the same for
1599 --exclude-dir, though that is not available in GNU grep, at least as of
1600 release 2.5.4.
1601
1602 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
1603 characters from a string of bytes) have been redefined so as not to use
1604 loops, in order to improve performance in some environments. At the same
1605 time, I abstracted some of the common code into auxiliary macros to save
1606 repetition (this should not affect the compiled code).
1607
1608 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
1609 compile-time error is now given if \c is not followed by an ASCII
1610 character, that is, a byte less than 128. (In EBCDIC mode, the code is
1611 different, and any byte value is allowed.)
1612
1613 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
1614 START_OPTIMIZE option, which is now allowed at compile time - but just
1615 passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
1616 to pcregrep and other applications that have no direct access to PCRE
1617 options. The new /Y option in pcretest sets this option when calling
1618 pcre_compile().
1619
1620 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
1621 back references. Groups containing recursive back references were forced to
1622 be atomic by that change, but in the case of named groups, the amount of
1623 memory required was incorrectly computed, leading to "Failed: internal
1624 error: code overflow". This has been fixed.
1625
1626 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
1627 pcretest.c, to avoid build problems in some Borland environments.
1628
1629
1630 Version 8.10 25-Jun-2010
1631 ------------------------
1632
1633 1. Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
1634 THEN.
1635
1636 2. (*ACCEPT) was not working when inside an atomic group.
1637
1638 3. Inside a character class, \B is treated as a literal by default, but
1639 faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
1640 causes the error). The code is unchanged, but I tidied the documentation.
1641
1642 4. Inside a character class, PCRE always treated \R and \X as literals,
1643 whereas Perl faults them if its -w option is set. I have changed PCRE so
1644 that it faults them when PCRE_EXTRA is set.
1645
1646 5. Added support for \N, which always matches any character other than
1647 newline. (It is the same as "." when PCRE_DOTALL is not set.)
1648
1649 6. When compiling pcregrep with newer versions of gcc which may have
1650 FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
1651 declared with attribute warn_unused_result" were given. Just casting the
1652 result to (void) does not stop the warnings; a more elaborate fudge is
1653 needed. I've used a macro to implement this.
1654
1655 7. Minor change to pcretest.c to avoid a compiler warning.
1656
1657 8. Added four artifical Unicode properties to help with an option to make
1658 \s etc use properties (see next item). The new properties are: Xan
1659 (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
1660
1661 9. Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
1662 use Unicode properties. (*UCP) at the start of a pattern can be used to set
1663 this option. Modified pcretest to add /W to test this facility. Added
1664 REG_UCP to make it available via the POSIX interface.
1665
1666 10. Added --line-buffered to pcregrep.
1667
1668 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
1669 studied, and the match started with a letter with a code point greater than
1670 127 whose first byte was different to the first byte of the other case of
1671 the letter, the other case of this starting letter was not recognized
1672 (#976).
1673
1674 12. If a pattern that was studied started with a repeated Unicode property
1675 test, for example, \p{Nd}+, there was the theoretical possibility of
1676 setting up an incorrect bitmap of starting bytes, but fortunately it could
1677 not have actually happened in practice until change 8 above was made (it
1678 added property types that matched character-matching opcodes).
1679
1680 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
1681 possible starting bytes for non-anchored patterns.
1682
1683 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
1684 \R, and also a number of cases that involve Unicode properties, both
1685 explicit and implicit when PCRE_UCP is set.
1686
1687 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
1688 input, it could crash or give wrong results if characters with values
1689 greater than 0xc0 were present in the subject string. (Detail: it assumed
1690 UTF-8 input when processing these items.)
1691
1692 16. Added a lot of (int) casts to avoid compiler warnings in systems where
1693 size_t is 64-bit (#991).
1694
1695 17. Added a check for running out of memory when PCRE is compiled with
1696 --disable-stack-for-recursion (#990).
1697
1698 18. If the last data line in a file for pcretest does not have a newline on
1699 the end, a newline was missing in the output.
1700
1701 19. The default pcre_chartables.c file recognizes only ASCII characters (values
1702 less than 128) in its various bitmaps. However, there is a facility for
1703 generating tables according to the current locale when PCRE is compiled. It
1704 turns out that in some environments, 0x85 and 0xa0, which are Unicode space
1705 characters, are recognized by isspace() and therefore were getting set in
1706 these tables, and indeed these tables seem to approximate to ISO 8859. This
1707 caused a problem in UTF-8 mode when pcre_study() was used to create a list
1708 of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
1709 which of course cannot start UTF-8 characters. I have changed the code so
1710 that only real ASCII characters (less than 128) and the correct starting
1711 bytes for UTF-8 encodings are set for characters greater than 127 when in
1712 UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
1713 altogether.)
1714
1715 20. Added the /T option to pcretest so as to be able to run tests with non-
1716 standard character tables, thus making it possible to include the tests
1717 used for 19 above in the standard set of tests.
1718
1719 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
1720 reference to a subpattern the other side of a comment that contains an
1721 opening parenthesis caused either an internal compiling error, or a
1722 reference to the wrong subpattern.
1723
1724
1725 Version 8.02 19-Mar-2010
1726 ------------------------
1727
1728 1. The Unicode data tables have been updated to Unicode 5.2.0.
1729
1730 2. Added the option --libs-cpp to pcre-config, but only when C++ support is
1731 configured.
1732
1733 3. Updated the licensing terms in the pcregexp.pas file, as agreed with the
1734 original author of that file, following a query about its status.
1735
1736 4. On systems that do not have stdint.h (e.g. Solaris), check for and include
1737 inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
1738
1739 5. A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
1740 quantifier applied to a forward-referencing subroutine call, could compile
1741 incorrect code or give the error "internal error: previously-checked
1742 referenced subpattern not found".
1743
1744 6. Both MS Visual Studio and Symbian OS have problems with initializing
1745 variables to point to external functions. For these systems, therefore,
1746 pcre_malloc etc. are now initialized to local functions that call the
1747 relevant global functions.
1748
1749 7. There were two entries missing in the vectors called coptable and poptable
1750 in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
1751 I've fixed the data, and added a kludgy way of testing at compile time that
1752 the lengths are correct (equal to the number of opcodes).
1753
1754 8. Following on from 7, I added a similar kludge to check the length of the
1755 eint vector in pcreposix.c.
1756
1757 9. Error texts for pcre_compile() are held as one long string to avoid too
1758 much relocation at load time. To find a text, the string is searched,
1759 counting zeros. There was no check for running off the end of the string,
1760 which could happen if a new error number was added without updating the
1761 string.
1762
1763 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
1764
1765 11. \K was not working if it appeared in an atomic group or in a group that
1766 was called as a "subroutine", or in an assertion. Perl 5.11 documents that
1767 \K is "not well defined" if used in an assertion. PCRE now accepts it if
1768 the assertion is positive, but not if it is negative.
1769
1770 12. Change 11 fortuitously reduced the size of the stack frame used in the
1771 "match()" function of pcre_exec.c by one pointer. Forthcoming
1772 implementation of support for (*MARK) will need an extra pointer on the
1773 stack; I have reserved it now, so that the stack frame size does not
1774 decrease.
1775
1776 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
1777 item in branch that calls a recursion is a subroutine call - as in the
1778 second branch in the above example - was incorrectly given the compile-
1779 time error "recursive call could loop indefinitely" because pcre_compile()
1780 was not correctly checking the subroutine for matching a non-empty string.
1781
1782 14. The checks for overrunning compiling workspace could trigger after an
1783 overrun had occurred. This is a "should never occur" error, but it can be
1784 triggered by pathological patterns such as hundreds of nested parentheses.
1785 The checks now trigger 100 bytes before the end of the workspace.
1786
1787 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
1788
1789
1790 Version 8.01 19-Jan-2010
1791 ------------------------
1792
1793 1. If a pattern contained a conditional subpattern with only one branch (in
1794 particular, this includes all (*DEFINE) patterns), a call to pcre_study()
1795 computed the wrong minimum data length (which is of course zero for such
1796 subpatterns). This could cause incorrect "no match" results.
1797
1798 2. For patterns such as (?i)a(?-i)b|c where an option setting at the start of
1799 the pattern is reset in the first branch, pcre_compile() failed with
1800 "internal error: code overflow at offset...". This happened only when
1801 the reset was to the original external option setting. (An optimization
1802 abstracts leading options settings into an external setting, which was the
1803 cause of this.)
1804
1805 3. A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
1806 of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
1807 assertion pattern did not match (meaning that the assertion was true), it
1808 was incorrectly treated as false if the SKIP had been reached during the
1809 matching. This also applied to assertions used as conditions.
1810
1811 4. If an item that is not supported by pcre_dfa_exec() was encountered in an
1812 assertion subpattern, including such a pattern used as a condition,
1813 unpredictable results occurred, instead of the error return
1814 PCRE_ERROR_DFA_UITEM.
1815
1816 5. The C++ GlobalReplace function was not working like Perl for the special
1817 situation when an empty string is matched. It now does the fancy magic
1818 stuff that is necessary.
1819
1820 6. In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
1821 removed. (These were left over from very, very early versions of PCRE.)
1822
1823 7. Some cosmetic changes to the code to make life easier when compiling it
1824 as part of something else:
1825
1826 (a) Change DEBUG to PCRE_DEBUG.
1827
1828 (b) In pcre_compile(), rename the member of the "branch_chain" structure
1829 called "current" as "current_branch", to prevent a collision with the
1830 Linux macro when compiled as a kernel module.
1831
1832 (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
1833 prevent a collision with the Linux macro when compiled as a kernel
1834 module.
1835
1836 8. In pcre_compile() there are some checks for integer overflows that used to
1837 cast potentially large values to (double). This has been changed to that
1838 when building, a check for int64_t is made, and if it is found, it is used
1839 instead, thus avoiding the use of floating point arithmetic. (There is no
1840 other use of FP in PCRE.) If int64_t is not found, the fallback is to
1841 double.
1842
1843 9. Added two casts to avoid signed/unsigned warnings from VS Studio Express
1844 2005 (difference between two addresses compared to an unsigned value).
1845
1846 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
1847 custom one, because of the following reported problem in Windows:
1848
1849 - libbz2 uses the Pascal calling convention (WINAPI) for the functions
1850 under Win32.
1851 - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
1852 therefore missing the function definition.
1853 - The compiler thus generates a "C" signature for the test function.
1854 - The linker fails to find the "C" function.
1855 - PCRE fails to configure if asked to do so against libbz2.
1856
1857 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
1858 messages were output:
1859
1860 Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
1861 rerunning libtoolize, to keep the correct libtool macros in-tree.
1862 Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
1863
1864 I have done both of these things.
1865
1866 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
1867 most of the time, it *can* run out if it is given a pattern that contains a
1868 runaway infinite recursion. I updated the discussion in the pcrestack man
1869 page.
1870
1871 13. Now that we have gone to the x.xx style of version numbers, the minor
1872 version may start with zero. Using 08 or 09 is a bad idea because users
1873 might check the value of PCRE_MINOR in their code, and 08 or 09 may be
1874 interpreted as invalid octal numbers. I've updated the previous comment in
1875 configure.ac, and also added a check that gives an error if 08 or 09 are
1876 used.
1877
1878 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
1879 causing partial matching to fail when the end of the subject matched \W
1880 in a UTF-8 pattern where \W was quantified with a minimum of 3.
1881
1882 15. There were some discrepancies between the declarations in pcre_internal.h
1883 of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
1884 their definitions. The declarations used "const uschar *" and the
1885 definitions used USPTR. Even though USPTR is normally defined as "const
1886 unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
1887 reported that: "This difference in casting confuses some C++ compilers, for
1888 example, SunCC recognizes above declarations as different functions and
1889 generates broken code for hbpcre." I have changed the declarations to use
1890 USPTR.
1891
1892 16. GNU libtool is named differently on some systems. The autogen.sh script now
1893 tries several variants such as glibtoolize (MacOSX) and libtoolize1x
1894 (FreeBSD).
1895
1896 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
1897 (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
1898 comment: "Figure out how to create a longlong from a string: strtoll and
1899 equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
1900 instance, but it only takes 2 args instead of 3!"
1901
1902 18. A subtle bug concerned with back references has been fixed by a change of
1903 specification, with a corresponding code fix. A pattern such as
1904 ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
1905 refers, was giving matches when it shouldn't. For example, xa=xaaa would
1906 match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
1907 same bug. Such groups have to be quantified to be useful, or contained
1908 inside another quantified group. (If there's no repetition, the reference
1909 can never match.) The problem arises because, having left the group and
1910 moved on to the rest of the pattern, a later failure that backtracks into
1911 the group uses the captured value from the final iteration of the group
1912 rather than the correct earlier one. I have fixed this in PCRE by forcing
1913 any group that contains a reference to itself to be an atomic group; that
1914 is, there cannot be any backtracking into it once it has completed. This is
1915 similar to recursive and subroutine calls.
1916
1917
1918 Version 8.00 19-Oct-09
1919 ----------------------
1920
1921 1. The table for translating pcre_compile() error codes into POSIX error codes
1922 was out-of-date, and there was no check on the pcre_compile() error code
1923 being within the table. This could lead to an OK return being given in
1924 error.
1925
1926 2. Changed the call to open a subject file in pcregrep from fopen(pathname,
1927 "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
1928 in a Windows environment.
1929
1930 3. The pcregrep --count option prints the count for each file even when it is
1931 zero, as does GNU grep. However, pcregrep was also printing all files when
1932 --files-with-matches was added. Now, when both options are given, it prints
1933 counts only for those files that have at least one match. (GNU grep just
1934 prints the file name in this circumstance, but including the count seems
1935 more useful - otherwise, why use --count?) Also ensured that the
1936 combination -clh just lists non-zero counts, with no names.
1937
1938 4. The long form of the pcregrep -F option was incorrectly implemented as
1939 --fixed_strings instead of --fixed-strings. This is an incompatible change,
1940 but it seems right to fix it, and I didn't think it was worth preserving
1941 the old behaviour.
1942
1943 5. The command line items --regex=pattern and --regexp=pattern were not
1944 recognized by pcregrep, which required --regex pattern or --regexp pattern
1945 (with a space rather than an '='). The man page documented the '=' forms,
1946 which are compatible with GNU grep; these now work.
1947
1948 6. No libpcreposix.pc file was created for pkg-config; there was just
1949 libpcre.pc and libpcrecpp.pc. The omission has been rectified.
1950
1951 7. Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
1952 when UCP support is not needed, by modifying the Python script that
1953 generates it from Unicode data files. This should not matter if the module
1954 is correctly used as a library, but I received one complaint about 50K of
1955 unwanted data. My guess is that the person linked everything into his
1956 program rather than using a library. Anyway, it does no harm.
1957
1958 8. A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
1959 was a minimum greater than 1 for a wide character in a possessive
1960 repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
1961 which had an unlimited repeat of a nested, fixed maximum repeat of a wide
1962 character. Chaos in the form of incorrect output or a compiling loop could
1963 result.
1964
1965 9. The restrictions on what a pattern can contain when partial matching is
1966 requested for pcre_exec() have been removed. All patterns can now be
1967 partially matched by this function. In addition, if there are at least two
1968 slots in the offset vector, the offset of the earliest inspected character
1969 for the match and the offset of the end of the subject are set in them when
1970 PCRE_ERROR_PARTIAL is returned.
1971
1972 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
1973 synonymous with PCRE_PARTIAL, for backwards compatibility, and
1974 PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
1975 and may be more useful for multi-segment matching.
1976
1977 11. Partial matching with pcre_exec() is now more intuitive. A partial match
1978 used to be given if ever the end of the subject was reached; now it is
1979 given only if matching could not proceed because another character was
1980 needed. This makes a difference in some odd cases such as Z(*FAIL) with the
1981 string "Z", which now yields "no match" instead of "partial match". In the
1982 case of pcre_dfa_exec(), "no match" is given if every matching path for the
1983 final character ended with (*FAIL).
1984
1985 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
1986 if the pattern had a "must contain" character that was already found in the
1987 earlier partial match, unless partial matching was again requested. For
1988 example, with the pattern /dog.(body)?/, the "must contain" character is
1989 "g". If the first part-match was for the string "dog", restarting with
1990 "sbody" failed. This bug has been fixed.
1991
1992 13. The string returned by pcre_dfa_exec() after a partial match has been
1993 changed so that it starts at the first inspected character rather than the
1994 first character of the match. This makes a difference only if the pattern
1995 starts with a lookbehind assertion or \b or \B (\K is not supported by
1996 pcre_dfa_exec()). It's an incompatible change, but it makes the two
1997 matching functions compatible, and I think it's the right thing to do.
1998
1999 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
2000 so that the demonstration program is easily available in environments where
2001 PCRE has not been installed from source.
2002
2003 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
2004 libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
2005 library.
2006
2007 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
2008 It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
2009 is not the first non-POSIX option to be added. Clearly some people find
2010 these options useful.
2011
2012 17. If a caller to the POSIX matching function regexec() passes a non-zero
2013 value for nmatch with a NULL value for pmatch, the value of
2014 nmatch is forced to zero.
2015
2016 18. RunGrepTest did not have a test for the availability of the -u option of
2017 the diff command, as RunTest does. It now checks in the same way as
2018 RunTest, and also checks for the -b option.
2019
2020 19. If an odd number of negated classes containing just a single character
2021 interposed, within parentheses, between a forward reference to a named
2022 subpattern and the definition of the subpattern, compilation crashed with
2023 an internal error, complaining that it could not find the referenced
2024 subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
2025 [The bug was that it was starting one character too far in when skipping
2026 over the character class, thus treating the ] as data rather than
2027 terminating the class. This meant it could skip too much.]
2028
2029 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
2030 /g option in pcretest when the pattern contains \K, which makes it possible
2031 to have an empty string match not at the start, even when the pattern is
2032 anchored. Updated pcretest and pcredemo to use this option.
2033
2034 21. If the maximum number of capturing subpatterns in a recursion was greater
2035 than the maximum at the outer level, the higher number was returned, but
2036 with unset values at the outer level. The correct (outer level) value is
2037 now given.
2038
2039 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
2040 PCRE did not set those parentheses (unlike Perl). I have now found a way to
2041 make it do so. The string so far is captured, making this feature
2042 compatible with Perl.
2043
2044 23. The tests have been re-organized, adding tests 11 and 12, to make it
2045 possible to check the Perl 5.10 features against Perl 5.10.
2046
2047 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
2048 pattern matches a fixed length string. PCRE did not allow this; now it
2049 does. Neither allows recursion.
2050
2051 25. I finally figured out how to implement a request to provide the minimum
2052 length of subject string that was needed in order to match a given pattern.
2053 (It was back references and recursion that I had previously got hung up
2054 on.) This code has now been added to pcre_study(); it finds a lower bound
2055 to the length of subject needed. It is not necessarily the greatest lower
2056 bound, but using it to avoid searching strings that are too short does give
2057 some useful speed-ups. The value is available to calling programs via
2058 pcre_fullinfo().
2059
2060 26. While implementing 25, I discovered to my embarrassment that pcretest had
2061 not been passing the result of pcre_study() to pcre_dfa_exec(), so the
2062 study optimizations had never been tested with that matching function.
2063 Oops. What is worse, even when it was passed study data, there was a bug in
2064 pcre_dfa_exec() that meant it never actually used it. Double oops. There
2065 were also very few tests of studied patterns with pcre_dfa_exec().
2066
2067 27. If (?| is used to create subpatterns with duplicate numbers, they are now
2068 allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
2069 on the other side of the coin, they are no longer allowed to have different
2070 names, because these cannot be distinguished in PCRE, and this has caused
2071 confusion. (This is a difference from Perl.)
2072
2073 28. When duplicate subpattern names are present (necessarily with different
2074 numbers, as required by 27 above), and a test is made by name in a
2075 conditional pattern, either for a subpattern having been matched, or for
2076 recursion in such a pattern, all the associated numbered subpatterns are
2077 tested, and the overall condition is true if the condition is true for any
2078 one of them. This is the way Perl works, and is also more like the way
2079 testing by number works.
2080
2081
2082 Version 7.9 11-Apr-09
2083 ---------------------
2084
2085 1. When building with support for bzlib/zlib (pcregrep) and/or readline
2086 (pcretest), all targets were linked against these libraries. This included
2087 libpcre, libpcreposix, and libpcrecpp, even though they do not use these
2088 libraries. This caused unwanted dependencies to be created. This problem
2089 has been fixed, and now only pcregrep is linked with bzlib/zlib and only
2090 pcretest is linked with readline.
2091
2092 2. The "typedef int BOOL" in pcre_internal.h that was included inside the
2093 "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
2094 moved outside it again, because FALSE and TRUE are already defined in AIX,
2095 but BOOL is not.
2096
2097 3. The pcre_config() function was treating the PCRE_MATCH_LIMIT and
2098 PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
2099
2100 4. The pcregrep documentation said spaces were inserted as well as colons (or
2101 hyphens) following file names and line numbers when outputting matching
2102 lines. This is not true; no spaces are inserted. I have also clarified the
2103 wording for the --colour (or --color) option.
2104
2105 5. In pcregrep, when --colour was used with -o, the list of matching strings
2106 was not coloured; this is different to GNU grep, so I have changed it to be
2107 the same.
2108
2109 6. When --colo(u)r was used in pcregrep, only the first matching substring in
2110 each matching line was coloured. Now it goes on to look for further matches
2111 of any of the test patterns, which is the same behaviour as GNU grep.
2112
2113 7. A pattern that could match an empty string could cause pcregrep to loop; it
2114 doesn't make sense to accept an empty string match in pcregrep, so I have
2115 locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
2116 seems to be how GNU grep behaves. [But see later change 40 for release
2117 8.33.]
2118
2119 8. The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
2120 start or after a newline", because the conditional assertion was not being
2121 correctly handled. The rule now is that both the assertion and what follows
2122 in the first alternative must satisfy the test.
2123
2124 9. If auto-callout was enabled in a pattern with a conditional group whose
2125 condition was an assertion, PCRE could crash during matching, both with
2126 pcre_exec() and pcre_dfa_exec().
2127
2128 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
2129 used for matching.
2130
2131 11. Unicode property support in character classes was not working for
2132 characters (bytes) greater than 127 when not in UTF-8 mode.
2133
2134 12. Added the -M command line option to pcretest.
2135
2136 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
2137
2138 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
2139
2140 16. Added comments and documentation about mis-use of no_arg in the C++
2141 wrapper.
2142
2143 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
2144 from Martin Jerabek that uses macro names for all relevant character and
2145 string constants.
2146
2147 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
2148 SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
2149 SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
2150 these, but not everybody uses configure.
2151
2152 19. A conditional group that had only one branch was not being correctly
2153 recognized as an item that could match an empty string. This meant that an
2154 enclosing group might also not be so recognized, causing infinite looping
2155 (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
2156 with the subject "ab", where knowledge that the repeated group can match
2157 nothing is needed in order to break the loop.
2158
2159 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
2160 exec(), but without supplying a callout function, matching went wrong.
2161
2162 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
2163 leak if the size of the offset vector was greater than 30. When the vector
2164 is smaller, the saved offsets during recursion go onto a local stack
2165 vector, but for larger vectors malloc() is used. It was failing to free
2166 when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
2167 error, in fact).
2168
2169 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
2170 heapframe that is used only when UTF-8 support is enabled. This caused no
2171 problem, but was untidy.
2172
2173 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
2174 CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
2175 included within another project.
2176
2177 24. Steven Van Ingelgem's patches to add more options to the CMake support,
2178 slightly modified by me:
2179
2180 (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
2181 not building pcregrep.
2182
2183 (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
2184 if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
2185
2186 25. Forward references, both numeric and by name, in patterns that made use of
2187 duplicate group numbers, could behave incorrectly or give incorrect errors,
2188 because when scanning forward to find the reference group, PCRE was not
2189 taking into account the duplicate group numbers. A pattern such as
2190 ^X(?3)(a)(?|(b)|(q))(Y) is an example.
2191
2192 26. Changed a few more instances of "const unsigned char *" to USPTR, making
2193 the feature of a custom pointer more persuasive (as requested by a user).
2194
2195 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
2196 pcretest.c, inside #ifndefs, because it seems they are sometimes already
2197 pre-defined.
2198
2199 28. Added support for (*UTF8) at the start of a pattern.
2200
2201 29. Arrange for flags added by the "release type" setting in CMake to be shown
2202 in the configuration summary.
2203
2204
2205 Version 7.8 05-Sep-08
2206 ---------------------
2207
2208 1. Replaced UCP searching code with optimized version as implemented for Ad
2209 Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
2210 stage table and inline lookup instead of a function, giving speed ups of 2
2211 to 5 times on some simple patterns that I tested. Permission was given to
2212 distribute the MultiStage2.py script that generates the tables (it's not in
2213 the tarball, but is in the Subversion repository).
2214
2215 2. Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
2216 scripts.
2217
2218 3. Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
2219 a group with a zero qualifier. The result of the study could be incorrect,
2220 or the function might crash, depending on the pattern.
2221
2222 4. Caseless matching was not working for non-ASCII characters in back
2223 references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
2224 It now works when Unicode Property Support is available.
2225
2226 5. In pcretest, an escape such as \x{de} in the data was always generating
2227 a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
2228 non-UTF-8 mode. If the value is greater than 255, it gives a warning about
2229 truncation.
2230
2231 6. Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
2232
2233 7. Added two (int) casts to pcregrep when printing the difference of two
2234 pointers, in case they are 64-bit values.
2235
2236 8. Added comments about Mac OS X stack usage to the pcrestack man page and to
2237 test 2 if it fails.
2238
2239 9. Added PCRE_CALL_CONVENTION just before the names of all exported functions,
2240 and a #define of that name to empty if it is not externally set. This is to
2241 allow users of MSVC to set it if necessary.
2242
2243 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
2244 the convenience functions in the pcre_get.c source file.
2245
2246 11. An option change at the start of a pattern that had top-level alternatives
2247 could cause overwriting and/or a crash. This command provoked a crash in
2248 some environments:
2249
2250 printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
2251
2252 This potential security problem was recorded as CVE-2008-2371.
2253
2254 12. For a pattern where the match had to start at the beginning or immediately
2255 after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
2256 pcre_dfa_exec() could read past the end of the passed subject if there was
2257 no match. To help with detecting such bugs (e.g. with valgrind), I modified
2258 pcretest so that it places the subject at the end of its malloc-ed buffer.
2259
2260 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
2261 exec() might read past the end of the data buffer in UTF-8 mode.
2262
2263 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
2264 the data contained the byte 0x85 as part of a UTF-8 character within its
2265 first line. This applied both to normal and DFA matching.
2266
2267 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
2268 /^[^d]*?$/8 failed to match "abc".
2269
2270 16. Added a missing copyright notice to pcrecpp_internal.h.
2271
2272 17. Make it more clear in the documentation that values returned from
2273 pcre_exec() in ovector are byte offsets, not character counts.
2274
2275 18. Tidied a few places to stop certain compilers from issuing warnings.
2276
2277 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
2278 supplied by Stefan Weber. I made a further small update for 7.8 because
2279 there is a change of source arrangements: the pcre_searchfuncs.c module is
2280 replaced by pcre_ucd.c.
2281
2282
2283 Version 7.7 07-May-08
2284 ---------------------
2285
2286 1. Applied Craig's patch to sort out a long long problem: "If we can't convert
2287 a string to a long long, pretend we don't even have a long long." This is
2288 done by checking for the strtoq, strtoll, and _strtoi64 functions.
2289
2290 2. Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
2291 pre-7.6 versions, which defined a global no_arg variable instead of putting
2292 it in the RE class. (See also #8 below.)
2293
2294 3. Remove a line of dead code, identified by coverity and reported by Nuno
2295 Lopes.
2296
2297 4. Fixed two related pcregrep bugs involving -r with --include or --exclude:
2298
2299 (1) The include/exclude patterns were being applied to the whole pathnames
2300 of files, instead of just to the final components.
2301
2302 (2) If there was more than one level of directory, the subdirectories were
2303 skipped unless they satisfied the include/exclude conditions. This is
2304 inconsistent with GNU grep (and could even be seen as contrary to the
2305 pcregrep specification - which I improved to make it absolutely clear).
2306 The action now is always to scan all levels of directory, and just
2307 apply the include/exclude patterns to regular files.
2308
2309 5. Added the --include_dir and --exclude_dir patterns to pcregrep, and used
2310 --exclude_dir in the tests to avoid scanning .svn directories.
2311
2312 6. Applied Craig's patch to the QuoteMeta function so that it escapes the
2313 NUL character as backslash + 0 rather than backslash + NUL, because PCRE
2314 doesn't support NULs in patterns.
2315
2316 7. Added some missing "const"s to declarations of static tables in
2317 pcre_compile.c and pcre_dfa_exec.c.
2318
2319 8. Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
2320 caused by fix #2 above. (Subsequently also a second patch to fix the
2321 first patch. And a third patch - this was a messy problem.)
2322
2323 9. Applied Craig's patch to remove the use of push_back().
2324
2325 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
2326 matching function regexec().
2327
2328 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
2329 which, however, unlike Perl's \g{...}, are subroutine calls, not back
2330 references. PCRE supports relative numbers with this syntax (I don't think
2331 Oniguruma does).
2332
2333 12. Previously, a group with a zero repeat such as (...){0} was completely
2334 omitted from the compiled regex. However, this means that if the group
2335 was called as a subroutine from elsewhere in the pattern, things went wrong
2336 (an internal error was given). Such groups are now left in the compiled
2337 pattern, with a new opcode that causes them to be skipped at execution
2338 time.
2339
2340 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
2341 to the way PCRE behaves:
2342
2343 (a) A lone ] character is dis-allowed (Perl treats it as data).
2344
2345 (b) A back reference to an unmatched subpattern matches an empty string
2346 (Perl fails the current match path).
2347
2348 (c) A data ] in a character class must be notated as \] because if the
2349 first data character in a class is ], it defines an empty class. (In
2350 Perl it is not possible to have an empty class.) The empty class []
2351 never matches; it forces failure and is equivalent to (*FAIL) or (?!).
2352 The negative empty class [^] matches any one character, independently
2353 of the DOTALL setting.
2354
2355 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
2356 non-existent subpattern following a character class starting with ']' and
2357 containing () gave an internal compiling error instead of "reference to
2358 non-existent subpattern". Fortunately, when the pattern did exist, the
2359 compiled code was correct. (When scanning forwards to check for the
2360 existence of the subpattern, it was treating the data ']' as terminating
2361 the class, so got the count wrong. When actually compiling, the reference
2362 was subsequently set up correctly.)
2363
2364 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
2365 it was being rejected as not supported by pcre_dfa_exec(), even though
2366 other assertions are supported. I have made pcre_dfa_exec() support
2367 (*FAIL).
2368
2369 16. The implementation of 13c above involved the invention of a new opcode,
2370 OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
2371 cannot be changed at match time, I realized I could make a small
2372 improvement to matching performance by compiling OP_ALLANY instead of
2373 OP_ANY for "." when DOTALL was set, and then removing the runtime tests
2374 on the OP_ANY path.
2375
2376 17. Compiling pcretest on Windows with readline support failed without the
2377 following two fixes: (1) Make the unistd.h include conditional on
2378 HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
2379
2380 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
2381 ncurses library to be included for pcretest when ReadLine support is
2382 requested, but also to allow for it to be overridden. This patch came from
2383 Daniel Bergström.
2384
2385 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
2386 as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
2387 any errors with the current Unicode tables. Thanks to Peter Kankowski for
2388 spotting this.
2389
2390
2391 Version 7.6 28-Jan-08
2392 ---------------------
2393
2394 1. A character class containing a very large number of characters with
2395 codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
2396 overflow.
2397
2398 2. Patch to cut out the "long long" test in pcrecpp_unittest when
2399 HAVE_LONG_LONG is not defined.
2400
2401 3. Applied Christian Ehrlicher's patch to update the CMake build files to
2402 bring them up to date and include new features. This patch includes:
2403
2404 - Fixed PH's badly added libz and libbz2 support.
2405 - Fixed a problem with static linking.
2406 - Added pcredemo. [But later removed - see 7 below.]
2407 - Fixed dftables problem and added an option.
2408 - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
2409 HAVE_LONG_LONG.
2410 - Added readline support for pcretest.
2411 - Added an listing of the option settings after cmake has run.
2412
2413 4. A user submitted a patch to Makefile that makes it easy to create
2414 "pcre.dll" under mingw when using Configure/Make. I added stuff to
2415 Makefile.am that cause it to include this special target, without
2416 affecting anything else. Note that the same mingw target plus all
2417 the other distribution libraries and programs are now supported
2418 when configuring with CMake (see 6 below) instead of with
2419 Configure/Make.
2420
2421 5. Applied Craig's patch that moves no_arg into the RE class in the C++ code.
2422 This is an attempt to solve the reported problem "pcrecpp::no_arg is not
2423 exported in the Windows port". It has not yet been confirmed that the patch
2424 solves the problem, but it does no harm.
2425
2426 6. Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
2427 NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
2428 with CMake, and also correct the comment about stack recursion.
2429
2430 7. Remove the automatic building of pcredemo from the ./configure system and
2431 from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
2432 of a program that users should build themselves after PCRE is installed, so
2433 building it automatically is not really right. What is more, it gave
2434 trouble in some build environments.
2435
2436 8. Further tidies to CMakeLists.txt from Sheri and Christian.
2437
2438
2439 Version 7.5 10-Jan-08
2440 ---------------------
2441
2442 1. Applied a patch from Craig: "This patch makes it possible to 'ignore'
2443 values in parens when parsing an RE using the C++ wrapper."
2444
2445 2. Negative specials like \S did not work in character classes in UTF-8 mode.
2446 Characters greater than 255 were excluded from the class instead of being
2447 included.
2448
2449 3. The same bug as (2) above applied to negated POSIX classes such as
2450 [:^space:].
2451
2452 4. PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
2453 defined or documented. It seems to have been a typo for PCRE_STATIC, so
2454 I have changed it.
2455
2456 5. The construct (?&) was not diagnosed as a syntax error (it referenced the
2457 first named subpattern) and a construct such as (?&a) would reference the
2458 first named subpattern whose name started with "a" (in other words, the
2459 length check was missing). Both these problems are fixed. "Subpattern name
2460 expected" is now given for (?&) (a zero-length name), and this patch also
2461 makes it give the same error for \k'' (previously it complained that that
2462 was a reference to a non-existent subpattern).
2463
2464 6. The erroneous patterns (?+-a) and (?-+a) give different error messages;
2465 this is right because (?- can be followed by option settings as well as by
2466 digits. I have, however, made the messages clearer.
2467
2468 7. Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
2469 than the number used in the conditional) now cause a compile-time error.
2470 This is actually not compatible with Perl, which accepts such patterns, but
2471 treats the conditional as always being FALSE (as PCRE used to), but it
2472 seems to me that giving a diagnostic is better.
2473
2474 8. Change "alphameric" to the more common word "alphanumeric" in comments
2475 and messages.
2476
2477 9. Fix two occurrences of "backslash" in comments that should have been
2478 "backspace".
2479
2480 10. Remove two redundant lines of code that can never be obeyed (their function
2481 was moved elsewhere).
2482
2483 11. The program that makes PCRE's Unicode character property table had a bug
2484 which caused it to generate incorrect table entries for sequences of
2485 characters that have the same character type, but are in different scripts.
2486 It amalgamated them into a single range, with the script of the first of
2487 them. In other words, some characters were in the wrong script. There were
2488 thirteen such cases, affecting characters in the following ranges:
2489
2490 U+002b0 - U+002c1
2491 U+0060c - U+0060d
2492 U+0061e - U+00612
2493 U+0064b - U+0065e
2494 U+0074d - U+0076d
2495 U+01800 - U+01805
2496 U+01d00 - U+01d77
2497 U+01d9b - U+01dbf
2498 U+0200b - U+0200f
2499 U+030fc - U+030fe
2500 U+03260 - U+0327f
2501 U+0fb46 - U+0fbb1
2502 U+10450 - U+1049d
2503
2504 12. The -o option (show only the matching part of a line) for pcregrep was not
2505 compatible with GNU grep in that, if there was more than one match in a
2506 line, it showed only the first of them. It now behaves in the same way as
2507 GNU grep.
2508
2509 13. If the -o and -v options were combined for pcregrep, it printed a blank
2510 line for every non-matching line. GNU grep prints nothing, and pcregrep now
2511 does the same. The return code can be used to tell if there were any
2512 non-matching lines.
2513
2514 14. Added --file-offsets and --line-offsets to pcregrep.
2515
2516 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
2517 infinitely looping recursion. The bug was that positive lookaheads were not
2518 being skipped when checking for a possible empty match (negative lookaheads
2519 and both kinds of lookbehind were skipped).
2520
2521 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
2522 inclusion of <windows.h> to before rather than after the definition of
2523 INVALID_FILE_ATTRIBUTES (patch from David Byron).
2524
2525 17. Specifying a possessive quantifier with a specific limit for a Unicode
2526 character property caused pcre_compile() to compile bad code, which led at
2527 runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
2528 are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
2529 caused the error; without that there was no problem.
2530
2531 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
2532
2533 19. Added --enable-pcretest-libreadline.
2534
2535 20. In pcrecpp.cc, the variable 'count' was incremented twice in
2536 RE::GlobalReplace(). As a result, the number of replacements returned was
2537 double what it should be. I removed one of the increments, but Craig sent a
2538 later patch that removed the other one (the right fix) and added unit tests
2539 that check the return values (which was not done before).
2540
2541 21. Several CMake things:
2542
2543 (1) Arranged that, when cmake is used on Unix, the libraries end up with
2544 the names libpcre and libpcreposix, not just pcre and pcreposix.
2545
2546 (2) The above change means that pcretest and pcregrep are now correctly
2547 linked with the newly-built libraries, not previously installed ones.
2548
2549 (3) Added PCRE_SUPPORT_LIBREADLINE, PCRE_SUPPORT_LIBZ, PCRE_SUPPORT_LIBBZ2.
2550
2551 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
2552 crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
2553 UTF-8 newline character). The key issue is that the pattern starts .*;
2554 this means that the match must be either at the beginning, or after a
2555 newline. The bug was in the code for advancing after a failed match and
2556 checking that the new position followed a newline. It was not taking
2557 account of UTF-8 characters correctly.
2558
2559 23. PCRE was behaving differently from Perl in the way it recognized POSIX
2560 character classes. PCRE was not treating the sequence [:...:] as a
2561 character class unless the ... were all letters. Perl, however, seems to
2562 allow any characters between [: and :], though of course it rejects as
2563 unknown any "names" that contain non-letters, because all the known class
2564 names consist only of letters. Thus, Perl gives an error for [[:1234:]],
2565 for example, whereas PCRE did not - it did not recognize a POSIX character
2566 class. This seemed a bit dangerous, so the code has been changed to be
2567 closer to Perl. The behaviour is not identical to Perl, because PCRE will
2568 diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
2569 treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
2570 Perl does, and where it didn't before.
2571
2572 24. Rewrite so as to remove the single use of %n from pcregrep because in some
2573 Windows environments %n is disabled by default.
2574
2575
2576 Version 7.4 21-Sep-07
2577 ---------------------
2578
2579 1. Change 7.3/28 was implemented for classes by looking at the bitmap. This
2580 means that a class such as [\s] counted as "explicit reference to CR or
2581 LF". That isn't really right - the whole point of the change was to try to
2582 help when there was an actual mention of one of the two characters. So now
2583 the change happens only if \r or \n (or a literal CR or LF) character is
2584 encountered.
2585
2586 2. The 32-bit options word was also used for 6 internal flags, but the numbers
2587 of both had grown to the point where there were only 3 bits left.
2588 Fortunately, there was spare space in the data structure, and so I have
2589 moved the internal flags into a new 16-bit field to free up more option
2590 bits.
2591
2592 3. The appearance of (?J) at the start of a pattern set the DUPNAMES option,
2593 but did not set the internal JCHANGED flag - either of these is enough to
2594 control the way the "get" function works - but the PCRE_INFO_JCHANGED
2595 facility is supposed to tell if (?J) was ever used, so now (?J) at the
2596 start sets both bits.
2597
2598 4. Added options (at build time, compile time, exec time) to change \R from
2599 matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
2600
2601 5. doc/pcresyntax.html was missing from the distribution.
2602
2603 6. Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
2604 compatibility, even though it is no longer used.
2605
2606 7. Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
2607 strtoull to pcrecpp.cc to select the available functions in WIN32 when the
2608 windows.h file is present (where different names are used). [This was
2609 reversed later after testing - see 16 below.]
2610
2611 8. Changed all #include <config.h> to #include "config.h". There were also
2612 some further <pcre.h> cases that I changed to "pcre.h".
2613
2614 9. When pcregrep was used with the --colour option, it missed the line ending
2615 sequence off the lines that it output.
2616
2617 10. It was pointed out to me that arrays of string pointers cause lots of
2618 relocations when a shared library is dynamically loaded. A technique of
2619 using a single long string with a table of offsets can drastically reduce
2620 these. I have refactored PCRE in four places to do this. The result is
2621 dramatic:
2622
2623 Originally: 290
2624 After changing UCP table: 187
2625 After changing error message table: 43
2626 After changing table of "verbs" 36
2627 After changing table of Posix names 22
2628
2629 Thanks to the folks working on Gregex for glib for this insight.
2630
2631 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
2632 unicode-properties was also set.
2633
2634 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
2635
2636 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
2637 checked only for CRLF.
2638
2639 14. Added casts to pcretest.c to avoid compiler warnings.
2640
2641 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
2642
2643 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
2644 and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
2645 entirely. This removes changes made in 7 above.
2646
2647 17. The CMake files have been updated, and there is now more information about
2648 building with CMake in the NON-UNIX-USE document.
2649
2650
2651 Version 7.3 28-Aug-07
2652 ---------------------
2653
2654 1. In the rejigging of the build system that eventually resulted in 7.1, the
2655 line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
2656 brackets there is not right, since it causes compilers to look for an
2657 installed pcre.h, not the version that is in the source that is being
2658 compiled (which of course may be different). I have changed it back to:
2659
2660 #include "pcre.h"
2661
2662 I have a vague recollection that the change was concerned with compiling in
2663 different directories, but in the new build system, that is taken care of
2664 by the VPATH setting the Makefile.
2665
2666 2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
2667 when the subject happened to end in the byte 0x85 (e.g. if the last
2668 character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
2669 characters but of course it shouldn't be taken as a newline when it is part
2670 of another character. The bug was that, for an unlimited repeat of . in
2671 not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
2672 characters when looking for a newline.
2673
2674 3. A small performance improvement in the DOTALL UTF-8 mode .* case.
2675
2676 4. Debugging: adjusted the names of opcodes for different kinds of parentheses
2677 in debug output.
2678
2679 5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
2680 long printing in the pcrecpp unittest when running under MinGW.
2681
2682 6. ESC_K was left out of the EBCDIC table.
2683
2684 7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
2685 parentheses; I made it 1000, which seemed large enough. Unfortunately, the
2686 limit also applies to "virtual nesting" when a pattern is recursive, and in
2687 this case 1000 isn't so big. I have been able to remove this limit at the
2688 expense of backing off one optimization in certain circumstances. Normally,
2689 when pcre_exec() would call its internal match() function recursively and
2690 immediately return the result unconditionally, it uses a "tail recursion"
2691 feature to save stack. However, when a subpattern that can match an empty
2692 string has an unlimited repetition quantifier, it no longer makes this
2693 optimization. That gives it a stack frame in which to save the data for
2694 checking that an empty string has been matched. Previously this was taken
2695 from the 1000-entry workspace that had been reserved. So now there is no
2696 explicit limit, but more stack is used.
2697
2698 8. Applied Daniel's patches to solve problems with the import/export magic
2699 syntax that is required for Windows, and which was going wrong for the
2700 pcreposix and pcrecpp parts of the library. These were overlooked when this
2701 problem was solved for the main library.
2702
2703 9. There were some crude static tests to avoid integer overflow when computing
2704 the size of patterns that contain repeated groups with explicit upper
2705 limits. As the maximum quantifier is 65535, the maximum group length was
2706 set at 30,000 so that the product of these two numbers did not overflow a
2707 32-bit integer. However, it turns out that people want to use groups that
2708 are longer than 30,000 bytes (though not repeat them that many times).
2709 Change 7.0/17 (the refactoring of the way the pattern size is computed) has
2710 made it possible to implement the integer overflow checks in a much more
2711 dynamic way, which I have now done. The artificial limitation on group
2712 length has been removed - we now have only the limit on the total length of
2713 the compiled pattern, which depends on the LINK_SIZE setting.
2714
2715 10. Fixed a bug in the documentation for get/copy named substring when
2716 duplicate names are permitted. If none of the named substrings are set, the
2717 functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
2718 empty string.
2719
2720 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
2721 instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
2722 because the ] is interpreted as the first data character and the
2723 terminating ] is not found. PCRE has been made compatible with Perl in this
2724 regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
2725 cause memory overwriting.
2726
2727 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
2728 string has been matched (to stop an infinite loop). It was not recognizing
2729 a conditional subpattern that could match an empty string if that
2730 subpattern was within another subpattern. For example, it looped when
2731 trying to match (((?(1)X|))*) but it was OK with ((?(1)X|)*) where the
2732 condition was not nested. This bug has been fixed.
2733
2734 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
2735 past the start of the subject in the presence of bytes with the top bit
2736 set, for example "\x8aBCD".
2737
2738 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
2739 (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
2740
2741 14. Optimized (?!) to (*FAIL).
2742
2743 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
2744 This restricts code points to be within the range 0 to 0x10FFFF, excluding
2745 the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
2746 full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
2747 does: it's just the validity check that is more restrictive.
2748
2749 16. Inserted checks for integer overflows during escape sequence (backslash)
2750 processing, and also fixed erroneous offset values for syntax errors during
2751 backslash processing.
2752
2753 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
2754 for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
2755
2756 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
2757 caused an overrun.
2758
2759 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
2760 something other than just ASCII characters) inside a group that had an
2761 unlimited repeat caused a loop at compile time (while checking to see
2762 whether the group could match an empty string).
2763
2764 20. Debugging a pattern containing \p or \P could cause a crash. For example,
2765 [\P{Any}] did so. (Error in the code for printing property names.)
2766
2767 21. An orphan \E inside a character class could cause a crash.
2768
2769 22. A repeated capturing bracket such as (A)? could cause a wild memory
2770 reference during compilation.
2771
2772 23. There are several functions in pcre_compile() that scan along a compiled
2773 expression for various reasons (e.g. to see if it's fixed length for look
2774 behind). There were bugs in these functions when a repeated \p or \P was
2775 present in the pattern. These operators have additional parameters compared
2776 with \d, etc, and these were not being taken into account when moving along
2777 the compiled data. Specifically:
2778
2779 (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
2780 length.
2781
2782 (b) An item such as \pL+ within a repeated group could cause crashes or
2783 loops.
2784
2785 (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
2786 "reference to non-existent subpattern" error.
2787
2788 (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
2789
2790 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
2791 characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
2792
2793 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
2794
2795 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
2796 character were causing crashes (broken optimization).
2797
2798 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
2799 \p or \P) caused a compile-time loop.
2800
2801 28. More problems have arisen in unanchored patterns when CRLF is a valid line
2802 break. For example, the unstudied pattern [\r\n]A does not match the string
2803 "\r\nA" because change 7.0/46 below moves the current point on by two
2804 characters after failing to match at the start. However, the pattern \nA
2805 *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
2806 the same is true. There doesn't seem any very clean way out of this, but
2807 what I have chosen to do makes the common cases work: PCRE now takes note
2808 of whether there can be an explicit match for \r or \n anywhere in the
2809 pattern, and if so, 7.0/46 no longer applies. As part of this change,
2810 there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
2811 pattern has explicit CR or LF references.
2812
2813 29. Added (*CR) etc for changing newline setting at start of pattern.
2814
2815
2816 Version 7.2 19-Jun-07
2817 ---------------------
2818
2819 1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
2820 which is apparently normally available under Windows.
2821
2822 2. Re-jig the pcregrep tests with different newline settings in an attempt
2823 to make them independent of the local environment's newline setting.
2824
2825 3. Add code to configure.ac to remove -g from the CFLAGS default settings.
2826
2827 4. Some of the "internals" tests were previously cut out when the link size
2828 was not 2, because the output contained actual offsets. The recent new
2829 "Z" feature of pcretest means that these can be cut out, making the tests
2830 usable with all link sizes.
2831
2832 5. Implemented Stan Switzer's goto replacement for longjmp() when not using
2833 stack recursion. This gives a massive performance boost under BSD, but just
2834 a small improvement under Linux. However, it saves one field in the frame
2835 in all cases.
2836
2837 6. Added more features from the forthcoming Perl 5.10:
2838
2839 (a) (?-n) (where n is a string of digits) is a relative subroutine or
2840 recursion call. It refers to the nth most recently opened parentheses.
2841
2842 (b) (?+n) is also a relative subroutine call; it refers to the nth next
2843 to be opened parentheses.
2844
2845 (c) Conditions that refer to capturing parentheses can be specified
2846 relatively, for example, (?(-2)... or (?(+3)...
2847
2848 (d) \K resets the start of the current match so that everything before
2849 is not part of it.
2850
2851 (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
2852
2853 (f) \g{name} is another synonym - part of Perl 5.10's unification of
2854 reference syntax.
2855
2856 (g) (?| introduces a group in which the numbering of parentheses in each
2857 alternative starts with the same number.
2858
2859 (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
2860
2861 7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
2862 PCRE_INFO_JCHANGED.
2863
2864 8. A pattern such as (.*(.)?)* caused pcre_exec() to fail by either not
2865 terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
2866 for detecting groups that can match an empty string.
2867
2868 9. A pattern with a very large number of alternatives (more than several
2869 hundred) was running out of internal workspace during the pre-compile
2870 phase, where pcre_compile() figures out how much memory will be needed. A
2871 bit of new cunning has reduced the workspace needed for groups with
2872 alternatives. The 1000-alternative test pattern now uses 12 bytes of
2873 workspace instead of running out of the 4096 that are available.
2874
2875 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
2876
2877 11. Applied patch from Google to remove an optimization that didn't quite work.
2878 The report of the bug said:
2879
2880 pcrecpp::RE("a*").FullMatch("aaa") matches, while
2881 pcrecpp::RE("a*?").FullMatch("aaa") does not, and
2882 pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
2883
2884 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
2885 it matched the wrong number of bytes.
2886
2887
2888 Version 7.1 24-Apr-07
2889 ---------------------
2890
2891 1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
2892 that is more "standard", making use of automake and other Autotools. There
2893 is some re-arrangement of the files and adjustment of comments consequent
2894 on this.
2895
2896 2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
2897 for recursive directory scanning broke on some systems because the files
2898 are not scanned in any specific order and on different systems the order
2899 was different. A call to "sort" has been inserted into RunGrepTest for the
2900 approprate test as a short-term fix. In the longer term there may be an
2901 alternative.
2902
2903 3. I had an email from Eric Raymond about problems translating some of PCRE's
2904 man pages to HTML (despite the fact that I distribute HTML pages, some
2905 people do their own conversions for various reasons). The problems
2906 concerned the use of low-level troff macros .br and .in. I have therefore
2907 removed all such uses from the man pages (some were redundant, some could
2908 be replaced by .nf/.fi pairs). The 132html script that I use to generate
2909 HTML has been updated to handle .nf/.fi and to complain if it encounters
2910 .br or .in.
2911
2912 4. Updated comments in configure.ac that get placed in config.h.in and also
2913 arranged for config.h to be included in the distribution, with the name
2914 config.h.generic, for the benefit of those who have to compile without
2915 Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
2916
2917 5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
2918 Weber: (1) pcre_internal.h was missing some function renames; (2) updated
2919 makevp.bat for the current PCRE, using the additional files
2920 makevp_c.txt, makevp_l.txt, and pcregexp.pas.
2921
2922 6. A Windows user reported a minor discrepancy with test 2, which turned out
2923 to be caused by a trailing space on an input line that had got lost in his
2924 copy. The trailing space was an accident, so I've just removed it.
2925
2926 7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
2927 that is needed.
2928
2929 8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
2930 as "const" (a) because they are and (b) because it helps the PHP
2931 maintainers who have recently made a script to detect big data structures
2932 in the php code that should be moved to the .rodata section. I remembered
2933 to update Builducptable as well, so it won't revert if ucptable.h is ever
2934 re-created.
2935
2936 9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
2937 pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
2938 order to be able to cut out the UTF-8 tables in the latter when UTF-8
2939 support is not required. This saves 1.5-2K of code, which is important in
2940 some applications.
2941
2942 Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
2943 so as not to refer to the tables, even though these functions will never be
2944 called when UTF-8 support is disabled. Otherwise there are problems with a
2945 shared library.
2946
2947 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
2948
2949 (a) It was defining its arguments as char * instead of void *.
2950
2951 (b) It was assuming that all moves were upwards in memory; this was true
2952 a long time ago when I wrote it, but is no longer the case.
2953
2954 The emulated memove() is provided for those environments that have neither
2955 memmove() nor bcopy(). I didn't think anyone used it these days, but that
2956 is clearly not the case, as these two bugs were recently reported.
2957
2958 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
2959 and Detrail to create the HTML documentation, the .txt form of the man
2960 pages, and it removes trailing spaces from listed files. It also creates
2961 pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
2962 case, it wraps all the #defines with #ifndefs. This script should be run
2963 before "make dist".
2964
2965 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
2966 with Unicode property support.
2967
2968 (a) For a maximizing quantifier, if the two different cases of the
2969 character were of different lengths in their UTF-8 codings (there are
2970 some cases like this - I found 11), and the matching function had to
2971 back up over a mixture of the two cases, it incorrectly assumed they
2972 were both the same length.
2973
2974 (b) When PCRE was configured to use the heap rather than the stack for
2975 recursion during matching, it was not correctly preserving the data for
2976 the other case of a UTF-8 character when checking ahead for a match
2977 while processing a minimizing repeat. If the check also involved
2978 matching a wide character, but failed, corruption could cause an
2979 erroneous result when trying to check for a repeat of the original
2980 character.
2981
2982 13. Some tidying changes to the testing mechanism:
2983
2984 (a) The RunTest script now detects the internal link size and whether there
2985 is UTF-8 and UCP support by running ./pcretest -C instead of relying on
2986 values substituted by "configure". (The RunGrepTest script already did
2987 this for UTF-8.) The configure.ac script no longer substitutes the
2988 relevant variables.
2989
2990 (b) The debugging options /B and /D in pcretest show the compiled bytecode
2991 with length and offset values. This means that the output is different
2992 for different internal link sizes. Test 2 is skipped for link sizes
2993 other than 2 because of this, bypassing the problem. Unfortunately,
2994 there was also a test in test 3 (the locale tests) that used /B and
2995 failed for link sizes other than 2. Rather than cut the whole test out,
2996 I have added a new /Z option to pcretest that replaces the length and
2997 offset values with spaces. This is now used to make test 3 independent
2998 of link size. (Test 2 will be tidied up later.)
2999
3000 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
3001 segmentation fault instead of returning the appropriate error message.
3002
3003 15. In multiline mode when the newline sequence was set to "any", the pattern
3004 ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
3005 This doesn't seem right; it now treats the CRLF combination as the line
3006 ending, and so does not match in that case. It's only a pattern such as ^$
3007 that would hit this one: something like ^ABC$ would have failed after \r
3008 and then tried again after \r\n.
3009
3010 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
3011 in an attempt to make files that differ only in their line terminators
3012 compare equal. This works on Linux.
3013
3014 17. Under certain error circumstances pcregrep might try to free random memory
3015 as it exited. This is now fixed, thanks to valgrind.
3016
3017 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
3018 "abc\r\n\r\n", it found an unwanted second match after the second \r. This
3019 was because its rules for how to advance for /g after matching an empty
3020 string at the end of a line did not allow for this case. They now check for
3021 it specially.
3022
3023 20. pcretest is supposed to handle patterns and data of any length, by
3024 extending its buffers when necessary. It was getting this wrong when the
3025 buffer for a data line had to be extended.
3026
3027 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
3028 CRLF as a newline sequence.
3029
3030 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
3031 out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
3032 I have nevertheless tidied it up.
3033
3034 23. Added some casts to kill warnings from HP-UX ia64 compiler.
3035
3036 24. Added a man page for pcre-config.
3037
3038
3039 Version 7.0 19-Dec-06
3040 ---------------------
3041
3042 1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
3043 moving to gcc 4.1.1.
3044
3045 2. The -S option for pcretest uses setrlimit(); I had omitted to #include
3046 sys/time.h, which is documented as needed for this function. It doesn't
3047 seem to matter on Linux, but it showed up on some releases of OS X.
3048
3049 3. It seems that there are systems where bytes whose values are greater than
3050 127 match isprint() in the "C" locale. The "C" locale should be the
3051 default when a C program starts up. In most systems, only ASCII printing
3052 characters match isprint(). This difference caused the output from pcretest
3053 to vary, making some of the tests fail. I have changed pcretest so that:
3054
3055 (a) When it is outputting text in the compiled version of a pattern, bytes
3056 other than 32-126 are always shown as hex escapes.
3057
3058 (b) When it is outputting text that is a matched part of a subject string,
3059 it does the same, unless a different locale has been set for the match
3060 (using the /L modifier). In this case, it uses isprint() to decide.
3061
3062 4. Fixed a major bug that caused incorrect computation of the amount of memory
3063 required for a compiled pattern when options that changed within the
3064 pattern affected the logic of the preliminary scan that determines the
3065 length. The relevant options are -x, and -i in UTF-8 mode. The result was
3066 that the computed length was too small. The symptoms of this bug were
3067 either the PCRE error "internal error: code overflow" from pcre_compile(),
3068 or a glibc crash with a message such as "pcretest: free(): invalid next
3069 size (fast)". Examples of patterns that provoked this bug (shown in
3070 pcretest format) are:
3071
3072 /(?-x: )/x
3073 /(?x)(?-x: \s*#\s*)/
3074 /((?i)[\x{c0}])/8
3075 /(?i:[\x{c0}])/8
3076
3077 HOWEVER: Change 17 below makes this fix obsolete as the memory computation
3078 is now done differently.
3079
3080 5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
3081 wrapper classes; (b) implement a new function in the C++ scanner that is
3082 more efficient than the old way of doing things because it avoids levels of
3083 recursion in the regex matching; (c) add a paragraph to the documentation
3084 for the FullMatch() function.
3085
3086 6. The escape sequence \n was being treated as whatever was defined as
3087 "newline". Not only was this contrary to the documentation, which states
3088 that \n is character 10 (hex 0A), but it also went horribly wrong when
3089 "newline" was defined as CRLF. This has been fixed.
3090
3091 7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
3092 was being set to -1 for the "end of line" case (supposedly a value that no
3093 character can have). Though this value is never used (the check for end of
3094 line is "zero bytes in current character"), it caused compiler complaints.
3095 I've changed it to 0xffffffff.
3096
3097 8. In pcre_version.c, the version string was being built by a sequence of
3098 C macros that, in the event of PCRE_PRERELEASE being defined as an empty
3099 string (as it is for production releases) called a macro with an empty
3100 argument. The C standard says the result of this is undefined. The gcc
3101 compiler treats it as an empty string (which was what was wanted) but it is
3102 reported that Visual C gives an error. The source has been hacked around to
3103 avoid this problem.
3104
3105 9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
3106 builds of pcretest, and changed the call to _setmode() to use _O_BINARY
3107 instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
3108 of them did).
3109
3110 10. Originally, pcretest opened its input and output without "b"; then I was
3111 told that "b" was needed in some environments, so it was added for release
3112 5.0 to both the input and output. (It makes no difference on Unix-like
3113 systems.) Later I was told that it is wrong for the input on Windows. I've
3114 now abstracted the modes into two macros, to make it easier to fiddle with
3115 them, and removed "b" from the input mode under Windows.
3116
3117 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
3118
3119 12. Added -help and --help to pcretest as an official way of being reminded
3120 of the options.
3121
3122 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
3123 and pcrecpp.cc because they annoy compilers at high warning levels.
3124
3125 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
3126
3127 15. Fixed an occurrence of == in configure.ac that should have been = (shell
3128 scripts are not C programs :-) and which was not noticed because it works
3129 on Linux.
3130
3131 16. pcretest is supposed to handle any length of pattern and data line (as one
3132 line or as a continued sequence of lines) by extending its input buffer if
3133 necessary. This feature was broken for very long pattern lines, leading to
3134 a string of junk being passed to pcre_compile() if the pattern was longer
3135 than about 50K.
3136
3137 17. I have done a major re-factoring of the way pcre_compile() computes the
3138 amount of memory needed for a compiled pattern. Previously, there was code
3139 that made a preliminary scan of the pattern in order to do this. That was
3140 OK when PCRE was new, but as the facilities have expanded, it has become
3141 harder and harder to keep it in step with the real compile phase, and there
3142 have been a number of bugs (see for example, 4 above). I have now found a
3143 cunning way of running the real compile function in a "fake" mode that
3144 enables it to compute how much memory it would need, while actually only
3145 ever using a few hundred bytes of working memory and without too many
3146 tests of the mode. This should make future maintenance and development
3147 easier. A side effect of this work is that the limit of 200 on the nesting
3148 depth of parentheses has been removed (though this was never a serious
3149 limitation, I suspect). However, there is a downside: pcre_compile() now
3150 runs more slowly than before (30% or more, depending on the pattern). I
3151 hope this isn't a big issue. There is no effect on runtime performance.
3152
3153 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
3154 newline (only possible for the last line of a file) and it was a
3155 pattern that set a locale (followed by /Lsomething), pcretest crashed.
3156
3157 19. Added additional timing features to pcretest. (1) The -tm option now times
3158 matching only, not compiling. (2) Both -t and -tm can be followed, as a
3159 separate command line item, by a number that specifies the number of
3160 repeats to use when timing. The default is 50000; this gives better
3161 precision, but takes uncomfortably long for very large patterns.
3162
3163 20. Extended pcre_study() to be more clever in cases where a branch of a
3164 subpattern has no definite first character. For example, (a*|b*)[cd] would
3165 previously give no result from pcre_study(). Now it recognizes that the
3166 first character must be a, b, c, or d.
3167
3168 21. There was an incorrect error "recursive call could loop indefinitely" if
3169 a subpattern (or the entire pattern) that was being tested for matching an
3170 empty string contained only one non-empty item after a nested subpattern.
3171 For example, the pattern (?>\x{100}*)\d(?R) provoked this error
3172 incorrectly, because the \d was being skipped in the check.
3173
3174 22. The pcretest program now has a new pattern option /B and a command line
3175 option -b, which is equivalent to adding /B to every pattern. This causes
3176 it to show the compiled bytecode, without the additional information that
3177 -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
3178 is the same as /B/I).
3179
3180 23. A new optimization is now able automatically to treat some sequences such
3181 as a*b as a*+b. More specifically, if something simple (such as a character
3182 or a simple class like \d) has an unlimited quantifier, and is followed by
3183 something that cannot possibly match the quantified thing, the quantifier
3184 is automatically "possessified".
3185
3186 24. A recursive reference to a subpattern whose number was greater than 39
3187 went wrong under certain circumstances in UTF-8 mode. This bug could also
3188 have affected the operation of pcre_study().
3189
3190 25. Realized that a little bit of performance could be had by replacing
3191 (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
3192
3193 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
3194
3195 27. Possessive quantifiers such as a++ were previously implemented by turning
3196 them into atomic groups such as ($>a+). Now they have their own opcodes,
3197 which improves performance. This includes the automatically created ones
3198 from 23 above.
3199
3200 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
3201 lookahead was broken if it was not anchored. PCRE was mistakenly expecting
3202 the first matched character to be a colon. This applied both to named and
3203 numbered groups.
3204
3205 29. The ucpinternal.h header file was missing its idempotency #ifdef.
3206
3207 30. I was sent a "project" file called libpcre.a.dev which I understand makes
3208 building PCRE on Windows easier, so I have included it in the distribution.
3209
3210 31. There is now a check in pcretest against a ridiculously large number being
3211 returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
3212 loop, the loop is abandoned.
3213
3214 32. Forward references to subpatterns in conditions such as (?(2)...) where
3215 subpattern 2 is defined later cause pcre_compile() to search forwards in
3216 the pattern for the relevant set of parentheses. This search went wrong
3217 when there were unescaped parentheses in a character class, parentheses
3218 escaped with \Q...\E, or parentheses in a #-comment in /x mode.
3219
3220 33. "Subroutine" calls and backreferences were previously restricted to
3221 referencing subpatterns earlier in the regex. This restriction has now
3222 been removed.
3223
3224 34. Added a number of extra features that are going to be in Perl 5.10. On the
3225 whole, these are just syntactic alternatives for features that PCRE had
3226 previously implemented using the Python syntax or my own invention. The
3227 other formats are all retained for compatibility.
3228
3229 (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
3230 as (?P<name>...). The new forms, as well as being in Perl 5.10, are
3231 also .NET compatible.
3232
3233 (b) A recursion or subroutine call to a named group can now be defined as
3234 (?&name) as well as (?P>name).
3235
3236 (c) A backreference to a named group can now be defined as \k<name> or
3237 \k'name' as well as (?P=name). The new forms, as well as being in Perl
3238 5.10, are also .NET compatible.
3239
3240 (d) A conditional reference to a named group can now use the syntax
3241 (?(<name>) or (?('name') as well as (?(name).
3242
3243 (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
3244 groups (named and numbered) that are never evaluated inline, but can be
3245 called as "subroutines" from elsewhere. In effect, the DEFINE condition
3246 is always false. There may be only one alternative in such a group.
3247
3248 (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
3249 as the simple (?(R). The condition is true only if the most recent
3250 recursion is that of the given number or name. It does not search out
3251 through the entire recursion stack.
3252
3253 (g) The escape \gN or \g{N} has been added, where N is a positive or
3254 negative number, specifying an absolute or relative reference.
3255
3256 35. Tidied to get rid of some further signed/unsigned compiler warnings and
3257 some "unreachable code" warnings.
3258
3259 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
3260 things, this adds five new scripts.
3261
3262 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
3263 There were also incompatibilities regarding the handling of \Q..\E inside
3264 character classes, for example with patterns like [\Qa\E-\Qz\E] where the
3265 hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
3266
3267 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
3268 matches an empty string, and forcibly breaks the loop. There were bugs in
3269 this code in non-simple cases. For a pattern such as ^(a()*)* matched
3270 against aaaa the result was just "a" rather than "aaaa", for example. Two
3271 separate and independent bugs (that affected different cases) have been
3272 fixed.
3273
3274 39. Refactored the code to abolish the use of different opcodes for small
3275 capturing bracket numbers. This is a tidy that I avoided doing when I
3276 removed the limit on the number of capturing brackets for 3.5 back in 2001.
3277 The new approach is not only tidier, it makes it possible to reduce the
3278 memory needed to fix the previous bug (38).
3279
3280 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
3281 sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
3282 processing dot, circumflex, or dollar metacharacters, or #-comments in /x
3283 mode.
3284
3285 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
3286 report.
3287
3288 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
3289 copy construction and assignment in the C++ wrapper.
3290
3291 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
3292 couple of bugs that could have given wrong results in the "--newline=crlf"
3293 case.
3294
3295 44. Added a number of casts and did some reorganization of signed/unsigned int
3296 variables following suggestions from Dair Grant. Also renamed the variable
3297 "this" as "item" because it is a C++ keyword.
3298
3299 45. Arranged for dftables to add
3300
3301 #include "pcre_internal.h"
3302
3303 to pcre_chartables.c because without it, gcc 4.x may remove the array
3304 definition from the final binary if PCRE is built into a static library and
3305 dead code stripping is activated.
3306
3307 46. For an unanchored pattern, if a match attempt fails at the start of a
3308 newline sequence, and the newline setting is CRLF or ANY, and the next two
3309 characters are CRLF, advance by two characters instead of one.
3310
3311
3312 Version 6.7 04-Jul-06
3313 ---------------------
3314
3315 1. In order to handle tests when input lines are enormously long, pcretest has
3316 been re-factored so that it automatically extends its buffers when
3317 necessary. The code is crude, but this _is_ just a test program. The
3318 default size has been increased from 32K to 50K.
3319
3320 2. The code in pcre_study() was using the value of the re argument before
3321 testing it for NULL. (Of course, in any sensible call of the function, it
3322 won't be NULL.)
3323
3324 3. The memmove() emulation function in pcre_internal.h, which is used on
3325 systems that lack both memmove() and bcopy() - that is, hardly ever -
3326 was missing a "static" storage class specifier.
3327
3328 4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
3329 containing an extended class (one that cannot be represented by a bitmap
3330 because it contains high-valued characters or Unicode property items, e.g.
3331 [\pZ]). Almost always one would set UTF-8 mode when processing such a
3332 pattern, but PCRE should not loop if you do not (it no longer does).
3333 [Detail: two cases were found: (a) a repeated subpattern containing an
3334 extended class; (b) a recursive reference to a subpattern that followed a
3335 previous extended class. It wasn't skipping over the extended class
3336 correctly when UTF-8 mode was not set.]
3337
3338 5. A negated single-character class was not being recognized as fixed-length
3339 in lookbehind assertions such as (?<=[^f]), leading to an incorrect
3340 compile error "lookbehind assertion is not fixed length".
3341
3342 6. The RunPerlTest auxiliary script was showing an unexpected difference
3343 between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
3344 write a Perl script that can interpret lines of an input file either as
3345 byte characters or as UTF-8, which is what "perltest" was being required to
3346 do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
3347 can't do is switch easily at run time between having the "use utf8;" pragma
3348 or not. In the end, I fudged it by using the RunPerlTest script to insert
3349 "use utf8;" explicitly for the UTF-8 tests.
3350
3351 7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
3352 the end of the subject string, contrary to the documentation and to what
3353 Perl does. This was true of both matching functions. Now it matches only at
3354 the start of the subject and immediately after *internal* newlines.
3355
3356 8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
3357 a pointer to an int instead of a pointer to an unsigned long int. This
3358 caused problems on 64-bit systems.
3359
3360 9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
3361 instance of the 'standard' template library not being so standard".
3362
3363 10. There was no check on the number of named subpatterns nor the maximum
3364 length of a subpattern name. The product of these values is used to compute
3365 the size of the memory block for a compiled pattern. By supplying a very
3366 long subpattern name and a large number of named subpatterns, the size
3367 computation could be caused to overflow. This is now prevented by limiting
3368 the length of names to 32 characters, and the number of named subpatterns
3369 to 10,000.
3370
3371 11. Subpatterns that are repeated with specific counts have to be replicated in
3372 the compiled pattern. The size of memory for this was computed from the
3373 length of the subpattern and the repeat count. The latter is limited to
3374 65535, but there was no limit on the former, meaning that integer overflow
3375 could in principle occur. The compiled length of a repeated subpattern is
3376 now limited to 30,000 bytes in order to prevent this.
3377
3378 12. Added the optional facility to have named substrings with the same name.
3379
3380 13. Added the ability to use a named substring as a condition, using the
3381 Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
3382 are numbers (not recommended). Forward references are permitted.
3383
3384 14. Added forward references in named backreferences (if you see what I mean).
3385
3386 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
3387 pattern could run off the end of the subject. For example, the pattern
3388 "(?s)(.{1,5})"8 did this with the subject "ab".
3389
3390 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
3391 PCRE_CASELESS was set when matching characters that were quantified with ?
3392 or *.
3393
3394 17. A character class other than a single negated character that had a minimum
3395 but no maximum quantifier - for example [ab]{6,} - was not handled
3396 correctly by pce_dfa_exec(). It would match only one character.
3397
3398 18. A valid (though odd) pattern that looked like a POSIX character
3399 class but used an invalid character after [ (for example [[,abc,]]) caused
3400 pcre_compile() to give the error "Failed: internal error: code overflow" or
3401 in some cases to crash with a glibc free() error. This could even happen if
3402 the pattern terminated after [[ but there just happened to be a sequence of
3403 letters, a binary zero, and a closing ] in the memory that followed.
3404
3405 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
3406 over the years. Originally (before any Unicode support), just the bottom 8
3407 bits were taken. Thus, for example, \500 really meant \100. Nowadays the
3408 output from "man perlunicode" includes this:
3409
3410 The regular expression compiler produces polymorphic opcodes. That
3411 is, the pattern adapts to the data and automatically switches to
3412 the Unicode character scheme when presented with Unicode data--or
3413 instead uses a traditional byte scheme when presented with byte
3414 data.
3415
3416 Sadly, a wide octal escape does not cause a switch, and in a string with
3417 no other multibyte characters, these octal escapes are treated as before.
3418 Thus, in Perl, the pattern /\500/ actually matches \100 but the pattern
3419 /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
3420 Unicode string.
3421
3422 I have not perpetrated such confusion in PCRE. Up till now, it took just
3423 the bottom 8 bits, as in old Perl. I have now made octal escapes with
3424 values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
3425 translate to the appropriate multibyte character.
3426
3427 29. Applied some refactoring to reduce the number of warnings from Microsoft
3428 and Borland compilers. This has included removing the fudge introduced
3429 seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
3430 a warning about an unused variable.
3431
3432 21. PCRE has not included VT (character 0x0b) in the set of whitespace
3433 characters since release 4.0, because Perl (from release 5.004) does not.
3434 [Or at least, is documented not to: some releases seem to be in conflict
3435 with the documentation.] However, when a pattern was studied with
3436 pcre_study() and all its branches started with \s, PCRE still included VT
3437 as a possible starting character. Of course, this did no harm; it just
3438 caused an unnecessary match attempt.
3439
3440 22. Removed a now-redundant internal flag bit that recorded the fact that case
3441 dependency changed within the pattern. This was once needed for "required
3442 byte" processing, but is no longer used. This recovers a now-scarce options
3443 bit. Also moved the least significant internal flag bit to the most-
3444 significant bit of the word, which was not previously used (hangover from
3445 the days when it was an int rather than a uint) to free up another bit for
3446 the future.
3447
3448 23. Added support for CRLF line endings as well as CR and LF. As well as the
3449 default being selectable at build time, it can now be changed at runtime
3450 via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
3451 specify that it is scanning data with non-default line endings.
3452
3453 24. Changed the definition of CXXLINK to make it agree with the definition of
3454 LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
3455
3456 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
3457 recursions. This makes a big different to stack usage for some patterns.
3458
3459 26. If a subpattern containing a named recursion or subroutine reference such
3460 as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
3461 the space required for the compiled pattern went wrong and gave too small a
3462 value. Depending on the environment, this could lead to "Failed: internal
3463 error: code overflow at offset 49" or "glibc detected double free or
3464 corruption" errors.
3465
3466 27. Applied patches from Google (a) to support the new newline modes and (b) to
3467 advance over multibyte UTF-8 characters in GlobalReplace.
3468
3469 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
3470 difference for some implementation of PCRE in some Windows version.
3471
3472 29. Added some extra testing facilities to pcretest:
3473
3474 \q<number> in a data line sets the "match limit" value
3475 \Q<number> in a data line sets the "match recursion limt" value
3476 -S <number> sets the stack size, where <number> is in megabytes
3477
3478 The -S option isn't available for Windows.
3479
3480
3481 Version 6.6 06-Feb-06
3482 ---------------------
3483
3484 1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
3485 in pcreposix.h. I have copied the definition from pcre.h.
3486
3487 2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
3488 because pcre.h is no longer a built file.
3489
3490 3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
3491 not normally included in the compiled code.
3492
3493
3494 Version 6.5 01-Feb-06
3495 ---------------------
3496
3497 1. When using the partial match feature with pcre_dfa_exec(), it was not
3498 anchoring the second and subsequent partial matches at the new starting
3499 point. This could lead to incorrect results. For example, with the pattern
3500 /1234/, partially matching against "123" and then "a4" gave a match.
3501
3502 2. Changes to pcregrep:
3503
3504 (a) All non-match returns from pcre_exec() were being treated as failures
3505 to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
3506 error message is output. Some extra information is given for the
3507 PCRE_ERROR_MATCHLIMIT and PCRE_ERROR_RECURSIONLIMIT errors, which are
3508 probably the only errors that are likely to be caused by users (by
3509 specifying a regex that has nested indefinite repeats, for instance).
3510 If there are more than 20 of these errors, pcregrep is abandoned.
3511
3512 (b) A binary zero was treated as data while matching, but terminated the
3513 output line if it was written out. This has been fixed: binary zeroes
3514 are now no different to any other data bytes.
3515
3516 (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
3517 used to set a locale for matching. The --locale=xxxx long option has
3518 been added (no short equivalent) to specify a locale explicitly on the
3519 pcregrep command, overriding the environment variables.
3520
3521 (d) When -B was used with -n, some line numbers in the output were one less
3522 than they should have been.
3523
3524 (e) Added the -o (--only-matching) option.
3525
3526 (f) If -A or -C was used with -c (count only), some lines of context were
3527 accidentally printed for the final match.
3528
3529 (g) Added the -H (--with-filename) option.
3530
3531 (h) The combination of options -rh failed to suppress file names for files
3532 that were found from directory arguments.
3533
3534 (i) Added the -D (--devices) and -d (--directories) options.
3535
3536 (j) Added the -F (--fixed-strings) option.
3537
3538 (k) Allow "-" to be used as a file name for -f as well as for a data file.
3539
3540 (l) Added the --colo(u)r option.
3541
3542 (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
3543 is not present by default.
3544
3545 3. A nasty bug was discovered in the handling of recursive patterns, that is,
3546 items such as (?R) or (?1), when the recursion could match a number of
3547 alternatives. If it matched one of the alternatives, but subsequently,
3548 outside the recursion, there was a failure, the code tried to back up into
3549 the recursion. However, because of the way PCRE is implemented, this is not
3550 possible, and the result was an incorrect result from the match.
3551
3552 In order to prevent this happening, the specification of recursion has
3553 been changed so that all such subpatterns are automatically treated as
3554 atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
3555
3556 4. I had overlooked the fact that, in some locales, there are characters for
3557 which isalpha() is true but neither isupper() nor islower() are true. In
3558 the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
3559 and ordfeminine) are like this. This affected the treatment of \w and \W
3560 when they appeared in character classes, but not when they appeared outside
3561 a character class. The bit map for "word" characters is now created
3562 separately from the results of isalnum() instead of just taking it from the
3563 upper, lower, and digit maps. (Plus the underscore character, of course.)
3564
3565 5. The above bug also affected the handling of POSIX character classes such as
3566 [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
3567 permanent tables. Instead, the bit maps for such a class were previously
3568 created as the appropriate unions of the upper, lower, and digit bitmaps.
3569 Now they are created by subtraction from the [[:word:]] class, which has
3570 its own bitmap.
3571
3572 6. The [[:blank:]] character class matches horizontal, but not vertical space.
3573 It is created by subtracting the vertical space characters (\x09, \x0a,
3574 \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
3575 subtraction was done in the overall bitmap for a character class, meaning
3576 that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
3577 be recognized. This bug has been fixed.
3578
3579 7. Patches from the folks at Google:
3580
3581 (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
3582 real life, but is still worth protecting against".
3583
3584 (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
3585 regular expressions".
3586
3587 (c) pcre_scanner.cc: avoid use of std::count() because not all systems
3588 have it.
3589
3590 (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
3591 "configure" and the latter not, in order to fix a problem somebody had
3592 with compiling the Arg class on HP-UX.
3593
3594 (e) Improve the error-handling of the C++ wrapper a little bit.
3595
3596 (f) New tests for checking recursion limiting.
3597
3598 8. The pcre_memmove() function, which is used only if the environment does not
3599 have a standard memmove() function (and is therefore rarely compiled),
3600 contained two bugs: (a) use of int instead of size_t, and (b) it was not
3601 returning a result (though PCRE never actually uses the result).
3602
3603 9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
3604 large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
3605 returned instead of calling malloc() with an overflowing number that would
3606 most likely cause subsequent chaos.
3607
3608 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
3609
3610 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
3611 with this option is matched, the nmatch and pmatch options of regexec() are
3612 ignored.
3613
3614 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
3615 provided in case anyone wants to the the POSIX interface with UTF-8
3616 strings.
3617
3618 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
3619 C++ linking (needed for some HP-UX environments).
3620
3621 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
3622 (unused parameter) and in the pcre_printint() function (omitted "default"
3623 switch label when the default is to do nothing).
3624
3625 15. Added some code to make it possible, when PCRE is compiled as a C++
3626 library, to replace subject pointers for pcre_exec() with a smart pointer
3627 class, thus making it possible to process discontinuous strings.
3628
3629 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
3630 much the same function. They were added by different people who were trying
3631 to make PCRE easy to compile on non-Unix systems. It has been suggested
3632 that PCRE_EXPORT be abolished now that there is more automatic apparatus
3633 for compiling on Windows systems. I have therefore replaced it with
3634 PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
3635 defaults to "extern" for C or "extern C" for C++, which works fine on
3636 Unix-like systems. It is now possible to override the value of PCRE_DATA_
3637 SCOPE with something explicit in config.h. In addition:
3638
3639 (a) pcreposix.h still had just "extern" instead of either of these macros;
3640 I have replaced it with PCRE_DATA_SCOPE.
3641
3642 (b) Functions such as _pcre_xclass(), which are internal to the library,
3643 but external in the C sense, all had PCRE_EXPORT in their definitions.
3644 This is apparently wrong for the Windows case, so I have removed it.
3645 (It makes no difference on Unix-like systems.)
3646
3647 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
3648 of recursive calls to match(). This is different to MATCH_LIMIT because
3649 that limits the total number of calls to match(), not all of which increase
3650 the depth of recursion. Limiting the recursion depth limits the amount of
3651 stack (or heap if NO_RECURSE is set) that is used. The default can be set
3652 when PCRE is compiled, and changed at run time. A patch from Google adds
3653 this functionality to the C++ interface.
3654
3655 18. Changes to the handling of Unicode character properties:
3656
3657 (a) Updated the table to Unicode 4.1.0.
3658
3659 (b) Recognize characters that are not in the table as "Cn" (undefined).
3660
3661 (c) I revised the way the table is implemented to a much improved format
3662 which includes recognition of ranges. It now supports the ranges that
3663 are defined in UnicodeData.txt, and it also amalgamates other
3664 characters into ranges. This has reduced the number of entries in the
3665 table from around 16,000 to around 3,000, thus reducing its size
3666 considerably. I realized I did not need to use a tree structure after
3667 all - a binary chop search is just as efficient. Having reduced the
3668 number of entries, I extended their size from 6 bytes to 8 bytes to
3669 allow for more data.
3670
3671 (d) Added support for Unicode script names via properties such as \p{Han}.
3672
3673 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
3674 matching that character.
3675
3676 20. When matching a repeated Unicode property with a minimum greater than zero,
3677 (for example \pL{2,}), PCRE could look past the end of the subject if it
3678 reached it while seeking the minimum number of characters. This could
3679 happen only if some of the characters were more than one byte long, because
3680 there is a check for at least the minimum number of bytes.
3681
3682 21. Refactored the implementation of \p and \P so as to be more general, to
3683 allow for more different types of property in future. This has changed the
3684 compiled form incompatibly. Anybody with saved compiled patterns that use
3685 \p or \P will have to recompile them.
3686
3687 22. Added "Any" and "L&" to the supported property types.
3688
3689 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
3690 but give a compile time error if the value is greater than 0xff.
3691
3692 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
3693 accidentally not being installed or uninstalled.
3694
3695 25. The pcre.h file was built from pcre.h.in, but the only changes that were
3696 made were to insert the current release number. This seemed silly, because
3697 it made things harder for people building PCRE on systems that don't run
3698 "configure". I have turned pcre.h into a distributed file, no longer built
3699 by "configure", with the version identification directly included. There is
3700 no longer a pcre.h.in file.
3701
3702 However, this change necessitated a change to the pcre-config script as
3703 well. It is built from pcre-config.in, and one of the substitutions was the
3704 release number. I have updated configure.ac so that ./configure now finds
3705 the release number by grepping pcre.h.
3706
3707 26. Added the ability to run the tests under valgrind.
3708
3709
3710 Version 6.4 05-Sep-05
3711 ---------------------
3712
3713 1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
3714 "--" to be printed when multiple files were scanned, even when none of the
3715 -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
3716 consider it to be a bug, and have restored the previous behaviour.
3717
3718 2. A couple of code tidies to get rid of compiler warnings.
3719
3720 3. The pcretest program used to cheat by referring to symbols in the library
3721 whose names begin with _pcre_. These are internal symbols that are not
3722 really supposed to be visible externally, and in some environments it is
3723 possible to suppress them. The cheating is now confined to including
3724 certain files from the library's source, which is a bit cleaner.
3725
3726 4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
3727 file's purpose clearer.
3728
3729 5. Reorganized pcre_ucp_findchar().
3730
3731
3732 Version 6.3 15-Aug-05
3733 ---------------------
3734
3735 1. The file libpcre.pc.in did not have general read permission in the tarball.
3736
3737 2. There were some problems when building without C++ support:
3738
3739 (a) If C++ support was not built, "make install" and "make test" still
3740 tried to test it.
3741
3742 (b) There were problems when the value of CXX was explicitly set. Some
3743 changes have been made to try to fix these, and ...
3744
3745 (c) --disable-cpp can now be used to explicitly disable C++ support.
3746
3747 (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
3748 backslash in a target when C++ was disabled. This confuses some
3749 versions of "make", apparently. Using an intermediate variable solves
3750 this. (Same for CPP_LOBJ.)
3751
3752 3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
3753 (non-Windows) now includes $(CFLAGS) because these flags are sometimes
3754 necessary on certain architectures.
3755
3756 4. Added a setting of -export-symbols-regex to the link command to remove
3757 those symbols that are exported in the C sense, but actually are local
3758 within the library, and not documented. Their names all begin with
3759 "_pcre_". This is not a perfect job, because (a) we have to except some
3760 symbols that pcretest ("illegally") uses, and (b) the facility isn't always
3761 available (and never for static libraries). I have made a note to try to
3762 find a way round (a) in the future.
3763
3764
3765 Version 6.2 01-Aug-05
3766 ---------------------
3767
3768 1. There was no test for integer overflow of quantifier values. A construction
3769 such as {1111111111111111} would give undefined results. What is worse, if
3770 a minimum quantifier for a parenthesized subpattern overflowed and became
3771 negative, the calculation of the memory size went wrong. This could have
3772 led to memory overwriting.
3773
3774 2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
3775
3776 3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
3777 operating environments where this matters.
3778
3779 4. Applied Giuseppe Maxia's patch to add additional features for controlling
3780 PCRE options from within the C++ wrapper.
3781
3782 5. Named capturing subpatterns were not being correctly counted when a pattern
3783 was compiled. This caused two problems: (a) If there were more than 100
3784 such subpatterns, the calculation of the memory needed for the whole
3785 compiled pattern went wrong, leading to an overflow error. (b) Numerical
3786 back references of the form \12, where the number was greater than 9, were
3787 not recognized as back references, even though there were sufficient
3788 previous subpatterns.
3789
3790 6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
3791 versions of gcc, e.g. 2.95.4.
3792
3793
3794 Version 6.1 21-Jun-05
3795 ---------------------
3796
3797 1. There was one reference to the variable "posix" in pcretest.c that was not
3798 surrounded by "#if !defined NOPOSIX".
3799
3800 2. Make it possible to compile pcretest without DFA support, UTF8 support, or
3801 the cross-check on the old pcre_info() function, for the benefit of the
3802 cut-down version of PCRE that is currently imported into Exim.
3803
3804 3. A (silly) pattern starting with (?i)(?-i) caused an internal space
3805 allocation error. I've done the easy fix, which wastes 2 bytes for sensible
3806 patterns that start (?i) but I don't think that matters. The use of (?i) is
3807 just an example; this all applies to the other options as well.
3808
3809 4. Since libtool seems to echo the compile commands it is issuing, the output
3810 from "make" can be reduced a bit by putting "@" in front of each libtool
3811 compile command.
3812
3813 5. Patch from the folks at Google for configure.in to be a bit more thorough
3814 in checking for a suitable C++ installation before trying to compile the
3815 C++ stuff. This should fix a reported problem when a compiler was present,
3816 but no suitable headers.
3817
3818 6. The man pages all had just "PCRE" as their title. I have changed them to
3819 be the relevant file name. I have also arranged that these names are
3820 retained in the file doc/pcre.txt, which is a concatenation in text format
3821 of all the man pages except the little individual ones for each function.
3822
3823 7. The NON-UNIX-USE file had not been updated for the different set of source
3824 files that come with release 6. I also added a few comments about the C++
3825 wrapper.
3826
3827
3828 Version 6.0 07-Jun-05
3829 ---------------------
3830
3831 1. Some minor internal re-organization to help with my DFA experiments.
3832
3833 2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
3834 didn't matter for the library itself when fully configured, but did matter
3835 when compiling without UCP support, or within Exim, where the ucp files are
3836 not imported.
3837
3838 3. Refactoring of the library code to split up the various functions into
3839 different source modules. The addition of the new DFA matching code (see
3840 below) to a single monolithic source would have made it really too
3841 unwieldy, quite apart from causing all the code to be include in a
3842 statically linked application, when only some functions are used. This is
3843 relevant even without the DFA addition now that patterns can be compiled in
3844 one application and matched in another.
3845
3846 The downside of splitting up is that there have to be some external
3847 functions and data tables that are used internally in different modules of
3848 the library but which are not part of the API. These have all had their
3849 names changed to start with "_pcre_" so that they are unlikely to clash
3850 with other external names.
3851
3852 4. Added an alternate matching function, pcre_dfa_exec(), which matches using
3853 a different (DFA) algorithm. Although it is slower than the original
3854 function, it does have some advantages for certain types of matching
3855 problem.
3856
3857 5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
3858 including restarting after a partial match.
3859
3860 6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
3861 defined when compiling for Windows was sent to me. I have put it into the
3862 code, though I have no means of testing or verifying it.
3863
3864 7. Added the pcre_refcount() auxiliary function.
3865
3866 8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
3867 match before or at the first newline in the subject string. In pcretest,
3868 the /f option on a pattern can be used to set this.
3869
3870 9. A repeated \w when used in UTF-8 mode with characters greater than 256
3871 would behave wrongly. This has been present in PCRE since release 4.0.
3872
3873 10. A number of changes to the pcregrep command:
3874
3875 (a) Refactored how -x works; insert ^(...)$ instead of setting
3876 PCRE_ANCHORED and checking the length, in preparation for adding
3877 something similar for -w.
3878
3879 (b) Added the -w (match as a word) option.
3880
3881 (c) Refactored the way lines are read and buffered so as to have more
3882 than one at a time available.
3883
3884 (d) Implemented a pcregrep test script.
3885
3886 (e) Added the -M (multiline match) option. This allows patterns to match
3887 over several lines of the subject. The buffering ensures that at least
3888 8K, or the rest of the document (whichever is the shorter) is available
3889 for matching (and similarly the previous 8K for lookbehind assertions).
3890
3891 (f) Changed the --help output so that it now says
3892
3893 -w, --word-regex(p)
3894
3895 instead of two lines, one with "regex" and the other with "regexp"
3896 because that confused at least one person since the short forms are the
3897 same. (This required a bit of code, as the output is generated
3898 automatically from a table. It wasn't just a text change.)
3899
3900 (g) -- can be used to terminate pcregrep options if the next thing isn't an
3901 option but starts with a hyphen. Could be a pattern or a path name
3902 starting with a hyphen, for instance.
3903
3904 (h) "-" can be given as a file name to represent stdin.
3905
3906 (i) When file names are being printed, "(standard input)" is used for
3907 the standard input, for compatibility with GNU grep. Previously
3908 "<stdin>" was used.
3909
3910 (j) The option --label=xxx can be used to supply a name to be used for
3911 stdin when file names are being printed. There is no short form.
3912
3913 (k) Re-factored the options decoding logic because we are going to add
3914 two more options that take data. Such options can now be given in four
3915 different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
3916
3917 (l) Added the -A, -B, and -C options for requesting that lines of context
3918 around matches be printed.
3919
3920 (m) Added the -L option to print the names of files that do not contain
3921 any matching lines, that is, the complement of -l.
3922
3923 (n) The return code is 2 if any file cannot be opened, but pcregrep does
3924 continue to scan other files.
3925
3926 (o) The -s option was incorrectly implemented. For compatibility with other
3927 greps, it now suppresses the error message for a non-existent or non-
3928 accessible file (but not the return code). There is a new option called
3929 -q that suppresses the output of matching lines, which was what -s was
3930 previously doing.
3931
3932 (p) Added --include and --exclude options to specify files for inclusion
3933 and exclusion when recursing.
3934
3935 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
3936 Hopefully, it now does.
3937
3938 12. Missing cast in pcre_study().
3939
3940 13. Added an "uninstall" target to the makefile.
3941
3942 14. Replaced "extern" in the function prototypes in Makefile.in with
3943 "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
3944 world, but is set differently for Windows.
3945
3946 15. Added a second compiling function called pcre_compile2(). The only
3947 difference is that it has an extra argument, which is a pointer to an
3948 integer error code. When there is a compile-time failure, this is set
3949 non-zero, in addition to the error test pointer being set to point to an
3950 error message. The new argument may be NULL if no error number is required
3951 (but then you may as well call pcre_compile(), which is now just a
3952 wrapper). This facility is provided because some applications need a
3953 numeric error indication, but it has also enabled me to tidy up the way
3954 compile-time errors are handled in the POSIX wrapper.
3955
3956 16. Added VPATH=.libs to the makefile; this should help when building with one
3957 prefix path and installing with another. (Or so I'm told by someone who
3958 knows more about this stuff than I do.)
3959
3960 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
3961 passes PCRE_DOTALL to the pcre_compile() function, making the "." character
3962 match everything, including newlines. This is not POSIX-compatible, but
3963 somebody wanted the feature. From pcretest it can be activated by using
3964 both the P and the s flags.
3965
3966 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
3967
3968 19. libpcre.pc was being incorrectly installed as executable.
3969
3970 20. A couple of places in pcretest check for end-of-line by looking for '\n';
3971 it now also looks for '\r' so that it will work unmodified on Windows.
3972
3973 21. Added Google's contributed C++ wrapper to the distribution.
3974
3975 22. Added some untidy missing memory free() calls in pcretest, to keep
3976 Electric Fence happy when testing.
3977
3978
3979
3980 Version 5.0 13-Sep-04
3981 ---------------------
3982
3983 1. Internal change: literal characters are no longer packed up into items
3984 containing multiple characters in a single byte-string. Each character
3985 is now matched using a separate opcode. However, there may be more than one
3986 byte in the character in UTF-8 mode.
3987
3988 2. The pcre_callout_block structure has two new fields: pattern_position and
3989 next_item_length. These contain the offset in the pattern to the next match
3990 item, and its length, respectively.
3991
3992 3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
3993 insertion of callouts before each pattern item. Added the /C option to
3994 pcretest to make use of this.
3995
3996 4. On the advice of a Windows user, the lines
3997
3998 #if defined(_WIN32) || defined(WIN32)
3999 _setmode( _fileno( stdout ), 0x8000 );
4000 #endif /* defined(_WIN32) || defined(WIN32) */
4001
4002 have been added to the source of pcretest. This apparently does useful
4003 magic in relation to line terminators.
4004
4005 5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
4006 for the benefit of those environments where the "b" makes a difference.
4007
4008 6. The icc compiler has the same options as gcc, but "configure" doesn't seem
4009 to know about it. I have put a hack into configure.in that adds in code
4010 to set GCC=yes if CC=icc. This seems to end up at a point in the
4011 generated configure script that is early enough to affect the setting of
4012 compiler options, which is what is needed, but I have no means of testing
4013 whether it really works. (The user who reported this had patched the
4014 generated configure script, which of course I cannot do.)
4015
4016 LATER: After change 22 below (new libtool files), the configure script
4017 seems to know about icc (and also ecc). Therefore, I have commented out
4018 this hack in configure.in.
4019
4020 7. Added support for pkg-config (2 patches were sent in).
4021
4022 8. Negated POSIX character classes that used a combination of internal tables
4023 were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
4024 [[:^ascii]]. Typically, they would match almost any characters. The other
4025 POSIX classes were not broken in this way.
4026
4027 9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
4028 to find the match, as PCRE was deluded into thinking that the match had to
4029 start at the start point or following a newline. The same bug applied to
4030 patterns with negative forward assertions or any backward assertions
4031 preceding ".*" at the start, unless the pattern required a fixed first
4032 character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
4033
4034 10. In UTF-8 mode, when moving forwards in the subject after a failed match
4035 starting at the last subject character, bytes beyond the end of the subject
4036 string were read.
4037
4038 11. Renamed the variable "class" as "classbits" to make life easier for C++
4039 users. (Previously there was a macro definition, but it apparently wasn't
4040 enough.)
4041
4042 12. Added the new field "tables" to the extra data so that tables can be passed
4043 in at exec time, or the internal tables can be re-selected. This allows
4044 a compiled regex to be saved and re-used at a later time by a different
4045 program that might have everything at different addresses.
4046
4047 13. Modified the pcre-config script so that, when run on Solaris, it shows a
4048 -R library as well as a -L library.
4049
4050 14. The debugging options of pcretest (-d on the command line or D on a
4051 pattern) showed incorrect output for anything following an extended class
4052 that contained multibyte characters and which was followed by a quantifier.
4053
4054 15. Added optional support for general category Unicode character properties
4055 via the \p, \P, and \X escapes. Unicode property support implies UTF-8
4056 support. It adds about 90K to the size of the library. The meanings of the
4057 inbuilt class escapes such as \d and \s have NOT been changed.
4058
4059 16. Updated pcredemo.c to include calls to free() to release the memory for the
4060 compiled pattern.
4061
4062 17. The generated file chartables.c was being created in the source directory
4063 instead of in the building directory. This caused the build to fail if the
4064 source directory was different from the building directory, and was
4065 read-only.
4066
4067 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
4068 file. No doubt somebody will tell me if they don't make sense... Also added
4069 Dan Mooney's comments about building on OpenVMS.
4070
4071 19. Added support for partial matching via the PCRE_PARTIAL option for
4072 pcre_exec() and the \P data escape in pcretest.
4073
4074 20. Extended pcretest with 3 new pattern features:
4075
4076 (i) A pattern option of the form ">rest-of-line" causes pcretest to
4077 write the compiled pattern to the file whose name is "rest-of-line".
4078 This is a straight binary dump of the data, with the saved pointer to
4079 the character tables forced to be NULL. The study data, if any, is
4080 written too. After writing, pcretest reads a new pattern.
4081
4082 (ii) If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
4083 compiled pattern from the given file. There must not be any
4084 occurrences of "<" in the file name (pretty unlikely); if there are,
4085 pcretest will instead treat the initial "<" as a pattern delimiter.
4086 After reading in the pattern, pcretest goes on to read data lines as
4087 usual.
4088
4089 (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
4090 and 16-bit fields in a compiled pattern, to simulate a pattern that
4091 was compiled on a host of opposite endianness.
4092
4093 21. The pcre-exec() function can now cope with patterns that were compiled on
4094 hosts of opposite endianness, with this restriction:
4095
4096 As for any compiled expression that is saved and used later, the tables
4097 pointer field cannot be preserved; the extra_data field in the arguments
4098 to pcre_exec() should be used to pass in a tables address if a value
4099 other than the default internal tables were used at compile time.
4100
4101 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
4102 now diagnosed as an error. Previously, most of the time, a negative number
4103 would have been treated as zero, but if in addition "ovector" was passed as
4104 NULL, a crash could occur.
4105
4106 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
4107 new versions from the libtool 1.5 distribution (the last one is a copy of
4108 a file called libtool.m4). This seems to have fixed the need to patch
4109 "configure" to support Darwin 1.3 (which I used to do). However, I still
4110 had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
4111 workstation).
4112
4113 24. Changed the PCRE licence to be the more standard "BSD" licence.
4114
4115
4116 Version 4.5 01-Dec-03
4117 ---------------------
4118
4119 1. There has been some re-arrangement of the code for the match() function so
4120 that it can be compiled in a version that does not call itself recursively.
4121 Instead, it keeps those local variables that need separate instances for
4122 each "recursion" in a frame on the heap, and gets/frees frames whenever it
4123 needs to "recurse". Keeping track of where control must go is done by means
4124 of setjmp/longjmp. The whole thing is implemented by a set of macros that
4125 hide most of the details from the main code, and operates only if
4126 NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
4127 "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
4128 operating.
4129
4130 To make it easier for callers to provide specially tailored get/free
4131 functions for this usage, two new functions, pcre_stack_malloc, and
4132 pcre_stack_free, are used. They are always called in strict stacking order,
4133 and the size of block requested is always the same.
4134
4135 The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
4136 PCRE has been compiled to use the stack or the heap for recursion. The
4137 -C option of pcretest uses this to show which version is compiled.
4138
4139 A new data escape \S, is added to pcretest; it causes the amounts of store
4140 obtained and freed by both kinds of malloc/free at match time to be added
4141 to the output.
4142
4143 2. Changed the locale test to use "fr_FR" instead of "fr" because that's
4144 what's available on my current Linux desktop machine.
4145
4146 3. When matching a UTF-8 string, the test for a valid string at the start has
4147 been extended. If start_offset is not zero, PCRE now checks that it points
4148 to a byte that is the start of a UTF-8 character. If not, it returns
4149 PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
4150 this is necessary because there may be backward assertions in the pattern.
4151 When matching the same subject several times, it may save resources to use
4152 PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
4153
4154 4. The code for checking the validity of UTF-8 strings has been tightened so
4155 that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
4156 containing "overlong sequences".
4157
4158 5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
4159 I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
4160 should have been "&", but it just so happened that all the cases this let
4161 through by mistake were picked up later in the function.
4162
4163 6. I had used a variable called "isblank" - this is a C99 function, causing
4164 some compilers to warn. To avoid this, I renamed it (as "blankclass").
4165
4166 7. Cosmetic: (a) only output another newline at the end of pcretest if it is
4167 prompting; (b) run "./pcretest /dev/null" at the start of the test script
4168 so the version is shown; (c) stop "make test" echoing "./RunTest".
4169
4170 8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
4171
4172 9. The prototype for memmove() for systems that don't have it was using
4173 size_t, but the inclusion of the header that defines size_t was later. I've
4174 moved the #includes for the C headers earlier to avoid this.
4175
4176 10. Added some adjustments to the code to make it easier to compiler on certain
4177 special systems:
4178
4179 (a) Some "const" qualifiers were missing.
4180 (b) Added the macro EXPORT before all exported functions; by default this
4181 is defined to be empty.
4182 (c) Changed the dftables auxiliary program (that builds chartables.c) so
4183 that it reads its output file name as an argument instead of writing
4184 to the standard output and assuming this can be redirected.
4185
4186 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
4187 class containing characters with values greater than 255, PCRE compilation
4188 went into a loop.
4189
4190 12. A recursive reference to a subpattern that was within another subpattern
4191 that had a minimum quantifier of zero caused PCRE to crash. For example,
4192 (x(y(?2))z)? provoked this bug with a subject that got as far as the
4193 recursion. If the recursively-called subpattern itself had a zero repeat,
4194 that was OK.
4195
4196 13. In pcretest, the buffer for reading a data line was set at 30K, but the
4197 buffer into which it was copied (for escape processing) was still set at
4198 1024, so long lines caused crashes.
4199
4200 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
4201 "internal error: code overflow...". This applied to any character class
4202 that was followed by a possessive quantifier.
4203
4204 15. Modified the Makefile to add libpcre.la as a prerequisite for
4205 libpcreposix.la because I was told this is needed for a parallel build to
4206 work.
4207
4208 16. If a pattern that contained .* following optional items at the start was
4209 studied, the wrong optimizing data was generated, leading to matching
4210 errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
4211 matching string must start with a or b or c. The correct conclusion for
4212 this pattern is that a match can start with any character.
4213
4214
4215 Version 4.4 13-Aug-03
4216 ---------------------
4217
4218 1. In UTF-8 mode, a character class containing characters with values between
4219 127 and 255 was not handled correctly if the compiled pattern was studied.
4220 In fixing this, I have also improved the studying algorithm for such
4221 classes (slightly).
4222
4223 2. Three internal functions had redundant arguments passed to them. Removal
4224 might give a very teeny performance improvement.
4225
4226 3. Documentation bug: the value of the capture_top field in a callout is *one
4227 more than* the number of the hightest numbered captured substring.
4228
4229 4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
4230 in incorrectly linking with a previously installed version. They now link
4231 explicitly with libpcre.la.
4232
4233 5. configure.in no longer needs to recognize Cygwin specially.
4234
4235 6. A problem in pcre.in for Windows platforms is fixed.
4236
4237 7. If a pattern was successfully studied, and the -d (or /D) flag was given to
4238 pcretest, it used to include the size of the study block as part of its
4239 output. Unfortunately, the structure contains a field that has a different
4240 size on different hardware architectures. This meant that the tests that
4241 showed this size failed. As the block is currently always of a fixed size,
4242 this information isn't actually particularly useful in pcretest output, so
4243 I have just removed it.
4244
4245 8. Three pre-processor statements accidentally did not start in column 1.
4246 Sadly, there are *still* compilers around that complain, even though
4247 standard C has not required this for well over a decade. Sigh.
4248
4249 9. In pcretest, the code for checking callouts passed small integers in the
4250 callout_data field, which is a void * field. However, some picky compilers
4251 complained about the casts involved for this on 64-bit systems. Now
4252 pcretest passes the address of the small integer instead, which should get
4253 rid of the warnings.
4254
4255 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
4256 both compile and run time, and gives an error if an invalid UTF-8 sequence
4257 is found. There is a option for disabling this check in cases where the
4258 string is known to be correct and/or the maximum performance is wanted.
4259
4260 11. In response to a bug report, I changed one line in Makefile.in from
4261
4262 -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
4263 to
4264 -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
4265
4266 to look similar to other lines, but I have no way of telling whether this
4267 is the right thing to do, as I do not use Windows. No doubt I'll get told
4268 if it's wrong...
4269
4270
4271 Version 4.3 21-May-03
4272 ---------------------
4273
4274 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
4275 Makefile.
4276
4277 2. Some refactoring to improve the quality of the code:
4278
4279 (i) The utf8_table... variables are now declared "const".
4280
4281 (ii) The code for \cx, which used the "case flipping" table to upper case
4282 lower case letters, now just substracts 32. This is ASCII-specific,
4283 but the whole concept of \cx is ASCII-specific, so it seems
4284 reasonable.
4285
4286 (iii) PCRE was using its character types table to recognize decimal and
4287 hexadecimal digits in the pattern. This is silly, because it handles
4288 only 0-9, a-f, and A-F, but the character types table is locale-
4289 specific, which means strange things might happen. A private
4290 table is now used for this - though it costs 256 bytes, a table is
4291 much faster than multiple explicit tests. Of course, the standard
4292 character types table is still used for matching digits in subject
4293 strings against \d.
4294
4295 (iv) Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
4296 ending in _t are). So I've renamed it as ESC_tee.
4297
4298 3. The first argument for regexec() in the POSIX wrapper should have been
4299 defined as "const".
4300
4301 4. Changed pcretest to use malloc() for its buffers so that they can be
4302 Electric Fenced for debugging.
4303
4304 5. There were several places in the code where, in UTF-8 mode, PCRE would try
4305 to read one or more bytes before the start of the subject string. Often this
4306 had no effect on PCRE's behaviour, but in some circumstances it could
4307 provoke a segmentation fault.
4308
4309 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
4310 to try to read one or more bytes before the start of the subject string.
4311
4312 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
4313 UTF-8 support could misbehave in various ways if the subject string
4314 contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
4315 area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
4316 back over UTF-8 characters.)
4317
4318
4319 Version 4.2 14-Apr-03
4320 ---------------------
4321
4322 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
4323
4324 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
4325 [ON_WINDOWS]: new variable, "#" on non-Windows platforms
4326 [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
4327 [WIN_PREFIX]: new variable, "cyg" for Cygwin
4328 * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
4329 and BUILD_EXEEXT
4330 Note: automatic setting of the BUILD variables is not yet working
4331 set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
4332 compile-time but not at link-time
4333 [LINK]: use for linking executables only
4334 make different versions for Windows and non-Windows
4335 [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
4336 libraries
4337 [LINK_FOR_BUILD]: new variable
4338 [OBJEXT]: use throughout
4339 [EXEEXT]: use throughout
4340 <winshared>: new target
4341 <wininstall>: new target
4342 <dftables.o>: use native compiler
4343 <dftables>: use native linker
4344 <install>: handle Windows platform correctly
4345 <clean>: ditto
4346 <check>: ditto
4347 copy DLL to top builddir before testing
4348
4349 As part of these changes, -no-undefined was removed again. This was reported
4350 to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
4351 in any case.
4352
4353 3. Some tidies to get rid of compiler warnings:
4354
4355 . In the match_data structure, match_limit was an unsigned long int, whereas
4356 match_call_count was an int. I've made them both unsigned long ints.
4357
4358 . In pcretest the fact that a const uschar * doesn't automatically cast to
4359 a void * provoked a warning.
4360
4361 . Turning on some more compiler warnings threw up some "shadow" variables
4362 and a few more missing casts.
4363
4364 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
4365 option, a class that contained a single character with a value between 128
4366 and 255 (e.g. /[\xFF]/) caused PCRE to crash.
4367
4368 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
4369 option, a class that contained several characters, but with at least one
4370 whose value was between 128 and 255 caused PCRE to crash.
4371
4372
4373 Version 4.1 12-Mar-03
4374 ---------------------
4375
4376 1. Compiling with gcc -pedantic found a couple of places where casts were
4377 needed, and a string in dftables.c that was longer than standard compilers are
4378 required to support.
4379
4380 2. Compiling with Sun's compiler found a few more places where the code could
4381 be tidied up in order to avoid warnings.
4382
4383 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
4384 first of these names is deprecated in the latest Autoconf in favour of the name
4385 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
4386 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
4387 analogy I have changed it to CFLAGS_FOR_BUILD.
4388
4389 4. Added -no-undefined to the linking command in the Makefile, because this is
4390 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
4391 linking step for the pcreposix library.
4392
4393 5. PCRE was failing to diagnose the case of two named groups with the same
4394 name.
4395
4396 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
4397 literal character that is needed in the subject for a match, and scans along to
4398 ensure that it is present before embarking on the full matching process. This
4399 saves time in cases of nested unlimited repeats that are never going to match.
4400 Problem: the scan can take a lot of time if the subject is very long (e.g.
4401 megabytes), thus penalizing straightforward matches. It is now done only if the
4402 amount of subject to be scanned is less than 1000 bytes.
4403
4404 7. A lesser problem with the same optimization is that it was recording the
4405 first character of an anchored pattern as "needed", thus provoking a search
4406 right along the subject, even when the first match of the pattern was going to
4407 fail. The "needed" character is now not set for anchored patterns, unless it
4408 follows something in the pattern that is of non-fixed length. Thus, it still
4409 fulfils its original purpose of finding quick non-matches in cases of nested
4410 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
4411
4412
4413 Version 4.0 17-Feb-03
4414 ---------------------
4415
4416 1. If a comment in an extended regex that started immediately after a meta-item
4417 extended to the end of string, PCRE compiled incorrect data. This could lead to
4418 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
4419
4420 2. Moved to autoconf 2.53 and libtool 1.4.2.
4421
4422 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
4423 the special perltest8 script is no longer needed - all the tests can be run
4424 from a single perltest script.
4425
4426 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
4427 by \s. It has now been removed in PCRE. This means it isn't recognized as
4428 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
4429 class [:space:] *does* include VT, thereby creating a mess.
4430
4431 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
4432 space and tab.
4433
4434 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
4435 its new features into the main test script, reducing the number of scripts.
4436
4437 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
4438 were backward compatible, and made the (?i) apply to the whole pattern, as if
4439 /i were given. Now it behaves more logically, and applies the option setting
4440 only to what follows. PCRE has been changed to follow suit. However, if it
4441 finds options settings right at the start of the pattern, it extracts them into
4442 the global options, as before. Thus, they show up in the info data.
4443
4444 8. Added support for the \Q...\E escape sequence. Characters in between are
4445 treated as literals. This is slightly different from Perl in that $ and @ are
4446 also handled as literals inside the quotes. In Perl, they will cause variable
4447 interpolation. Note the following examples:
4448
4449 Pattern PCRE matches Perl matches
4450
4451 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
4452 \Qabc\$xyz\E abc\$xyz abc\$xyz
4453 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
4454
4455 For compatibility with Perl, \Q...\E sequences are recognized inside character
4456 classes as well as outside them.
4457
4458 9. Re-organized 3 code statements in pcretest to avoid "overflow in
4459 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
4460 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
4461 signed/unsigned warnings.
4462
4463 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
4464 option for pcretest, so I've replaced it by a simple function that does just
4465 that job.
4466
4467 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
4468 "pcregrep -".
4469
4470 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
4471 Java package. This provides some syntactic sugar for simple cases of what my
4472 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
4473 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
4474 item, you can use this simplified notation. Note that only makes sense with
4475 greedy quantifiers. Consequently, the use of the possessive quantifier forces
4476 greediness, whatever the setting of the PCRE_UNGREEDY option.
4477
4478 13. A change of greediness default within a pattern was not taking effect at
4479 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
4480 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
4481 was abstracted outside.
4482
4483 14. PCRE now supports the \G assertion. It is true when the current matching
4484 position is at the start point of the match. This differs from \A when the
4485 starting offset is non-zero. Used with the /g option of pcretest (or similar
4486 code), it works in the same way as it does for Perl's /g option. If all
4487 alternatives of a regex begin with \G, the expression is anchored to the start
4488 match position, and the "anchored" flag is set in the compiled expression.
4489
4490 15. Some bugs concerning the handling of certain option changes within patterns
4491 have been fixed. These applied to options other than (?ims). For example,
4492 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
4493 been the other way round. Some of this was related to change 7 above.
4494
4495 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
4496 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
4497 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
4498 POSIX classes only within a class (e.g. /[[:alpha:]]/).
4499
4500 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
4501 mode. Unlike ".", it always matches newline, whatever the setting of
4502 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
4503 assertions. Perl allows it, but it doesn't (in general) work because it can't
4504 calculate the length of the lookbehind. At least, that's the case for Perl
4505 5.8.0 - I've been told they are going to document that it doesn't work in
4506 future.
4507
4508 18. Added an error diagnosis for escapes that PCRE does not support: these are
4509 \L, \l, \N, \P, \p, \U, \u, and \X.
4510
4511 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
4512 reading past the end of the pattern in cases such as /[abcd/.
4513
4514 20. PCRE was getting more memory than necessary for patterns with classes that
4515 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
4516
4517 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
4518 compiling PCRE for use with Virtual Pascal.
4519
4520 22. Small fix to the Makefile to make it work properly if the build is done
4521 outside the source tree.
4522
4523 23. Added a new extension: a condition to go with recursion. If a conditional
4524 subpattern starts with (?(R) the "true" branch is used if recursion has
4525 happened, whereas the "false" branch is used only at the top level.
4526
4527 24. When there was a very long string of literal characters (over 255 bytes
4528 without UTF support, over 250 bytes with UTF support), the computation of how
4529 much memory was required could be incorrect, leading to segfaults or other
4530 strange effects.
4531
4532 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
4533 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
4534 there was a subsequent back reference to those brackets. This meant that, for
4535 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
4536 possible to check for precisely this case. All we can do is abandon the
4537 optimization if .* occurs inside capturing brackets when there are any back
4538 references whatsoever. (See below for a better fix that came later.)
4539
4540 26. The handling of the optimization for finding the first character of a
4541 non-anchored pattern, and for finding a character that is required later in the
4542 match were failing in some cases. This didn't break the matching; it just
4543 failed to optimize when it could. The way this is done has been re-implemented.
4544
4545 27. Fixed typo in error message for invalid (?R item (it said "(?p").
4546
4547 28. Added a new feature that provides some of the functionality that Perl
4548 provides with (?{...}). The facility is termed a "callout". The way it is done
4549 in PCRE is for the caller to provide an optional function, by setting
4550 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
4551 global variable. By default it is unset, which disables all calling out. To get
4552 the function called, the regex must include (?C) at appropriate points. This
4553 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
4554 This provides a means of identifying different callout points. When PCRE
4555 reaches such a point in the regex, if pcre_callout has been set, the external
4556 function is called. It is provided with data in a structure called
4557 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
4558 matching continues; if it returns a non-zero value, the match at the current
4559 point fails. However, backtracking will occur if possible. [This was changed
4560 later and other features added - see item 49 below.]
4561
4562 29. pcretest is upgraded to test the callout functionality. It provides a
4563 callout function that displays information. By default, it shows the start of
4564 the match and the current position in the text. There are some new data escapes
4565 to vary what happens:
4566
4567 \C+ in addition, show current contents of captured substrings
4568 \C- do not supply a callout function
4569 \C!n return 1 when callout number n is reached
4570 \C!n!m return 1 when callout number n is reached for the mth time
4571
4572 30. If pcregrep was called with the -l option and just a single file name, it
4573 output "<stdin>" if a match was found, instead of the file name.
4574
4575 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
4576 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
4577 pcre_exec(). This saves a malloc/free per call. The default value of
4578 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
4579 when configuring.
4580
4581 32. The default maximum size of a compiled pattern is 64K. There have been a
4582 few cases of people hitting this limit. The code now uses macros to handle the
4583 storing of links as offsets within the compiled pattern. It defaults to 2-byte
4584 links, but this can be changed to 3 or 4 bytes by --with-link-size when
4585 configuring. Tests 2 and 5 work only with 2-byte links because they output
4586 debugging information about compiled patterns.
4587
4588 33. Internal code re-arrangements:
4589
4590 (a) Moved the debugging function for printing out a compiled regex into
4591 its own source file (printint.c) and used #include to pull it into
4592 pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
4593 separate copies.
4594
4595 (b) Defined the list of op-code names for debugging as a macro in
4596 internal.h so that it is next to the definition of the opcodes.
4597
4598 (c) Defined a table of op-code lengths for simpler skipping along compiled
4599 code. This is again a macro in internal.h so that it is next to the
4600 definition of the opcodes.
4601
4602 34. Added support for recursive calls to individual subpatterns, along the
4603 lines of Robin Houston's patch (but implemented somewhat differently).
4604
4605 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
4606 allow it to read and process whole directories in Win32. This code was
4607 contributed by Lionel Fourquaux; it has not been tested by me.
4608
4609 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
4610 used to name a group. Names consist of alphanumerics and underscores, and must
4611 be unique. Back references use the syntax (?P=name) and recursive calls use
4612 (?P>name) which is a PCRE extension to the Python extension. Groups still have
4613 numbers. The function pcre_fullinfo() can be used after compilation to extract
4614 a name/number map. There are three relevant calls:
4615
4616 PCRE_INFO_NAMEENTRYSIZE yields the size of each entry in the map
4617 PCRE_INFO_NAMECOUNT yields the number of entries
4618 PCRE_INFO_NAMETABLE yields a pointer to the map.
4619
4620 The map is a vector of fixed-size entries. The size of each entry depends on
4621 the length of the longest name used. The first two bytes of each entry are the
4622 group number, most significant byte first. There follows the corresponding
4623 name, zero terminated. The names are in alphabetical order.
4624
4625 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
4626 case instead of 255. Making it the same both with and without UTF-8 support
4627 means that the same test output works with both.
4628
4629 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
4630 calling malloc() with a zero argument.
4631
4632 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
4633 optimization. I've improved things by keeping a bitmap of backreferences with
4634 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
4635 fact referenced, the optimization can be applied. It is unlikely that a
4636 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
4637 the match to follow \n) will appear inside brackets with a number greater than
4638 31, but if it does, any back reference > 31 suppresses the optimization.
4639
4640 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
4641 of disabling numbered capturing parentheses. Any opening parenthesis that is
4642 not followed by ? behaves as if it were followed by ?: but named parentheses
4643 can still be used for capturing (and they will acquire numbers in the usual
4644 way).
4645
4646 41. Redesigned the return codes from the match() function into yes/no/error so
4647 that errors can be passed back from deep inside the nested calls. A malloc
4648 failure while inside a recursive subpattern call now causes the
4649 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
4650
4651 42. It is now possible to set a limit on the number of times the match()
4652 function is called in a call to pcre_exec(). This facility makes it possible to
4653 limit the amount of recursion and backtracking, though not in a directly
4654 obvious way, because the match() function is used in a number of different
4655 circumstances. The count starts from zero for each position in the subject
4656 string (for non-anchored patterns). The default limit is, for compatibility, a
4657 large number, namely 10 000 000. You can change this in two ways:
4658
4659 (a) When configuring PCRE before making, you can use --with-match-limit=n
4660 to set a default value for the compiled library.
4661
4662 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
4663 a different value is set. See 45 below.
4664
4665 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
4666
4667 43. Added a new function pcre_config(int, void *) to enable run-time extraction
4668 of things that can be changed at compile time. The first argument specifies
4669 what is wanted and the second points to where the information is to be placed.
4670 The current list of available information is:
4671
4672 PCRE_CONFIG_UTF8
4673
4674 The output is an integer that is set to one if UTF-8 support is available;
4675 otherwise it is set to zero.
4676
4677 PCRE_CONFIG_NEWLINE
4678
4679 The output is an integer that it set to the value of the code that is used for
4680 newline. It is either LF (10) or CR (13).
4681
4682 PCRE_CONFIG_LINK_SIZE
4683
4684 The output is an integer that contains the number of bytes used for internal
4685 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
4686
4687 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
4688
4689 The output is an integer that contains the threshold above which the POSIX
4690 interface uses malloc() for output vectors. See item 31 above.
4691
4692 PCRE_CONFIG_MATCH_LIMIT
4693
4694 The output is an unsigned integer that contains the default limit of the number
4695 of match() calls in a pcre_exec() execution. See 42 above.
4696
4697 44. pcretest has been upgraded by the addition of the -C option. This causes it
4698 to extract all the available output from the new pcre_config() function, and to
4699 output it. The program then exits immediately.
4700
4701 45. A need has arisen to pass over additional data with calls to pcre_exec() in
4702 order to support additional features. One way would have been to define
4703 pcre_exec2() (for example) with extra arguments, but this would not have been
4704 extensible, and would also have required all calls to the original function to
4705 be mapped to the new one. Instead, I have chosen to extend the mechanism that
4706 is used for passing in "extra" data from pcre_study().
4707
4708 The pcre_extra structure is now exposed and defined in pcre.h. It currently
4709 contains the following fields:
4710
4711 flags a bitmap indicating which of the following fields are set
4712 study_data opaque data from pcre_study()
4713 match_limit a way of specifying a limit on match() calls for a specific
4714 call to pcre_exec()
4715 callout_data data for callouts (see 49 below)
4716
4717 The flag bits are also defined in pcre.h, and are
4718
4719 PCRE_EXTRA_STUDY_DATA
4720 PCRE_EXTRA_MATCH_LIMIT
4721 PCRE_EXTRA_CALLOUT_DATA
4722
4723 The pcre_study() function now returns one of these new pcre_extra blocks, with
4724 the actual study data pointed to by the study_data field, and the
4725 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
4726 before. That is, this change is entirely upwards-compatible and requires no
4727 change to existing code.
4728
4729 If you want to pass in additional data to pcre_exec(), you can either place it
4730 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
4731 block.
4732
4733 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
4734 data string contains the escape sequence \M, pcretest calls pcre_exec() several
4735 times with different match limits, until it finds the minimum value needed for
4736 pcre_exec() to complete. The value is then output. This can be instructive; for
4737 most simple matches the number is quite small, but for pathological cases it
4738 gets very large very quickly.
4739
4740 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
4741 returns the size of the data block pointed to by the study_data field in a
4742 pcre_extra block, that is, the value that was passed as the argument to
4743 pcre_malloc() when PCRE was getting memory in which to place the information
4744 created by pcre_study(). The fourth argument should point to a size_t variable.
4745 pcretest has been extended so that this information is shown after a successful
4746 pcre_study() call when information about the compiled regex is being displayed.
4747
4748 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
4749 because what follows is always an absolute path. (Later: it turns out that this
4750 is more than cosmetic for MinGW, because it doesn't like empty path
4751 components.)
4752
4753 49. Some changes have been made to the callout feature (see 28 above):
4754
4755 (i) A callout function now has three choices for what it returns:
4756
4757 0 => success, carry on matching
4758 > 0 => failure at this point, but backtrack if possible
4759 < 0 => serious error, return this value from pcre_exec()
4760
4761 Negative values should normally be chosen from the set of PCRE_ERROR_xxx
4762 values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
4763 "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
4764 use by callout functions. It will never be used by PCRE itself.
4765
4766 (ii) The pcre_extra structure (see 45 above) has a void * field called
4767 callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
4768 pcre_callout_block structure has a field of the same name. The contents of
4769 the field passed in the pcre_extra structure are passed to the callout
4770 function in the corresponding field in the callout block. This makes it
4771 easier to use the same callout-containing regex from multiple threads. For
4772 testing, the pcretest program has a new data escape
4773
4774 \C*n pass the number n (may be negative) as callout_data
4775
4776 If the callout function in pcretest receives a non-zero value as
4777 callout_data, it returns that value.
4778
4779 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
4780 there were some redundant $(CFLAGS) in commands that are now specified as
4781 $(LINK), which already includes $(CFLAGS).
4782
4783 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
4784 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
4785 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
4786 one-byte characters throughout. Note that case-insensitive matching applies
4787 only to characters whose values are less than 256. PCRE doesn't support the
4788 notion of cases for higher-valued characters.
4789
4790 (i) A character class whose characters are all within 0-255 is handled as
4791 a bit map, and the map is inverted for negative classes. Previously, a
4792 character > 255 always failed to match such a class; however it should
4793 match if the class was a negative one (e.g. [^ab]). This has been fixed.
4794
4795 (ii) A negated character class with a single character < 255 is coded as
4796 "not this character" (OP_NOT). This wasn't working properly when the test
4797 character was multibyte, either singly or repeated.
4798
4799 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
4800 mode, for example: \x{100}{2,3}.
4801
4802 (iv) The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
4803 singly or repeated) now correctly test multibyte characters. However,
4804 PCRE doesn't recognize any characters with values greater than 255 as
4805 digits, spaces, or word characters. Such characters always match \D, \S,
4806 and \W, and never match \d, \s, or \w.
4807
4808 (v) Classes may now contain characters and character ranges with values
4809 greater than 255. For example: [ab\x{100}-\x{400}].
4810
4811 (vi) pcregrep now has a --utf-8 option (synonym -u) which makes it call
4812 PCRE in UTF-8 mode.
4813
4814 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
4815 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
4816 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
4817 value.)
4818
4819 53. The single man page has become too large. I have therefore split it up into
4820 a number of separate man pages. These also give rise to individual HTML pages;
4821 these are now put in a separate directory, and there is an index.html page that
4822 lists them all. Some hyperlinking between the pages has been installed.
4823
4824 54. Added convenience functions for handling named capturing parentheses.
4825
4826 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
4827 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
4828 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
4829 are faulted.
4830
4831 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
4832 calling configure. These values are used when compiling the dftables.c program
4833 which is run to generate the source of the default character tables. They
4834 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
4835 you will need to set these values.
4836
4837 57. Updated the building process for Windows DLL, as provided by Fred Cox.
4838
4839
4840 Version 3.9 02-Jan-02
4841 ---------------------
4842
4843 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
4844
4845 2. If --disable-static was given, the building process failed when trying to
4846 build pcretest and pcregrep. (For some reason it was using libtool to compile
4847 them, which is not right, as they aren't part of the library.)
4848
4849
4850 Version 3.8 18-Dec-01
4851 ---------------------
4852
4853 1. The experimental UTF-8 code was completely screwed up. It was packing the
4854 bytes in the wrong order. How dumb can you get?
4855
4856
4857 Version 3.7 29-Oct-01
4858 ---------------------
4859
4860 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
4861 This caused pcretest, when used on the test data, to segfault. Unfortunately,
4862 this didn't happen under Solaris 8, where I normally test things.
4863
4864 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
4865 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
4866 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
4867 this fix an hour or so after the initial 3.7 release.)
4868
4869
4870 Version 3.6 23-Oct-01
4871 ---------------------
4872
4873 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
4874 offsets passed as NULL with zero offset count.
4875
4876 2. The config.guess and config.sub files had not been updated when I moved to
4877 the latest autoconf.
4878
4879
4880 Version 3.5 15-Aug-01
4881 ---------------------
4882
4883 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
4884 had been forgotten.
4885
4886 2. By using declared but undefined structures, we can avoid using "void"
4887 definitions in pcre.h while keeping the internal definitions of the structures
4888 private.
4889
4890 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
4891 user point of view, this means that both static and shared libraries are built
4892 by default, but this can be individually controlled. More of the work of
4893 handling this static/shared cases is now inside libtool instead of PCRE's make
4894 file.
4895
4896 4. The pcretest utility is now installed along with pcregrep because it is
4897 useful for users (to test regexs) and by doing this, it automatically gets
4898 relinked by libtool. The documentation has been turned into a man page, so
4899 there are now .1, .txt, and .html versions in /doc.
4900
4901 5. Upgrades to pcregrep:
4902 (i) Added long-form option names like gnu grep.
4903 (ii) Added --help to list all options with an explanatory phrase.
4904 (iii) Added -r, --recursive to recurse into sub-directories.
4905 (iv) Added -f, --file to read patterns from a file.
4906
4907 6. pcre_exec() was referring to its "code" argument before testing that
4908 argument for NULL (and giving an error if it was NULL).
4909
4910 7. Upgraded Makefile.in to allow for compiling in a different directory from
4911 the source directory.
4912
4913 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
4914 options bits, the pointer it was passed was to an int instead of to an unsigned
4915 long int. This mattered only on 64-bit systems.
4916
4917 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
4918 generated) instead of pcre.in, which it its source. Also made the same change
4919 in several of the .c files.
4920
4921 10. A new release of gcc defines printf() as a macro, which broke pcretest
4922 because it had an ifdef in the middle of a string argument for printf(). Fixed
4923 by using separate calls to printf().
4924
4925 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
4926 script, to force use of CR or LF instead of \n in the source. On non-Unix
4927 systems, the value can be set in config.h.
4928
4929 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
4930 absolute limit. Changed the text of the error message to make this clear, and
4931 likewise updated the man page.
4932
4933 13. The limit of 99 on the number of capturing subpatterns has been removed.
4934 The new limit is 65535, which I hope will not be a "real" limit.
4935
4936
4937 Version 3.4 22-Aug-00
4938 ---------------------
4939
4940 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
4941
4942 2. Diagnose condition (?(0) as an error instead of crashing on matching.
4943
4944
4945 Version 3.3 01-Aug-00
4946 ---------------------
4947
4948 1. If an octal character was given, but the value was greater than \377, it
4949 was not getting masked to the least significant bits, as documented. This could
4950 lead to crashes in some systems.
4951
4952 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
4953 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
4954
4955 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
4956 These just pass their arguments on to (pcre_free)(), but they are provided
4957 because some uses of PCRE bind it to non-C systems that can call its functions,
4958 but cannot call free() or pcre_free() directly.
4959
4960 4. Add "make test" as a synonym for "make check". Corrected some comments in
4961 the Makefile.
4962
4963 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
4964 Makefile.
4965
4966 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
4967 command called pgrep for grepping around the active processes.
4968
4969 7. Added the beginnings of support for UTF-8 character strings.
4970
4971 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
4972 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
4973 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
4974 out for the ar command.)
4975
4976
4977 Version 3.2 12-May-00
4978 ---------------------
4979
4980 This is purely a bug fixing release.
4981
4982 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
4983 of ZA. This was just one example of several cases that could provoke this bug,
4984 which was introduced by change 9 of version 2.00. The code for breaking
4985 infinite loops after an iteration that matches an empty string was't working
4986 correctly.
4987
4988 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
4989 when matched against abbab (for example). After matching an empty string, it
4990 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
4991 caused it to match further down the string than it should.
4992
4993 3. The code contained an inclusion of sys/types.h. It isn't clear why this
4994 was there because it doesn't seem to be needed, and it causes trouble on some
4995 systems, as it is not a Standard C header. It has been removed.
4996
4997 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
4998 were reported on the Macintosh. The changes were from
4999
5000 while ((c = *(++ptr)) != 0 && c != '\n');
5001 to
5002 while ((c = *(++ptr)) != 0 && c != '\n') ;
5003
5004 Totally extraordinary, but if that's what it takes...
5005
5006 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
5007 available. Added HAVE_BCOPY and an autoconf test for it; if neither
5008 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
5009 assumes the way PCRE uses memmove() (always moving upwards).
5010
5011 6. PCRE is being used in one environment where strchr() is not available. There
5012 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
5013 faster code anyway.
5014
5015
5016 Version 3.1 09-Feb-00
5017 ---------------------
5018
5019 The only change in this release is the fixing of some bugs in Makefile.in for
5020 the "install" target:
5021
5022 (1) It was failing to install pcreposix.h.
5023
5024 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
5025
5026
5027 Version 3.0 01-Feb-00
5028 ---------------------
5029
5030 1. Add support for the /+ modifier to perltest (to output $` like it does in
5031 pcretest).
5032
5033 2. Add support for the /g modifier to perltest.
5034
5035 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
5036 matches null strings.
5037
5038 4. Fix perltest so that it doesn't do unwanted things when fed an empty
5039 pattern. Perl treats empty patterns specially - it reuses the most recent
5040 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
5041 effect.
5042
5043 5. The POSIX interface was broken in that it was just handing over the POSIX
5044 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
5045 required a bigger vector, with some working space on the end. This means that
5046 the POSIX wrapper now has to get and free some memory, and copy the results.
5047
5048 6. Added some simple autoconf support, placing the test data and the
5049 documentation in separate directories, re-organizing some of the
5050 information files, and making it build pcre-config (a GNU standard). Also added
5051 libtool support for building PCRE as a shared library, which is now the
5052 default.
5053
5054 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
5055 09 are not valid octal constants. Single digits will be used for minor values
5056 less than 10.
5057
5058 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
5059 existing programs that set these in the POSIX interface can use PCRE without
5060 modification.
5061
5062 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
5063 return all that pcre_info() returns, plus additional data. The pcre_info()
5064 function is retained for compatibility, but is considered to be obsolete.
5065
5066 10. Added experimental recursion feature (?R) to handle one common case that
5067 Perl 5.6 will be able to do with (?p{...}).
5068
5069 11. Added support for POSIX character classes like [:alpha:], which Perl is
5070 adopting.
5071
5072
5073 Version 2.08 31-Aug-99
5074 ----------------------
5075
5076 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
5077 trying to match at the startoffset position, but instead was moving forward to
5078 the next newline as if a previous match had failed.
5079
5080 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
5081 and could get into a loop if a null string was matched other than at the start
5082 of the subject.
5083
5084 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
5085 be distinguished at compile time, and for completeness also added PCRE_DATE.
5086
5087 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
5088 in GnuWin32 environments.
5089
5090
5091 Version 2.07 29-Jul-99
5092 ----------------------
5093
5094 1. The documentation is now supplied in plain text form and HTML as well as in
5095 the form of man page sources.
5096
5097 2. C++ compilers don't like assigning (void *) values to other pointer types.
5098 In particular this affects malloc(). Although there is no problem in Standard
5099 C, I've put in casts to keep C++ compilers happy.
5100
5101 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
5102 should be (const char *).
5103
5104 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
5105 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
5106 However, I haven't made this a standard facility. The documentation doesn't
5107 mention it, and the Makefile doesn't support it.
5108
5109 5. The Makefile now contains an "install" target, with editable destinations at
5110 the top of the file. The pcretest program is not installed.
5111
5112 6. pgrep -V now gives the PCRE version number and date.
5113
5114 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
5115 causing the entire string to be ignored, instead of just the last character.
5116
5117 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
5118 non-matching string, it can take a very, very long time, even for strings of
5119 quite modest length, because of the nested recursion. PCRE now does better in
5120 some of these cases. It does this by remembering the last required literal
5121 character in the pattern, and pre-searching the subject to ensure it is present
5122 before running the real match. In other words, it applies a heuristic to detect
5123 some types of certain failure quickly, and in the above example, if presented
5124 with a string that has no trailing " it gives "no match" very quickly.
5125
5126 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
5127 other alternatives are tried instead.
5128
5129
5130 Version 2.06 09-Jun-99
5131 ----------------------
5132
5133 1. Change pcretest's output for amount of store used to show just the code
5134 space, because the remainder (the data block) varies in size between 32-bit and
5135 64-bit systems.
5136
5137 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
5138 start matching at. This allows lookbehinds to work when searching for multiple
5139 occurrences in a string.
5140
5141 3. Added additional options to pcretest for testing multiple occurrences:
5142
5143 /+ outputs the rest of the string that follows a match
5144 /g loops for multiple occurrences, using the new startoffset argument
5145 /G loops for multiple occurrences by passing an incremented pointer
5146
5147 4. PCRE wasn't doing the "first character" optimization for patterns starting
5148 with \b or \B, though it was doing it for other lookbehind assertions. That is,
5149 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
5150 the letter 'x'. On long subject strings, this gives a significant speed-up.
5151
5152
5153 Version 2.05 21-Apr-99
5154 ----------------------
5155
5156 1. Changed the type of magic_number from int to long int so that it works
5157 properly on 16-bit systems.
5158
5159 2. Fixed a bug which caused patterns starting with .* not to work correctly
5160 when the subject string contained newline characters. PCRE was assuming
5161 anchoring for such patterns in all cases, which is not correct because .* will
5162 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
5163 DOTALL is set at top level; otherwise it knows that patterns starting with .*
5164 must be retried after every newline in the subject.
5165
5166
5167 Version 2.04 18-Feb-99
5168 ----------------------
5169
5170 1. For parenthesized subpatterns with repeats whose minimum was zero, the
5171 computation of the store needed to hold the pattern was incorrect (too large).
5172 If such patterns were nested a few deep, this could multiply and become a real
5173 problem.
5174
5175 2. Added /M option to pcretest to show the memory requirement of a specific
5176 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
5177
5178 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
5179 compiled in such a way that the backtracking after subsequent failure was
5180 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
5181 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
5182
5183
5184 Version 2.03 02-Feb-99
5185 ----------------------
5186
5187 1. Fixed typo and small mistake in man page.
5188
5189 2. Added 4th condition (GPL supersedes if conflict) and created separate
5190 LICENCE file containing the conditions.
5191
5192 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
5193 Perl, that is the internal \ allows the delimiter to be included in the
5194 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
5195 the final delimiter, add \ to the end of the pattern (to test the error).
5196
5197 4. Added the convenience functions for extracting substrings after a successful
5198 match. Updated pcretest to make it able to test these functions.
5199
5200
5201 Version 2.02 14-Jan-99
5202 ----------------------
5203
5204 1. Initialized the working variables associated with each extraction so that
5205 their saving and restoring doesn't refer to uninitialized store.
5206
5207 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
5208 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
5209 fix the problem.
5210
5211 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
5212 calls, and wasn't printing the correct value for compiling calls. Increased the
5213 default value of LOOPREPEAT, and the number of significant figures in the
5214 times.
5215
5216 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
5217
5218 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
5219 a building problem on Windows NT with a FAT file system.
5220
5221
5222 Version 2.01 21-Oct-98
5223 ----------------------
5224
5225 1. Changed the API for pcre_compile() to allow for the provision of a pointer
5226 to character tables built by pcre_maketables() in the current locale. If NULL
5227 is passed, the default tables are used.
5228
5229
5230 Version 2.00 24-Sep-98
5231 ----------------------
5232
5233 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
5234 it any more.
5235
5236 2. Allow quantification of (?>) groups, and make it work correctly.
5237
5238 3. The first character computation wasn't working for (?>) groups.
5239
5240 4. Correct the implementation of \Z (it is permitted to match on the \n at the
5241 end of the subject) and add 5.005's \z, which really does match only at the
5242 very end of the subject.
5243
5244 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
5245
5246 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
5247 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
5248 localized options. All options to pcre_study() were also removed.
5249
5250 7. Add other new features from 5.005:
5251
5252 $(?<= positive lookbehind
5253 $(?<! negative lookbehind
5254 (?imsx-imsx) added the unsetting capability
5255 such a setting is global if at outer level; local otherwise
5256 (?imsx-imsx:) non-capturing groups with option setting
5257 (?(cond)re|re) conditional pattern matching
5258
5259 A backreference to itself in a repeated group matches the previous
5260 captured string.
5261
5262 8. General tidying up of studying (both automatic and via "study")
5263 consequential on the addition of new assertions.
5264
5265 9. As in 5.005, unlimited repeated groups that could match an empty substring
5266 are no longer faulted at compile time. Instead, the loop is forcibly broken at
5267 runtime if any iteration does actually match an empty substring.
5268
5269 10. Include the RunTest script in the distribution.
5270
5271 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
5272 discrepancies, some of which were old and were also with respect to 5.004. They
5273 have now been fixed.
5274
5275
5276 Version 1.09 28-Apr-98
5277 ----------------------
5278
5279 1. A negated single character class followed by a quantifier with a minimum
5280 value of one (e.g. [^x]{1,6} ) was not compiled correctly. This could lead to
5281 program crashes, or just wrong answers. This did not apply to negated classes
5282 containing more than one character, or to minima other than one.
5283
5284
5285 Version 1.08 27-Mar-98
5286 ----------------------
5287
5288 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
5289
5290 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
5291 latter must appear before anything that relies on it in the pattern.
5292
5293
5294 Version 1.07 16-Feb-98
5295 ----------------------
5296
5297 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
5298 repeat of a potentially empty string).
5299
5300
5301 Version 1.06 23-Jan-98
5302 ----------------------
5303
5304 1. Added Markus Oberhumer's little patches for C++.
5305
5306 2. Literal strings longer than 255 characters were broken.
5307
5308
5309 Version 1.05 23-Dec-97
5310 ----------------------
5311
5312 1. Negated character classes containing more than one character were failing if
5313 PCRE_CASELESS was set at run time.
5314
5315
5316 Version 1.04 19-Dec-97
5317 ----------------------
5318
5319 1. Corrected the man page, where some "const" qualifiers had been omitted.
5320
5321 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
5322 input syntax.
5323
5324 3. Fixed memory leak which occurred when a regex with back references was
5325 matched with an offsets vector that wasn't big enough. The temporary memory
5326 that is used in this case wasn't being freed if the match failed.
5327
5328 4. Tidied pcretest to ensure it frees memory that it gets.
5329
5330 5. Temporary memory was being obtained in the case where the passed offsets
5331 vector was exactly big enough.
5332
5333 6. Corrected definition of offsetof() from change 5 below.
5334
5335 7. I had screwed up change 6 below and broken the rules for the use of
5336 setjmp(). Now fixed.
5337
5338
5339 Version 1.03 18-Dec-97
5340 ----------------------
5341
5342 1. A erroneous regex with a missing opening parenthesis was correctly
5343 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
5344 on some systems.
5345
5346 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
5347 it was reported that one broken compiler failed on the former because "code" is
5348 also an independent variable.
5349
5350 3. The erroneous regex a[]b caused an array overrun reference.
5351
5352 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
5353 fail on data ending with that character. (It was going on too far, and checking
5354 the next character, typically a binary zero.) This was specific to the
5355 optimized code for single-character negative classes.
5356
5357 5. Added a contributed patch from the TIN world which does the following:
5358
5359 + Add an undef for memmove, in case the the system defines a macro for it.
5360
5361 + Add a definition of offsetof(), in case there isn't one. (I don't know
5362 the reason behind this - offsetof() is part of the ANSI standard - but
5363 it does no harm).
5364
5365 + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
5366 most of the places where whitespace preceded '#'. I have given up and
5367 allowed the remaining 2 cases to be at the margin.
5368
5369 + Rename some variables in pcre to eliminate shadowing. This seems very
5370 pedantic, but does no harm, of course.
5371
5372 6. Moved the call to setjmp() into its own function, to get rid of warnings
5373 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
5374
5375 7. Constructs such as \d{8,} were compiling into the equivalent of
5376 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
5377 outcome, but in this particular case used more store than had been allocated,
5378 which caused the bug to be discovered because it threw up an internal error.
5379
5380 8. The debugging code in both pcre and pcretest for outputting the compiled
5381 form of a regex was going wrong in the case of back references followed by
5382 curly-bracketed repeats.
5383
5384
5385 Version 1.02 12-Dec-97
5386 ----------------------
5387
5388 1. Typos in pcre.3 and comments in the source fixed.
5389
5390 2. Applied a contributed patch to get rid of places where it used to remove
5391 'const' from variables, and fixed some signed/unsigned and uninitialized
5392 variable warnings.
5393
5394 3. Added the "runtest" target to Makefile.
5395
5396 4. Set default compiler flag to -O2 rather than just -O.
5397
5398
5399 Version 1.01 19-Nov-97
5400 ----------------------
5401
5402 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
5403 like /([ab]*)*/, that is, for classes with more than one character in them.
5404
5405 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
5406 as /((?>a*))*/ (a PCRE_EXTRA facility).
5407
5408
5409 Version 1.00 18-Nov-97
5410 ----------------------
5411
5412 1. Added compile-time macros to support systems such as SunOS4 which don't have
5413 memmove() or strerror() but have other things that can be used instead.
5414
5415 2. Arranged that "make clean" removes the executables.
5416
5417
5418 Version 0.99 27-Oct-97
5419 ----------------------
5420
5421 1. Fixed bug in code for optimizing classes with only one character. It was
5422 initializing a 32-byte map regardless, which could cause it to run off the end
5423 of the memory it had got.
5424
5425 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
5426
5427
5428 Version 0.98 22-Oct-97
5429 ----------------------
5430
5431 1. Fixed bug in code for handling temporary memory usage when there are more
5432 back references than supplied space in the ovector. This could cause segfaults.
5433
5434
5435 Version 0.97 21-Oct-97
5436 ----------------------
5437
5438 1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
5439
5440 2. Optimized negated single characters not to use a bit map.
5441
5442 3. Brought error texts together as macro definitions; clarified some of them;
5443 fixed one that was wrong - it said "range out of order" when it meant "invalid
5444 escape sequence".
5445
5446 4. Changed some char * arguments to const char *.
5447
5448 5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
5449
5450 6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
5451 pcretest.
5452
5453
5454 Version 0.96 16-Oct-97
5455 ----------------------
5456
5457 1. Added a simple "pgrep" utility to the distribution.
5458
5459 2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
5460 unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
5461 where "ddd" means "one or more decimal digits".
5462
5463 3. Fixed serious bug. If a pattern had a back reference, but the call to
5464 pcre_exec() didn't supply a large enough ovector to record the related
5465 identifying subpattern, the match always failed. PCRE now remembers the number
5466 of the largest back reference, and gets some temporary memory in which to save
5467 the offsets during matching if necessary, in order to ensure that
5468 backreferences always work.
5469
5470 4. Increased the compatibility with Perl in a number of ways:
5471
5472 (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
5473 to request this handling. The option can be set at compile or exec time.
5474
5475 (b) $ matches before a terminating newline by default; an option
5476 PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
5477 mode). The option can be set at compile or exec time.
5478
5479 (c) The handling of \ followed by a digit other than 0 is now supposed to be
5480 the same as Perl's. If the decimal number it represents is less than 10
5481 or there aren't that many previous left capturing parentheses, an octal
5482 escape is read. Inside a character class, it's always an octal escape,
5483 even if it is a single digit.
5484
5485 (d) An escaped but undefined alphabetic character is taken as a literal,
5486 unless PCRE_EXTRA is set. Currently this just reserves the remaining
5487 escapes.
5488
5489 (e) {0} is now permitted. (The previous item is removed from the compiled
5490 pattern).
5491
5492 5. Changed all the names of code files so that the basic parts are no longer
5493 than 10 characters, and abolished the teeny "globals.c" file.
5494
5495 6. Changed the handling of character classes; they are now done with a 32-byte
5496 bit map always.
5497
5498 7. Added the -d and /D options to pcretest to make it possible to look at the
5499 internals of compilation without having to recompile pcre.
5500
5501
5502 Version 0.95 23-Sep-97
5503 ----------------------
5504
5505 1. Fixed bug in pre-pass concerning escaped "normal" characters such as \x5c or
5506 \x20 at the start of a run of normal characters. These were being treated as
5507 real characters, instead of the source characters being re-checked.
5508
5509
5510 Version 0.94 18-Sep-97
5511 ----------------------
5512
5513 1. The functions are now thread-safe, with the caveat that the global variables
5514 containing pointers to malloc() and free() or alternative functions are the
5515 same for all threads.
5516
5517 2. Get pcre_study() to generate a bitmap of initial characters for non-
5518 anchored patterns when this is possible, and use it if passed to pcre_exec().
5519
5520
5521 Version 0.93 15-Sep-97
5522 ----------------------
5523
5524 1. /(b)|(:+)/ was computing an incorrect first character.
5525
5526 2. Add pcre_study() to the API and the passing of pcre_extra to pcre_exec(),
5527 but not actually doing anything yet.
5528
5529 3. Treat "-" characters in classes that cannot be part of ranges as literals,
5530 as Perl does (e.g. [-az] or [az-]).
5531
5532 4. Set the anchored flag if a branch starts with .* or .*? because that tests
5533 all possible positions.
5534
5535 5. Split up into different modules to avoid including unneeded functions in a
5536 compiled binary. However, compile and exec are still in one module. The "study"
5537 function is split off.
5538
5539 6. The character tables are now in a separate module whose source is generated
5540 by an auxiliary program - but can then be edited by hand if required. There are
5541 now no calls to isalnum(), isspace(), isdigit(), isxdigit(), tolower() or
5542 toupper() in the code.
5543
5544 7. Turn the malloc/free funtions variables into pcre_malloc and pcre_free and
5545 make them global. Abolish the function for setting them, as the caller can now
5546 set them directly.
5547
5548
5549 Version 0.92 11-Sep-97
5550 ----------------------
5551
5552 1. A repeat with a fixed maximum and a minimum of 1 for an ordinary character
5553 (e.g. /a{1,3}/) was broken (I mis-optimized it).
5554
5555 2. Caseless matching was not working in character classes if the characters in
5556 the pattern were in upper case.
5557
5558 3. Make ranges like [W-c] work in the same way as Perl for caseless matching.
5559
5560 4. Make PCRE_ANCHORED public and accept as a compile option.
5561
5562 5. Add an options word to pcre_exec() and accept PCRE_ANCHORED and
5563 PCRE_CASELESS at run time. Add escapes \A and \I to pcretest to cause it to
5564 pass them.
5565
5566 6. Give an error if bad option bits passed at compile or run time.
5567
5568 7. Add PCRE_MULTILINE at compile and exec time, and (?m) as well. Add \M to
5569 pcretest to cause it to pass that flag.
5570
5571 8. Add pcre_info(), to get the number of identifying subpatterns, the stored
5572 options, and the first character, if set.
5573
5574 9. Recognize C+ or C{n,m} where n >= 1 as providing a fixed starting character.
5575
5576
5577 Version 0.91 10-Sep-97
5578 ----------------------
5579
5580 1. PCRE was failing to diagnose unlimited repeats of subpatterns that could
5581 match the empty string as in /(a*)*/. It was looping and ultimately crashing.
5582
5583 2. PCRE was looping on encountering an indefinitely repeated back reference to
5584 a subpattern that had matched an empty string, e.g. /(a|)\1*/. It now does what
5585 Perl does - treats the match as successful.
5586
5587 ****

Properties

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

  ViewVC Help
Powered by ViewVC 1.1.5