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

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1519 - (show annotations)
Fri Feb 6 17:55:54 2015 UTC (4 years, 10 months ago) by ph10
File size: 269968 byte(s)
Error occurred while calculating annotation data.
Fix compile loop for subroutine reference when preceded by a non-ASCII 
single-character class such as [^\ff].
1 ChangeLog for PCRE
2 ------------------
3
4 Version 8.37 xx-xxx-2015
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 2. If an assertion condition was quantified with a minimum of zero (an odd
15 thing to do, but it happened), SIGSEGV or other misbehaviour could occur.
16
17 3. If a pattern in pcretest input had the P (POSIX) modifier followed by an
18 unrecognized modifier, a crash could occur.
19
20 4. An attempt to do global matching in pcretest with a zero-length ovector
21 caused a crash.
22
23 5. Fixed a memory leak during matching that could occur for a subpattern
24 subroutine call (recursive or otherwise) if the number of captured groups
25 that had to be saved was greater than ten.
26
27 6. Catch a bad opcode during auto-possessification after compiling a bad UTF
28 string with NO_UTF_CHECK. This is a tidyup, not a bug fix, as passing bad
29 UTF with NO_UTF_CHECK is documented as having an undefined outcome.
30
31 7. A UTF pattern containing a "not" match of a non-ASCII character and a
32 subroutine reference could loop at compile time. Example: /[^\xff]((?1))/.
33
34
35 Version 8.36 26-September-2014
36 ------------------------------
37
38 1. Got rid of some compiler warnings in the C++ modules that were shown up by
39 -Wmissing-field-initializers and -Wunused-parameter.
40
41 2. The tests for quantifiers being too big (greater than 65535) were being
42 applied after reading the number, and stupidly assuming that integer
43 overflow would give a negative number. The tests are now applied as the
44 numbers are read.
45
46 3. Tidy code in pcre_exec.c where two branches that used to be different are
47 now the same.
48
49 4. The JIT compiler did not generate match limit checks for certain
50 bracketed expressions with quantifiers. This may lead to exponential
51 backtracking, instead of returning with PCRE_ERROR_MATCHLIMIT. This
52 issue should be resolved now.
53
54 5. Fixed an issue, which occures when nested alternatives are optimized
55 with table jumps.
56
57 6. Inserted two casts and changed some ints to size_t in the light of some
58 reported 64-bit compiler warnings (Bugzilla 1477).
59
60 7. Fixed a bug concerned with zero-minimum possessive groups that could match
61 an empty string, which sometimes were behaving incorrectly in the
62 interpreter (though correctly in the JIT matcher). This pcretest input is
63 an example:
64
65 '\A(?:[^"]++|"(?:[^"]*+|"")*+")++'
66 NON QUOTED "QUOT""ED" AFTER "NOT MATCHED
67
68 the interpreter was reporting a match of 'NON QUOTED ' only, whereas the
69 JIT matcher and Perl both matched 'NON QUOTED "QUOT""ED" AFTER '. The test
70 for an empty string was breaking the inner loop and carrying on at a lower
71 level, when possessive repeated groups should always return to a higher
72 level as they have no backtrack points in them. The empty string test now
73 occurs at the outer level.
74
75 8. Fixed a bug that was incorrectly auto-possessifying \w+ in the pattern
76 ^\w+(?>\s*)(?<=\w) which caused it not to match "test test".
77
78 9. Give a compile-time error for \o{} (as Perl does) and for \x{} (which Perl
79 doesn't).
80
81 10. Change 8.34/15 introduced a bug that caused the amount of memory needed
82 to hold a pattern to be incorrectly computed (too small) when there were
83 named back references to duplicated names. This could cause "internal
84 error: code overflow" or "double free or corruption" or other memory
85 handling errors.
86
87 11. When named subpatterns had the same prefixes, back references could be
88 confused. For example, in this pattern:
89
90 /(?P<Name>a)?(?P<Name2>b)?(?(<Name>)c|d)*l/
91
92 the reference to 'Name' was incorrectly treated as a reference to a
93 duplicate name.
94
95 12. A pattern such as /^s?c/mi8 where the optional character has more than
96 one "other case" was incorrectly compiled such that it would only try to
97 match starting at "c".
98
99 13. When a pattern starting with \s was studied, VT was not included in the
100 list of possible starting characters; this should have been part of the
101 8.34/18 patch.
102
103 14. If a character class started [\Qx]... where x is any character, the class
104 was incorrectly terminated at the ].
105
106 15. If a pattern that started with a caseless match for a character with more
107 than one "other case" was studied, PCRE did not set up the starting code
108 unit bit map for the list of possible characters. Now it does. This is an
109 optimization improvement, not a bug fix.
110
111 16. The Unicode data tables have been updated to Unicode 7.0.0.
112
113 17. Fixed a number of memory leaks in pcregrep.
114
115 18. Avoid a compiler warning (from some compilers) for a function call with
116 a cast that removes "const" from an lvalue by using an intermediate
117 variable (to which the compiler does not object).
118
119 19. Incorrect code was compiled if a group that contained an internal recursive
120 back reference was optional (had quantifier with a minimum of zero). This
121 example compiled incorrect code: /(((a\2)|(a*)\g<-1>))*/ and other examples
122 caused segmentation faults because of stack overflows at compile time.
123
124 20. A pattern such as /((?(R)a|(?1)))+/, which contains a recursion within a
125 group that is quantified with an indefinite repeat, caused a compile-time
126 loop which used up all the system stack and provoked a segmentation fault.
127 This was not the same bug as 19 above.
128
129 21. Add PCRECPP_EXP_DECL declaration to operator<< in pcre_stringpiece.h.
130 Patch by Mike Frysinger.
131
132
133 Version 8.35 04-April-2014
134 --------------------------
135
136 1. A new flag is set, when property checks are present in an XCLASS.
137 When this flag is not set, PCRE can perform certain optimizations
138 such as studying these XCLASS-es.
139
140 2. The auto-possessification of character sets were improved: a normal
141 and an extended character set can be compared now. Furthermore
142 the JIT compiler optimizes more character set checks.
143
144 3. Got rid of some compiler warnings for potentially uninitialized variables
145 that show up only when compiled with -O2.
146
147 4. A pattern such as (?=ab\K) that uses \K in an assertion can set the start
148 of a match later then the end of the match. The pcretest program was not
149 handling the case sensibly - it was outputting from the start to the next
150 binary zero. It now reports this situation in a message, and outputs the
151 text from the end to the start.
152
153 5. Fast forward search is improved in JIT. Instead of the first three
154 characters, any three characters with fixed position can be searched.
155 Search order: first, last, middle.
156
157 6. Improve character range checks in JIT. Characters are read by an inprecise
158 function now, which returns with an unknown value if the character code is
159 above a certain threshold (e.g: 256). The only limitation is that the value
160 must be bigger than the threshold as well. This function is useful when
161 the characters above the threshold are handled in the same way.
162
163 7. The macros whose names start with RAWUCHAR are placeholders for a future
164 mode in which only the bottom 21 bits of 32-bit data items are used. To
165 make this more memorable for those maintaining the code, the names have
166 been changed to start with UCHAR21, and an extensive comment has been added
167 to their definition.
168
169 8. Add missing (new) files sljitNativeTILEGX.c and sljitNativeTILEGX-encoder.c
170 to the export list in Makefile.am (they were accidentally omitted from the
171 8.34 tarball).
172
173 9. The informational output from pcretest used the phrase "starting byte set"
174 which is inappropriate for the 16-bit and 32-bit libraries. As the output
175 for "first char" and "need char" really means "non-UTF-char", I've changed
176 "byte" to "char", and slightly reworded the output. The documentation about
177 these values has also been (I hope) clarified.
178
179 10. Another JIT related optimization: use table jumps for selecting the correct
180 backtracking path, when more than four alternatives are present inside a
181 bracket.
182
183 11. Empty match is not possible, when the minimum length is greater than zero,
184 and there is no \K in the pattern. JIT should avoid empty match checks in
185 such cases.
186
187 12. In a caseless character class with UCP support, when a character with more
188 than one alternative case was not the first character of a range, not all
189 the alternative cases were added to the class. For example, s and \x{17f}
190 are both alternative cases for S: the class [RST] was handled correctly,
191 but [R-T] was not.
192
193 13. The configure.ac file always checked for pthread support when JIT was
194 enabled. This is not used in Windows, so I have put this test inside a
195 check for the presence of windows.h (which was already tested for).
196
197 14. Improve pattern prefix search by a simplified Boyer-Moore algorithm in JIT.
198 The algorithm provides a way to skip certain starting offsets, and usually
199 faster than linear prefix searches.
200
201 15. Change 13 for 8.20 updated RunTest to check for the 'fr' locale as well
202 as for 'fr_FR' and 'french'. For some reason, however, it then used the
203 Windows-specific input and output files, which have 'french' screwed in.
204 So this could never have worked. One of the problems with locales is that
205 they aren't always the same. I have now updated RunTest so that it checks
206 the output of the locale test (test 3) against three different output
207 files, and it allows the test to pass if any one of them matches. With luck
208 this should make the test pass on some versions of Solaris where it was
209 failing. Because of the uncertainty, the script did not used to stop if
210 test 3 failed; it now does. If further versions of a French locale ever
211 come to light, they can now easily be added.
212
213 16. If --with-pcregrep-bufsize was given a non-integer value such as "50K",
214 there was a message during ./configure, but it did not stop. This now
215 provokes an error. The invalid example in README has been corrected.
216 If a value less than the minimum is given, the minimum value has always
217 been used, but now a warning is given.
218
219 17. If --enable-bsr-anycrlf was set, the special 16/32-bit test failed. This
220 was a bug in the test system, which is now fixed. Also, the list of various
221 configurations that are tested for each release did not have one with both
222 16/32 bits and --enable-bar-anycrlf. It now does.
223
224 18. pcretest was missing "-C bsr" for displaying the \R default setting.
225
226 19. Little endian PowerPC systems are supported now by the JIT compiler.
227
228 20. The fast forward newline mechanism could enter to an infinite loop on
229 certain invalid UTF-8 input. Although we don't support these cases
230 this issue can be fixed by a performance optimization.
231
232 21. Change 33 of 8.34 is not sufficient to ensure stack safety because it does
233 not take account if existing stack usage. There is now a new global
234 variable called pcre_stack_guard that can be set to point to an external
235 function to check stack availability. It is called at the start of
236 processing every parenthesized group.
237
238 22. A typo in the code meant that in ungreedy mode the max/min qualifier
239 behaved like a min-possessive qualifier, and, for example, /a{1,3}b/U did
240 not match "ab".
241
242 23. When UTF was disabled, the JIT program reported some incorrect compile
243 errors. These messages are silenced now.
244
245 24. Experimental support for ARM-64 and MIPS-64 has been added to the JIT
246 compiler.
247
248 25. Change all the temporary files used in RunGrepTest to be different to those
249 used by RunTest so that the tests can be run simultaneously, for example by
250 "make -j check".
251
252
253 Version 8.34 15-December-2013
254 -----------------------------
255
256 1. Add pcre[16|32]_jit_free_unused_memory to forcibly free unused JIT
257 executable memory. Patch inspired by Carsten Klein.
258
259 2. ./configure --enable-coverage defined SUPPORT_GCOV in config.h, although
260 this macro is never tested and has no effect, because the work to support
261 coverage involves only compiling and linking options and special targets in
262 the Makefile. The comment in config.h implied that defining the macro would
263 enable coverage support, which is totally false. There was also support for
264 setting this macro in the CMake files (my fault, I just copied it from
265 configure). SUPPORT_GCOV has now been removed.
266
267 3. Make a small performance improvement in strlen16() and strlen32() in
268 pcretest.
269
270 4. Change 36 for 8.33 left some unreachable statements in pcre_exec.c,
271 detected by the Solaris compiler (gcc doesn't seem to be able to diagnose
272 these cases). There was also one in pcretest.c.
273
274 5. Cleaned up a "may be uninitialized" compiler warning in pcre_exec.c.
275
276 6. In UTF mode, the code for checking whether a group could match an empty
277 string (which is used for indefinitely repeated groups to allow for
278 breaking an infinite loop) was broken when the group contained a repeated
279 negated single-character class with a character that occupied more than one
280 data item and had a minimum repetition of zero (for example, [^\x{100}]* in
281 UTF-8 mode). The effect was undefined: the group might or might not be
282 deemed as matching an empty string, or the program might have crashed.
283
284 7. The code for checking whether a group could match an empty string was not
285 recognizing that \h, \H, \v, \V, and \R must match a character.
286
287 8. Implemented PCRE_INFO_MATCH_EMPTY, which yields 1 if the pattern can match
288 an empty string. If it can, pcretest shows this in its information output.
289
290 9. Fixed two related bugs that applied to Unicode extended grapheme clusters
291 that were repeated with a maximizing qualifier (e.g. \X* or \X{2,5}) when
292 matched by pcre_exec() without using JIT:
293
294 (a) If the rest of the pattern did not match after a maximal run of
295 grapheme clusters, the code for backing up to try with fewer of them
296 did not always back up over a full grapheme when characters that do not
297 have the modifier quality were involved, e.g. Hangul syllables.
298
299 (b) If the match point in a subject started with modifier character, and
300 there was no match, the code could incorrectly back up beyond the match
301 point, and potentially beyond the first character in the subject,
302 leading to a segfault or an incorrect match result.
303
304 10. A conditional group with an assertion condition could lead to PCRE
305 recording an incorrect first data item for a match if no other first data
306 item was recorded. For example, the pattern (?(?=ab)ab) recorded "a" as a
307 first data item, and therefore matched "ca" after "c" instead of at the
308 start.
309
310 11. Change 40 for 8.33 (allowing pcregrep to find empty strings) showed up a
311 bug that caused the command "echo a | ./pcregrep -M '|a'" to loop.
312
313 12. The source of pcregrep now includes z/OS-specific code so that it can be
314 compiled for z/OS as part of the special z/OS distribution.
315
316 13. Added the -T and -TM options to pcretest.
317
318 14. The code in pcre_compile.c for creating the table of named capturing groups
319 has been refactored. Instead of creating the table dynamically during the
320 actual compiling pass, the information is remembered during the pre-compile
321 pass (on the stack unless there are more than 20 named groups, in which
322 case malloc() is used) and the whole table is created before the actual
323 compile happens. This has simplified the code (it is now nearly 150 lines
324 shorter) and prepared the way for better handling of references to groups
325 with duplicate names.
326
327 15. A back reference to a named subpattern when there is more than one of the
328 same name now checks them in the order in which they appear in the pattern.
329 The first one that is set is used for the reference. Previously only the
330 first one was inspected. This change makes PCRE more compatible with Perl.
331
332 16. Unicode character properties were updated from Unicode 6.3.0.
333
334 17. The compile-time code for auto-possessification has been refactored, based
335 on a patch by Zoltan Herczeg. It now happens after instead of during
336 compilation. The code is cleaner, and more cases are handled. The option
337 PCRE_NO_AUTO_POSSESS is added for testing purposes, and the -O and /O
338 options in pcretest are provided to set it. It can also be set by
339 (*NO_AUTO_POSSESS) at the start of a pattern.
340
341 18. The character VT has been added to the default ("C" locale) set of
342 characters that match \s and are generally treated as white space,
343 following this same change in Perl 5.18. There is now no difference between
344 "Perl space" and "POSIX space". Whether VT is treated as white space in
345 other locales depends on the locale.
346
347 19. The code for checking named groups as conditions, either for being set or
348 for being recursed, has been refactored (this is related to 14 and 15
349 above). Processing unduplicated named groups should now be as fast at
350 numerical groups, and processing duplicated groups should be faster than
351 before.
352
353 20. Two patches to the CMake build system, by Alexander Barkov:
354
355 (1) Replace the "source" command by "." in CMakeLists.txt because
356 "source" is a bash-ism.
357
358 (2) Add missing HAVE_STDINT_H and HAVE_INTTYPES_H to config-cmake.h.in;
359 without these the CMake build does not work on Solaris.
360
361 21. Perl has changed its handling of \8 and \9. If there is no previously
362 encountered capturing group of those numbers, they are treated as the
363 literal characters 8 and 9 instead of a binary zero followed by the
364 literals. PCRE now does the same.
365
366 22. Following Perl, added \o{} to specify codepoints in octal, making it
367 possible to specify values greater than 0777 and also making them
368 unambiguous.
369
370 23. Perl now gives an error for missing closing braces after \x{... instead of
371 treating the string as literal. PCRE now does the same.
372
373 24. RunTest used to grumble if an inappropriate test was selected explicitly,
374 but just skip it when running all tests. This make it awkward to run ranges
375 of tests when one of them was inappropriate. Now it just skips any
376 inappropriate tests, as it always did when running all tests.
377
378 25. If PCRE_AUTO_CALLOUT and PCRE_UCP were set for a pattern that contained
379 character types such as \d or \w, too many callouts were inserted, and the
380 data that they returned was rubbish.
381
382 26. In UCP mode, \s was not matching two of the characters that Perl matches,
383 namely NEL (U+0085) and MONGOLIAN VOWEL SEPARATOR (U+180E), though they
384 were matched by \h. The code has now been refactored so that the lists of
385 the horizontal and vertical whitespace characters used for \h and \v (which
386 are defined only in one place) are now also used for \s.
387
388 27. Add JIT support for the 64 bit TileGX architecture.
389 Patch by Jiong Wang (Tilera Corporation).
390
391 28. Possessive quantifiers for classes (both explicit and automatically
392 generated) now use special opcodes instead of wrapping in ONCE brackets.
393
394 29. Whereas an item such as A{4}+ ignored the possessivenes of the quantifier
395 (because it's meaningless), this was not happening when PCRE_CASELESS was
396 set. Not wrong, but inefficient.
397
398 30. Updated perltest.pl to add /u (force Unicode mode) when /W (use Unicode
399 properties for \w, \d, etc) is present in a test regex. Otherwise if the
400 test contains no characters greater than 255, Perl doesn't realise it
401 should be using Unicode semantics.
402
403 31. Upgraded the handling of the POSIX classes [:graph:], [:print:], and
404 [:punct:] when PCRE_UCP is set so as to include the same characters as Perl
405 does in Unicode mode.
406
407 32. Added the "forbid" facility to pcretest so that putting tests into the
408 wrong test files can sometimes be quickly detected.
409
410 33. There is now a limit (default 250) on the depth of nesting of parentheses.
411 This limit is imposed to control the amount of system stack used at compile
412 time. It can be changed at build time by --with-parens-nest-limit=xxx or
413 the equivalent in CMake.
414
415 34. Character classes such as [A-\d] or [a-[:digit:]] now cause compile-time
416 errors. Perl warns for these when in warning mode, but PCRE has no facility
417 for giving warnings.
418
419 35. Change 34 for 8.13 allowed quantifiers on assertions, because Perl does.
420 However, this was not working for (?!) because it is optimized to (*FAIL),
421 for which PCRE does not allow quantifiers. The optimization is now disabled
422 when a quantifier follows (?!). I can't see any use for this, but it makes
423 things uniform.
424
425 36. Perl no longer allows group names to start with digits, so I have made this
426 change also in PCRE. It simplifies the code a bit.
427
428 37. In extended mode, Perl ignores spaces before a + that indicates a
429 possessive quantifier. PCRE allowed a space before the quantifier, but not
430 before the possessive +. It now does.
431
432 38. The use of \K (reset reported match start) within a repeated possessive
433 group such as (a\Kb)*+ was not working.
434
435 40. Document that the same character tables must be used at compile time and
436 run time, and that the facility to pass tables to pcre_exec() and
437 pcre_dfa_exec() is for use only with saved/restored patterns.
438
439 41. Applied Jeff Trawick's patch CMakeLists.txt, which "provides two new
440 features for Builds with MSVC:
441
442 1. Support pcre.rc and/or pcreposix.rc (as is already done for MinGW
443 builds). The .rc files can be used to set FileDescription and many other
444 attributes.
445
446 2. Add an option (-DINSTALL_MSVC_PDB) to enable installation of .pdb files.
447 This allows higher-level build scripts which want .pdb files to avoid
448 hard-coding the exact files needed."
449
450 42. Added support for [[:<:]] and [[:>:]] as used in the BSD POSIX library to
451 mean "start of word" and "end of word", respectively, as a transition aid.
452
453 43. A minimizing repeat of a class containing codepoints greater than 255 in
454 non-UTF 16-bit or 32-bit modes caused an internal error when PCRE was
455 compiled to use the heap for recursion.
456
457 44. Got rid of some compiler warnings for unused variables when UTF but not UCP
458 is configured.
459
460
461 Version 8.33 28-May-2013
462 ------------------------
463
464 1. Added 'U' to some constants that are compared to unsigned integers, to
465 avoid compiler signed/unsigned warnings. Added (int) casts to unsigned
466 variables that are added to signed variables, to ensure the result is
467 signed and can be negated.
468
469 2. Applied patch by Daniel Richard G for quashing MSVC warnings to the
470 CMake config files.
471
472 3. Revise the creation of config.h.generic so that all boolean macros are
473 #undefined, whereas non-boolean macros are #ifndef/#endif-ed. This makes
474 overriding via -D on the command line possible.
475
476 4. Changing the definition of the variable "op" in pcre_exec.c from pcre_uchar
477 to unsigned int is reported to make a quite noticeable speed difference in
478 a specific Windows environment. Testing on Linux did also appear to show
479 some benefit (and it is clearly not harmful). Also fixed the definition of
480 Xop which should be unsigned.
481
482 5. Related to (4), changing the definition of the intermediate variable cc
483 in repeated character loops from pcre_uchar to pcre_uint32 also gave speed
484 improvements.
485
486 6. Fix forward search in JIT when link size is 3 or greater. Also removed some
487 unnecessary spaces.
488
489 7. Adjust autogen.sh and configure.ac to lose warnings given by automake 1.12
490 and later.
491
492 8. Fix two buffer over read issues in 16 and 32 bit modes. Affects JIT only.
493
494 9. Optimizing fast_forward_start_bits in JIT.
495
496 10. Adding support for callouts in JIT, and fixing some issues revealed
497 during this work. Namely:
498
499 (a) Unoptimized capturing brackets incorrectly reset on backtrack.
500
501 (b) Minimum length was not checked before the matching is started.
502
503 11. The value of capture_last that is passed to callouts was incorrect in some
504 cases when there was a capture on one path that was subsequently abandoned
505 after a backtrack. Also, the capture_last value is now reset after a
506 recursion, since all captures are also reset in this case.
507
508 12. The interpreter no longer returns the "too many substrings" error in the
509 case when an overflowing capture is in a branch that is subsequently
510 abandoned after a backtrack.
511
512 13. In the pathological case when an offset vector of size 2 is used, pcretest
513 now prints out the matched string after a yield of 0 or 1.
514
515 14. Inlining subpatterns in recursions, when certain conditions are fulfilled.
516 Only supported by the JIT compiler at the moment.
517
518 15. JIT compiler now supports 32 bit Macs thanks to Lawrence Velazquez.
519
520 16. Partial matches now set offsets[2] to the "bumpalong" value, that is, the
521 offset of the starting point of the matching process, provided the offsets
522 vector is large enough.
523
524 17. The \A escape now records a lookbehind value of 1, though its execution
525 does not actually inspect the previous character. This is to ensure that,
526 in partial multi-segment matching, at least one character from the old
527 segment is retained when a new segment is processed. Otherwise, if there
528 are no lookbehinds in the pattern, \A might match incorrectly at the start
529 of a new segment.
530
531 18. Added some #ifdef __VMS code into pcretest.c to help VMS implementations.
532
533 19. Redefined some pcre_uchar variables in pcre_exec.c as pcre_uint32; this
534 gives some modest performance improvement in 8-bit mode.
535
536 20. Added the PCRE-specific property \p{Xuc} for matching characters that can
537 be expressed in certain programming languages using Universal Character
538 Names.
539
540 21. Unicode validation has been updated in the light of Unicode Corrigendum #9,
541 which points out that "non characters" are not "characters that may not
542 appear in Unicode strings" but rather "characters that are reserved for
543 internal use and have only local meaning".
544
545 22. When a pattern was compiled with automatic callouts (PCRE_AUTO_CALLOUT) and
546 there was a conditional group that depended on an assertion, if the
547 assertion was false, the callout that immediately followed the alternation
548 in the condition was skipped when pcre_exec() was used for matching.
549
550 23. Allow an explicit callout to be inserted before an assertion that is the
551 condition for a conditional group, for compatibility with automatic
552 callouts, which always insert a callout at this point.
553
554 24. In 8.31, (*COMMIT) was confined to within a recursive subpattern. Perl also
555 confines (*SKIP) and (*PRUNE) in the same way, and this has now been done.
556
557 25. (*PRUNE) is now supported by the JIT compiler.
558
559 26. Fix infinite loop when /(?<=(*SKIP)ac)a/ is matched against aa.
560
561 27. Fix the case where there are two or more SKIPs with arguments that may be
562 ignored.
563
564 28. (*SKIP) is now supported by the JIT compiler.
565
566 29. (*THEN) is now supported by the JIT compiler.
567
568 30. Update RunTest with additional test selector options.
569
570 31. The way PCRE handles backtracking verbs has been changed in two ways.
571
572 (1) Previously, in something like (*COMMIT)(*SKIP), COMMIT would override
573 SKIP. Now, PCRE acts on whichever backtracking verb is reached first by
574 backtracking. In some cases this makes it more Perl-compatible, but Perl's
575 rather obscure rules do not always do the same thing.
576
577 (2) Previously, backtracking verbs were confined within assertions. This is
578 no longer the case for positive assertions, except for (*ACCEPT). Again,
579 this sometimes improves Perl compatibility, and sometimes does not.
580
581 32. A number of tests that were in test 2 because Perl did things differently
582 have been moved to test 1, because either Perl or PCRE has changed, and
583 these tests are now compatible.
584
585 32. Backtracking control verbs are now handled in the same way in JIT and
586 interpreter.
587
588 33. An opening parenthesis in a MARK/PRUNE/SKIP/THEN name in a pattern that
589 contained a forward subroutine reference caused a compile error.
590
591 34. Auto-detect and optimize limited repetitions in JIT.
592
593 35. Implement PCRE_NEVER_UTF to lock out the use of UTF, in particular,
594 blocking (*UTF) etc.
595
596 36. In the interpreter, maximizing pattern repetitions for characters and
597 character types now use tail recursion, which reduces stack usage.
598
599 37. The value of the max lookbehind was not correctly preserved if a compiled
600 and saved regex was reloaded on a host of different endianness.
601
602 38. Implemented (*LIMIT_MATCH) and (*LIMIT_RECURSION). As part of the extension
603 of the compiled pattern block, expand the flags field from 16 to 32 bits
604 because it was almost full.
605
606 39. Try madvise first before posix_madvise.
607
608 40. Change 7 for PCRE 7.9 made it impossible for pcregrep to find empty lines
609 with a pattern such as ^$. It has taken 4 years for anybody to notice! The
610 original change locked out all matches of empty strings. This has been
611 changed so that one match of an empty string per line is recognized.
612 Subsequent searches on the same line (for colouring or for --only-matching,
613 for example) do not recognize empty strings.
614
615 41. Applied a user patch to fix a number of spelling mistakes in comments.
616
617 42. Data lines longer than 65536 caused pcretest to crash.
618
619 43. Clarified the data type for length and startoffset arguments for pcre_exec
620 and pcre_dfa_exec in the function-specific man pages, where they were
621 explicitly stated to be in bytes, never having been updated. I also added
622 some clarification to the pcreapi man page.
623
624 44. A call to pcre_dfa_exec() with an output vector size less than 2 caused
625 a segmentation fault.
626
627
628 Version 8.32 30-November-2012
629 -----------------------------
630
631 1. Improved JIT compiler optimizations for first character search and single
632 character iterators.
633
634 2. Supporting IBM XL C compilers for PPC architectures in the JIT compiler.
635 Patch by Daniel Richard G.
636
637 3. Single character iterator optimizations in the JIT compiler.
638
639 4. Improved JIT compiler optimizations for character ranges.
640
641 5. Rename the "leave" variable names to "quit" to improve WinCE compatibility.
642 Reported by Giuseppe D'Angelo.
643
644 6. The PCRE_STARTLINE bit, indicating that a match can occur only at the start
645 of a line, was being set incorrectly in cases where .* appeared inside
646 atomic brackets at the start of a pattern, or where there was a subsequent
647 *PRUNE or *SKIP.
648
649 7. Improved instruction cache flush for POWER/PowerPC.
650 Patch by Daniel Richard G.
651
652 8. Fixed a number of issues in pcregrep, making it more compatible with GNU
653 grep:
654
655 (a) There is now no limit to the number of patterns to be matched.
656
657 (b) An error is given if a pattern is too long.
658
659 (c) Multiple uses of --exclude, --exclude-dir, --include, and --include-dir
660 are now supported.
661
662 (d) --exclude-from and --include-from (multiple use) have been added.
663
664 (e) Exclusions and inclusions now apply to all files and directories, not
665 just to those obtained from scanning a directory recursively.
666
667 (f) Multiple uses of -f and --file-list are now supported.
668
669 (g) In a Windows environment, the default for -d has been changed from
670 "read" (the GNU grep default) to "skip", because otherwise the presence
671 of a directory in the file list provokes an error.
672
673 (h) The documentation has been revised and clarified in places.
674
675 9. Improve the matching speed of capturing brackets.
676
677 10. Changed the meaning of \X so that it now matches a Unicode extended
678 grapheme cluster.
679
680 11. Patch by Daniel Richard G to the autoconf files to add a macro for sorting
681 out POSIX threads when JIT support is configured.
682
683 12. Added support for PCRE_STUDY_EXTRA_NEEDED.
684
685 13. In the POSIX wrapper regcomp() function, setting re_nsub field in the preg
686 structure could go wrong in environments where size_t is not the same size
687 as int.
688
689 14. Applied user-supplied patch to pcrecpp.cc to allow PCRE_NO_UTF8_CHECK to be
690 set.
691
692 15. The EBCDIC support had decayed; later updates to the code had included
693 explicit references to (e.g.) \x0a instead of CHAR_LF. There has been a
694 general tidy up of EBCDIC-related issues, and the documentation was also
695 not quite right. There is now a test that can be run on ASCII systems to
696 check some of the EBCDIC-related things (but is it not a full test).
697
698 16. The new PCRE_STUDY_EXTRA_NEEDED option is now used by pcregrep, resulting
699 in a small tidy to the code.
700
701 17. Fix JIT tests when UTF is disabled and both 8 and 16 bit mode are enabled.
702
703 18. If the --only-matching (-o) option in pcregrep is specified multiple
704 times, each one causes appropriate output. For example, -o1 -o2 outputs the
705 substrings matched by the 1st and 2nd capturing parentheses. A separating
706 string can be specified by --om-separator (default empty).
707
708 19. Improving the first n character searches.
709
710 20. Turn case lists for horizontal and vertical white space into macros so that
711 they are defined only once.
712
713 21. This set of changes together give more compatible Unicode case-folding
714 behaviour for characters that have more than one other case when UCP
715 support is available.
716
717 (a) The Unicode property table now has offsets into a new table of sets of
718 three or more characters that are case-equivalent. The MultiStage2.py
719 script that generates these tables (the pcre_ucd.c file) now scans
720 CaseFolding.txt instead of UnicodeData.txt for character case
721 information.
722
723 (b) The code for adding characters or ranges of characters to a character
724 class has been abstracted into a generalized function that also handles
725 case-independence. In UTF-mode with UCP support, this uses the new data
726 to handle characters with more than one other case.
727
728 (c) A bug that is fixed as a result of (b) is that codepoints less than 256
729 whose other case is greater than 256 are now correctly matched
730 caselessly. Previously, the high codepoint matched the low one, but not
731 vice versa.
732
733 (d) The processing of \h, \H, \v, and \ in character classes now makes use
734 of the new class addition function, using character lists defined as
735 macros alongside the case definitions of 20 above.
736
737 (e) Caseless back references now work with characters that have more than
738 one other case.
739
740 (f) General caseless matching of characters with more than one other case
741 is supported.
742
743 22. Unicode character properties were updated from Unicode 6.2.0
744
745 23. Improved CMake support under Windows. Patch by Daniel Richard G.
746
747 24. Add support for 32-bit character strings, and UTF-32
748
749 25. Major JIT compiler update (code refactoring and bugfixing).
750 Experimental Sparc 32 support is added.
751
752 26. Applied a modified version of Daniel Richard G's patch to create
753 pcre.h.generic and config.h.generic by "make" instead of in the
754 PrepareRelease script.
755
756 27. Added a definition for CHAR_NULL (helpful for the z/OS port), and use it in
757 pcre_compile.c when checking for a zero character.
758
759 28. Introducing a native interface for JIT. Through this interface, the compiled
760 machine code can be directly executed. The purpose of this interface is to
761 provide fast pattern matching, so several sanity checks are not performed.
762 However, feature tests are still performed. The new interface provides
763 1.4x speedup compared to the old one.
764
765 29. If pcre_exec() or pcre_dfa_exec() was called with a negative value for
766 the subject string length, the error given was PCRE_ERROR_BADOFFSET, which
767 was confusing. There is now a new error PCRE_ERROR_BADLENGTH for this case.
768
769 30. In 8-bit UTF-8 mode, pcretest failed to give an error for data codepoints
770 greater than 0x7fffffff (which cannot be represented in UTF-8, even under
771 the "old" RFC 2279). Instead, it ended up passing a negative length to
772 pcre_exec().
773
774 31. Add support for GCC's visibility feature to hide internal functions.
775
776 32. Running "pcretest -C pcre8" or "pcretest -C pcre16" gave a spurious error
777 "unknown -C option" after outputting 0 or 1.
778
779 33. There is now support for generating a code coverage report for the test
780 suite in environments where gcc is the compiler and lcov is installed. This
781 is mainly for the benefit of the developers.
782
783 34. If PCRE is built with --enable-valgrind, certain memory regions are marked
784 unaddressable using valgrind annotations, allowing valgrind to detect
785 invalid memory accesses. This is mainly for the benefit of the developers.
786
787 25. (*UTF) can now be used to start a pattern in any of the three libraries.
788
789 26. Give configure error if --enable-cpp but no C++ compiler found.
790
791
792 Version 8.31 06-July-2012
793 -------------------------
794
795 1. Fixing a wrong JIT test case and some compiler warnings.
796
797 2. Removed a bashism from the RunTest script.
798
799 3. Add a cast to pcre_exec.c to fix the warning "unary minus operator applied
800 to unsigned type, result still unsigned" that was given by an MS compiler
801 on encountering the code "-sizeof(xxx)".
802
803 4. Partial matching support is added to the JIT compiler.
804
805 5. Fixed several bugs concerned with partial matching of items that consist
806 of more than one character:
807
808 (a) /^(..)\1/ did not partially match "aba" because checking references was
809 done on an "all or nothing" basis. This also applied to repeated
810 references.
811
812 (b) \R did not give a hard partial match if \r was found at the end of the
813 subject.
814
815 (c) \X did not give a hard partial match after matching one or more
816 characters at the end of the subject.
817
818 (d) When newline was set to CRLF, a pattern such as /a$/ did not recognize
819 a partial match for the string "\r".
820
821 (e) When newline was set to CRLF, the metacharacter "." did not recognize
822 a partial match for a CR character at the end of the subject string.
823
824 6. If JIT is requested using /S++ or -s++ (instead of just /S+ or -s+) when
825 running pcretest, the text "(JIT)" added to the output whenever JIT is
826 actually used to run the match.
827
828 7. Individual JIT compile options can be set in pcretest by following -s+[+]
829 or /S+[+] with a digit between 1 and 7.
830
831 8. OP_NOT now supports any UTF character not just single-byte ones.
832
833 9. (*MARK) control verb is now supported by the JIT compiler.
834
835 10. The command "./RunTest list" lists the available tests without actually
836 running any of them. (Because I keep forgetting what they all are.)
837
838 11. Add PCRE_INFO_MAXLOOKBEHIND.
839
840 12. Applied a (slightly modified) user-supplied patch that improves performance
841 when the heap is used for recursion (compiled with --disable-stack-for-
842 recursion). Instead of malloc and free for each heap frame each time a
843 logical recursion happens, frames are retained on a chain and re-used where
844 possible. This sometimes gives as much as 30% improvement.
845
846 13. As documented, (*COMMIT) is now confined to within a recursive subpattern
847 call.
848
849 14. As documented, (*COMMIT) is now confined to within a positive assertion.
850
851 15. It is now possible to link pcretest with libedit as an alternative to
852 libreadline.
853
854 16. (*COMMIT) control verb is now supported by the JIT compiler.
855
856 17. The Unicode data tables have been updated to Unicode 6.1.0.
857
858 18. Added --file-list option to pcregrep.
859
860 19. Added binary file support to pcregrep, including the -a, --binary-files,
861 -I, and --text options.
862
863 20. The madvise function is renamed for posix_madvise for QNX compatibility
864 reasons. Fixed by Giuseppe D'Angelo.
865
866 21. Fixed a bug for backward assertions with REVERSE 0 in the JIT compiler.
867
868 22. Changed the option for creating symbolic links for 16-bit man pages from
869 -s to -sf so that re-installing does not cause issues.
870
871 23. Support PCRE_NO_START_OPTIMIZE in JIT as (*MARK) support requires it.
872
873 24. Fixed a very old bug in pcretest that caused errors with restarted DFA
874 matches in certain environments (the workspace was not being correctly
875 retained). Also added to pcre_dfa_exec() a simple plausibility check on
876 some of the workspace data at the beginning of a restart.
877
878 25. \s*\R was auto-possessifying the \s* when it should not, whereas \S*\R
879 was not doing so when it should - probably a typo introduced by SVN 528
880 (change 8.10/14).
881
882 26. When PCRE_UCP was not set, \w+\x{c4} was incorrectly auto-possessifying the
883 \w+ when the character tables indicated that \x{c4} was a word character.
884 There were several related cases, all because the tests for doing a table
885 lookup were testing for characters less than 127 instead of 255.
886
887 27. If a pattern contains capturing parentheses that are not used in a match,
888 their slots in the ovector are set to -1. For those that are higher than
889 any matched groups, this happens at the end of processing. In the case when
890 there were back references that the ovector was too small to contain
891 (causing temporary malloc'd memory to be used during matching), and the
892 highest capturing number was not used, memory off the end of the ovector
893 was incorrectly being set to -1. (It was using the size of the temporary
894 memory instead of the true size.)
895
896 28. To catch bugs like 27 using valgrind, when pcretest is asked to specify an
897 ovector size, it uses memory at the end of the block that it has got.
898
899 29. Check for an overlong MARK name and give an error at compile time. The
900 limit is 255 for the 8-bit library and 65535 for the 16-bit library.
901
902 30. JIT compiler update.
903
904 31. JIT is now supported on jailbroken iOS devices. Thanks for Ruiger
905 Rill for the patch.
906
907 32. Put spaces around SLJIT_PRINT_D in the JIT compiler. Required by CXX11.
908
909 33. Variable renamings in the PCRE-JIT compiler. No functionality change.
910
911 34. Fixed typos in pcregrep: in two places there was SUPPORT_LIBZ2 instead of
912 SUPPORT_LIBBZ2. This caused a build problem when bzip2 but not gzip (zlib)
913 was enabled.
914
915 35. Improve JIT code generation for greedy plus quantifier.
916
917 36. When /((?:a?)*)*c/ or /((?>a?)*)*c/ was matched against "aac", it set group
918 1 to "aa" instead of to an empty string. The bug affected repeated groups
919 that could potentially match an empty string.
920
921 37. Optimizing single character iterators in JIT.
922
923 38. Wide characters specified with \uxxxx in JavaScript mode are now subject to
924 the same checks as \x{...} characters in non-JavaScript mode. Specifically,
925 codepoints that are too big for the mode are faulted, and in a UTF mode,
926 disallowed codepoints are also faulted.
927
928 39. If PCRE was compiled with UTF support, in three places in the DFA
929 matcher there was code that should only have been obeyed in UTF mode, but
930 was being obeyed unconditionally. In 8-bit mode this could cause incorrect
931 processing when bytes with values greater than 127 were present. In 16-bit
932 mode the bug would be provoked by values in the range 0xfc00 to 0xdc00. In
933 both cases the values are those that cannot be the first data item in a UTF
934 character. The three items that might have provoked this were recursions,
935 possessively repeated groups, and atomic groups.
936
937 40. Ensure that libpcre is explicitly listed in the link commands for pcretest
938 and pcregrep, because some OS require shared objects to be explicitly
939 passed to ld, causing the link step to fail if they are not.
940
941 41. There were two incorrect #ifdefs in pcre_study.c, meaning that, in 16-bit
942 mode, patterns that started with \h* or \R* might be incorrectly matched.
943
944
945 Version 8.30 04-February-2012
946 -----------------------------
947
948 1. Renamed "isnumber" as "is_a_number" because in some Mac environments this
949 name is defined in ctype.h.
950
951 2. Fixed a bug in fixed-length calculation for lookbehinds that would show up
952 only in quite long subpatterns.
953
954 3. Removed the function pcre_info(), which has been obsolete and deprecated
955 since it was replaced by pcre_fullinfo() in February 2000.
956
957 4. For a non-anchored pattern, if (*SKIP) was given with a name that did not
958 match a (*MARK), and the match failed at the start of the subject, a
959 reference to memory before the start of the subject could occur. This bug
960 was introduced by fix 17 of release 8.21.
961
962 5. A reference to an unset group with zero minimum repetition was giving
963 totally wrong answers (in non-JavaScript-compatibility mode). For example,
964 /(another)?(\1?)test/ matched against "hello world test". This bug was
965 introduced in release 8.13.
966
967 6. Add support for 16-bit character strings (a large amount of work involving
968 many changes and refactorings).
969
970 7. RunGrepTest failed on msys because \r\n was replaced by whitespace when the
971 command "pattern=`printf 'xxx\r\njkl'`" was run. The pattern is now taken
972 from a file.
973
974 8. Ovector size of 2 is also supported by JIT based pcre_exec (the ovector size
975 rounding is not applied in this particular case).
976
977 9. The invalid Unicode surrogate codepoints U+D800 to U+DFFF are now rejected
978 if they appear, or are escaped, in patterns.
979
980 10. Get rid of a number of -Wunused-but-set-variable warnings.
981
982 11. The pattern /(?=(*:x))(q|)/ matches an empty string, and returns the mark
983 "x". The similar pattern /(?=(*:x))((*:y)q|)/ did not return a mark at all.
984 Oddly, Perl behaves the same way. PCRE has been fixed so that this pattern
985 also returns the mark "x". This bug applied to capturing parentheses,
986 non-capturing parentheses, and atomic parentheses. It also applied to some
987 assertions.
988
989 12. Stephen Kelly's patch to CMakeLists.txt allows it to parse the version
990 information out of configure.ac instead of relying on pcre.h.generic, which
991 is not stored in the repository.
992
993 13. Applied Dmitry V. Levin's patch for a more portable method for linking with
994 -lreadline.
995
996 14. ZH added PCRE_CONFIG_JITTARGET; added its output to pcretest -C.
997
998 15. Applied Graycode's patch to put the top-level frame on the stack rather
999 than the heap when not using the stack for recursion. This gives a
1000 performance improvement in many cases when recursion is not deep.
1001
1002 16. Experimental code added to "pcretest -C" to output the stack frame size.
1003
1004
1005 Version 8.21 12-Dec-2011
1006 ------------------------
1007
1008 1. Updating the JIT compiler.
1009
1010 2. JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
1011 are added as well.
1012
1013 3. Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
1014 PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
1015 calling _pcre_jit_exec. Some extra comments are added.
1016
1017 4. (*MARK) settings inside atomic groups that do not contain any capturing
1018 parentheses, for example, (?>a(*:m)), were not being passed out. This bug
1019 was introduced by change 18 for 8.20.
1020
1021 5. Supporting of \x, \U and \u in JavaScript compatibility mode based on the
1022 ECMA-262 standard.
1023
1024 6. Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
1025 erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
1026 This bug was probably introduced by change 9 of 8.13.
1027
1028 7. While fixing 6 above, I noticed that a number of other items were being
1029 incorrectly rejected as "not fixed length". This arose partly because newer
1030 opcodes had not been added to the fixed-length checking code. I have (a)
1031 corrected the bug and added tests for these items, and (b) arranged for an
1032 error to occur if an unknown opcode is encountered while checking for fixed
1033 length instead of just assuming "not fixed length". The items that were
1034 rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
1035 (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
1036 repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
1037
1038 8. A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
1039 being incorrectly compiled and would have given unpredicatble results.
1040
1041 9. A possessively repeated subpattern with minimum repeat count greater than
1042 one behaved incorrectly. For example, (A){2,}+ behaved as if it was
1043 (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
1044 the first (A) could occur when it should not.
1045
1046 10. Add a cast and remove a redundant test from the code.
1047
1048 11. JIT should use pcre_malloc/pcre_free for allocation.
1049
1050 12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
1051 best practice nowadays, and helps with cross-compiling. (If the exec_prefix
1052 is anything other than /usr, -L is still shown).
1053
1054 13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
1055
1056 14. Perl does not support \N without a following name in a [] class; PCRE now
1057 also gives an error.
1058
1059 15. If a forward reference was repeated with an upper limit of around 2000,
1060 it caused the error "internal error: overran compiling workspace". The
1061 maximum number of forward references (including repeats) was limited by the
1062 internal workspace, and dependent on the LINK_SIZE. The code has been
1063 rewritten so that the workspace expands (via pcre_malloc) if necessary, and
1064 the default depends on LINK_SIZE. There is a new upper limit (for safety)
1065 of around 200,000 forward references. While doing this, I also speeded up
1066 the filling in of repeated forward references.
1067
1068 16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
1069 incorrectly expecting the subject to contain another "a" after the start.
1070
1071 17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
1072 in the match, the SKIP should be ignored. This was not happening; instead
1073 the SKIP was being treated as NOMATCH. For patterns such as
1074 /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
1075 tested.
1076
1077 18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
1078 now much more compatible with Perl, in particular in cases where the result
1079 is a non-match for a non-anchored pattern. For example, if
1080 /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
1081 "m", where previously it did not return a name. A side effect of this
1082 change is that for partial matches, the last encountered mark name is
1083 returned, as for non matches. A number of tests that were previously not
1084 Perl-compatible have been moved into the Perl-compatible test files. The
1085 refactoring has had the pleasing side effect of removing one argument from
1086 the match() function, thus reducing its stack requirements.
1087
1088 19. If the /S+ option was used in pcretest to study a pattern using JIT,
1089 subsequent uses of /S (without +) incorrectly behaved like /S+.
1090
1091 21. Retrieve executable code size support for the JIT compiler and fixing
1092 some warnings.
1093
1094 22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
1095 not work when the shorter character appeared right at the end of the
1096 subject string.
1097
1098 23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
1099 systems.
1100
1101 24. Added PCRE_INFO_JITSIZE to pass on the value from (21) above, and also
1102 output it when the /M option is used in pcretest.
1103
1104 25. The CheckMan script was not being included in the distribution. Also, added
1105 an explicit "perl" to run Perl scripts from the PrepareRelease script
1106 because this is reportedly needed in Windows.
1107
1108 26. If study data was being save in a file and studying had not found a set of
1109 "starts with" bytes for the pattern, the data written to the file (though
1110 never used) was taken from uninitialized memory and so caused valgrind to
1111 complain.
1112
1113 27. Updated RunTest.bat as provided by Sheri Pierce.
1114
1115 28. Fixed a possible uninitialized memory bug in pcre_jit_compile.c.
1116
1117 29. Computation of memory usage for the table of capturing group names was
1118 giving an unnecessarily large value.
1119
1120
1121 Version 8.20 21-Oct-2011
1122 ------------------------
1123
1124 1. Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
1125 a POSIX class. After further experiments with Perl, which convinced me that
1126 Perl has bugs and confusions, a closing square bracket is no longer allowed
1127 in a POSIX name. This bug also affected patterns with classes that started
1128 with full stops.
1129
1130 2. If a pattern such as /(a)b|ac/ is matched against "ac", there is no
1131 captured substring, but while checking the failing first alternative,
1132 substring 1 is temporarily captured. If the output vector supplied to
1133 pcre_exec() was not big enough for this capture, the yield of the function
1134 was still zero ("insufficient space for captured substrings"). This cannot
1135 be totally fixed without adding another stack variable, which seems a lot
1136 of expense for a edge case. However, I have improved the situation in cases
1137 such as /(a)(b)x|abc/ matched against "abc", where the return code
1138 indicates that fewer than the maximum number of slots in the ovector have
1139 been set.
1140
1141 3. Related to (2) above: when there are more back references in a pattern than
1142 slots in the output vector, pcre_exec() uses temporary memory during
1143 matching, and copies in the captures as far as possible afterwards. It was
1144 using the entire output vector, but this conflicts with the specification
1145 that only 2/3 is used for passing back captured substrings. Now it uses
1146 only the first 2/3, for compatibility. This is, of course, another edge
1147 case.
1148
1149 4. Zoltan Herczeg's just-in-time compiler support has been integrated into the
1150 main code base, and can be used by building with --enable-jit. When this is
1151 done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
1152 runtime --no-jit option is given.
1153
1154 5. When the number of matches in a pcre_dfa_exec() run exactly filled the
1155 ovector, the return from the function was zero, implying that there were
1156 other matches that did not fit. The correct "exactly full" value is now
1157 returned.
1158
1159 6. If a subpattern that was called recursively or as a subroutine contained
1160 (*PRUNE) or any other control that caused it to give a non-standard return,
1161 invalid errors such as "Error -26 (nested recursion at the same subject
1162 position)" or even infinite loops could occur.
1163
1164 7. If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
1165 computing the minimum length on reaching *ACCEPT, and so ended up with the
1166 wrong value of 1 rather than 0. Further investigation indicates that
1167 computing a minimum subject length in the presence of *ACCEPT is difficult
1168 (think back references, subroutine calls), and so I have changed the code
1169 so that no minimum is registered for a pattern that contains *ACCEPT.
1170
1171 8. If (*THEN) was present in the first (true) branch of a conditional group,
1172 it was not handled as intended. [But see 16 below.]
1173
1174 9. Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
1175 Sheri Pierce.
1176
1177 10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
1178 the first byte in a match must be "a".
1179
1180 11. Change 17 for 8.13 increased the recursion depth for patterns like
1181 /a(?:.)*?a/ drastically. I've improved things by remembering whether a
1182 pattern contains any instances of (*THEN). If it does not, the old
1183 optimizations are restored. It would be nice to do this on a per-group
1184 basis, but at the moment that is not feasible.
1185
1186 12. In some environments, the output of pcretest -C is CRLF terminated. This
1187 broke RunTest's code that checks for the link size. A single white space
1188 character after the value is now allowed for.
1189
1190 13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
1191 For "fr", it uses the Windows-specific input and output files.
1192
1193 14. If (*THEN) appeared in a group that was called recursively or as a
1194 subroutine, it did not work as intended. [But see next item.]
1195
1196 15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
1197 pattern fragments (but not containing any | characters). If A and B are
1198 matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
1199 was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
1200 D. In other words, Perl considers parentheses that do not contain any |
1201 characters to be part of a surrounding alternative, whereas PCRE was
1202 treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
1203 differently. PCRE now behaves in the same way as Perl, except in the case
1204 of subroutine/recursion calls such as (?1) which have in any case always
1205 been different (but PCRE had them first :-).
1206
1207 16. Related to 15 above: Perl does not treat the | in a conditional group as
1208 creating alternatives. Such a group is treated in the same way as an
1209 ordinary group without any | characters when processing (*THEN). PCRE has
1210 been changed to match Perl's behaviour.
1211
1212 17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
1213 RunGrepTest script failed.
1214
1215 18. Change 22 for version 13 caused atomic groups to use more stack. This is
1216 inevitable for groups that contain captures, but it can lead to a lot of
1217 stack use in large patterns. The old behaviour has been restored for atomic
1218 groups that do not contain any capturing parentheses.
1219
1220 19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
1221 suppress the check for a minimum subject length at run time. (If it was
1222 given to pcre_exec() or pcre_dfa_exec() it did work.)
1223
1224 20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
1225 fail to work when decoding hex characters in data strings in EBCDIC
1226 environments.
1227
1228 21. It appears that in at least one Mac OS environment, the isxdigit() function
1229 is implemented as a macro that evaluates to its argument more than once,
1230 contravening the C 90 Standard (I haven't checked a later standard). There
1231 was an instance in pcretest which caused it to go wrong when processing
1232 \x{...} escapes in subject strings. The has been rewritten to avoid using
1233 things like p++ in the argument of isxdigit().
1234
1235
1236 Version 8.13 16-Aug-2011
1237 ------------------------
1238
1239 1. The Unicode data tables have been updated to Unicode 6.0.0.
1240
1241 2. Two minor typos in pcre_internal.h have been fixed.
1242
1243 3. Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
1244 pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
1245 in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
1246
1247 4. There were a number of related bugs in the code for matching backrefences
1248 caselessly in UTF-8 mode when codes for the characters concerned were
1249 different numbers of bytes. For example, U+023A and U+2C65 are an upper
1250 and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
1251 (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
1252 code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
1253 2-byte code at the end of the subject (it thought there wasn't enough data
1254 left).
1255
1256 5. Comprehensive information about what went wrong is now returned by
1257 pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
1258 as the output vector has at least 2 elements. The offset of the start of
1259 the failing character and a reason code are placed in the vector.
1260
1261 6. When the UTF-8 string check fails for pcre_compile(), the offset that is
1262 now returned is for the first byte of the failing character, instead of the
1263 last byte inspected. This is an incompatible change, but I hope it is small
1264 enough not to be a problem. It makes the returned offset consistent with
1265 pcre_exec() and pcre_dfa_exec().
1266
1267 7. pcretest now gives a text phrase as well as the error number when
1268 pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
1269 failure, the offset and reason code are output.
1270
1271 8. When \R was used with a maximizing quantifier it failed to skip backwards
1272 over a \r\n pair if the subsequent match failed. Instead, it just skipped
1273 back over a single character (\n). This seems wrong (because it treated the
1274 two characters as a single entity when going forwards), conflicts with the
1275 documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
1276 behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
1277 has been changed.
1278
1279 9. Some internal refactoring has changed the processing so that the handling
1280 of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
1281 time (the PCRE_DOTALL option was changed this way some time ago: version
1282 7.7 change 16). This has made it possible to abolish the OP_OPT op code,
1283 which was always a bit of a fudge. It also means that there is one less
1284 argument for the match() function, which reduces its stack requirements
1285 slightly. This change also fixes an incompatibility with Perl: the pattern
1286 (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
1287
1288 10. More internal refactoring has drastically reduced the number of recursive
1289 calls to match() for possessively repeated groups such as (abc)++ when
1290 using pcre_exec().
1291
1292 11. While implementing 10, a number of bugs in the handling of groups were
1293 discovered and fixed:
1294
1295 (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
1296 (a|)*(?1) gave a compile-time internal error.
1297 ((a|)+)+ did not notice that the outer group could match an empty string.
1298 (^a|^)+ was not marked as anchored.
1299 (.*a|.*)+ was not marked as matching at start or after a newline.
1300
1301 12. Yet more internal refactoring has removed another argument from the match()
1302 function. Special calls to this function are now indicated by setting a
1303 value in a variable in the "match data" data block.
1304
1305 13. Be more explicit in pcre_study() instead of relying on "default" for
1306 opcodes that mean there is no starting character; this means that when new
1307 ones are added and accidentally left out of pcre_study(), testing should
1308 pick them up.
1309
1310 14. The -s option of pcretest has been documented for ages as being an old
1311 synonym of -m (show memory usage). I have changed it to mean "force study
1312 for every regex", that is, assume /S for every regex. This is similar to -i
1313 and -d etc. It's slightly incompatible, but I'm hoping nobody is still
1314 using it. It makes it easier to run collections of tests with and without
1315 study enabled, and thereby test pcre_study() more easily. All the standard
1316 tests are now run with and without -s (but some patterns can be marked as
1317 "never study" - see 20 below).
1318
1319 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
1320 restoration of the capturing data to the outer values was not happening
1321 correctly.
1322
1323 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
1324 empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
1325 pattern had matched an empty string, and so incorrectly returned a no
1326 match.
1327
1328 17. There was optimizing code for the last branch of non-capturing parentheses,
1329 and also for the obeyed branch of a conditional subexpression, which used
1330 tail recursion to cut down on stack usage. Unfortunately, now that there is
1331 the possibility of (*THEN) occurring in these branches, tail recursion is
1332 no longer possible because the return has to be checked for (*THEN). These
1333 two optimizations have therefore been removed. [But see 8.20/11 above.]
1334
1335 18. If a pattern containing \R was studied, it was assumed that \R always
1336 matched two bytes, thus causing the minimum subject length to be
1337 incorrectly computed because \R can also match just one byte.
1338
1339 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
1340 was incorrectly computed.
1341
1342 20. If /S is present twice on a test pattern in pcretest input, it now
1343 *disables* studying, thereby overriding the use of -s on the command line
1344 (see 14 above). This is necessary for one or two tests to keep the output
1345 identical in both cases.
1346
1347 21. When (*ACCEPT) was used in an assertion that matched an empty string and
1348 PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
1349
1350 22. When an atomic group that contained a capturing parenthesis was
1351 successfully matched, but the branch in which it appeared failed, the
1352 capturing was not being forgotten if a higher numbered group was later
1353 captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
1354 group 1 to "a", when in fact it should be unset. This applied to multi-
1355 branched capturing and non-capturing groups, repeated or not, and also to
1356 positive assertions (capturing in negative assertions does not happen
1357 in PCRE) and also to nested atomic groups.
1358
1359 23. Add the ++ qualifier feature to pcretest, to show the remainder of the
1360 subject after a captured substring, to make it easier to tell which of a
1361 number of identical substrings has been captured.
1362
1363 24. The way atomic groups are processed by pcre_exec() has been changed so that
1364 if they are repeated, backtracking one repetition now resets captured
1365 values correctly. For example, if ((?>(a+)b)+aabab) is matched against
1366 "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
1367 "aaa". Previously, it would have been "a". As part of this code
1368 refactoring, the way recursive calls are handled has also been changed.
1369
1370 25. If an assertion condition captured any substrings, they were not passed
1371 back unless some other capturing happened later. For example, if
1372 (?(?=(a))a) was matched against "a", no capturing was returned.
1373
1374 26. When studying a pattern that contained subroutine calls or assertions,
1375 the code for finding the minimum length of a possible match was handling
1376 direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
1377 group 1 called group 2 while simultaneously a separate group 2 called group
1378 1). A stack overflow occurred in this case. I have fixed this by limiting
1379 the recursion depth to 10.
1380
1381 27. Updated RunTest.bat in the distribution to the version supplied by Tom
1382 Fortmann. This supports explicit test numbers on the command line, and has
1383 argument validation and error reporting.
1384
1385 28. An instance of \X with an unlimited repeat could fail if at any point the
1386 first character it looked at was a mark character.
1387
1388 29. Some minor code refactoring concerning Unicode properties and scripts
1389 should reduce the stack requirement of match() slightly.
1390
1391 30. Added the '=' option to pcretest to check the setting of unused capturing
1392 slots at the end of the pattern, which are documented as being -1, but are
1393 not included in the return count.
1394
1395 31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
1396 compiled something random. Now it gives a compile-time error (as does
1397 Perl).
1398
1399 32. A *MARK encountered during the processing of a positive assertion is now
1400 recorded and passed back (compatible with Perl).
1401
1402 33. If --only-matching or --colour was set on a pcregrep call whose pattern
1403 had alternative anchored branches, the search for a second match in a line
1404 was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
1405 matched the line "0102" twice. The same bug affected patterns that started
1406 with a backwards assertion. For example /\b01|\b02/ also matched "0102"
1407 twice.
1408
1409 34. Previously, PCRE did not allow quantification of assertions. However, Perl
1410 does, and because of capturing effects, quantifying parenthesized
1411 assertions may at times be useful. Quantifiers are now allowed for
1412 parenthesized assertions.
1413
1414 35. A minor code tidy in pcre_compile() when checking options for \R usage.
1415
1416 36. \g was being checked for fancy things in a character class, when it should
1417 just be a literal "g".
1418
1419 37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
1420 appearance of a nested POSIX class supersedes an apparent external class.
1421 For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
1422 unescaped square brackets may also appear as part of class names. For
1423 example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
1424 more like Perl. (But see 8.20/1 above.)
1425
1426 38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
1427 was because it thought it was \N{name}, which is not supported).
1428
1429 39. Add minix to OS list not supporting the -S option in pcretest.
1430
1431 40. PCRE tries to detect cases of infinite recursion at compile time, but it
1432 cannot analyze patterns in sufficient detail to catch mutual recursions
1433 such as ((?1))((?2)). There is now a runtime test that gives an error if a
1434 subgroup is called recursively as a subpattern for a second time at the
1435 same position in the subject string. In previous releases this might have
1436 been caught by the recursion limit, or it might have run out of stack.
1437
1438 41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
1439 happen only once. PCRE was, however incorrectly giving a compile time error
1440 "recursive call could loop indefinitely" because it cannot analyze the
1441 pattern in sufficient detail. The compile time test no longer happens when
1442 PCRE is compiling a conditional subpattern, but actual runaway loops are
1443 now caught at runtime (see 40 above).
1444
1445 42. It seems that Perl allows any characters other than a closing parenthesis
1446 to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
1447 has been changed to be the same.
1448
1449 43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
1450 as not to get warnings when autogen.sh is called. Also changed
1451 AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
1452
1453 44. To help people who use pcregrep to scan files containing exceedingly long
1454 lines, the following changes have been made:
1455
1456 (a) The default value of the buffer size parameter has been increased from
1457 8K to 20K. (The actual buffer used is three times this size.)
1458
1459 (b) The default can be changed by ./configure --with-pcregrep-bufsize when
1460 PCRE is built.
1461
1462 (c) A --buffer-size=n option has been added to pcregrep, to allow the size
1463 to be set at run time.
1464
1465 (d) Numerical values in pcregrep options can be followed by K or M, for
1466 example --buffer-size=50K.
1467
1468 (e) If a line being scanned overflows pcregrep's buffer, an error is now
1469 given and the return code is set to 2.
1470
1471 45. Add a pointer to the latest mark to the callout data block.
1472
1473 46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
1474 partial match of an empty string instead of no match. This was specific to
1475 the use of ".".
1476
1477 47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
1478 complete match instead of a partial match. This bug was dependent on both
1479 the PCRE_UTF8 and PCRE_DOTALL options being set.
1480
1481 48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
1482 starting byte set, because \b was not being ignored.
1483
1484
1485 Version 8.12 15-Jan-2011
1486 ------------------------
1487
1488 1. Fixed some typos in the markup of the man pages, and wrote a script that
1489 checks for such things as part of the documentation building process.
1490
1491 2. On a big-endian 64-bit system, pcregrep did not correctly process the
1492 --match-limit and --recursion-limit options (added for 8.11). In
1493 particular, this made one of the standard tests fail. (The integer value
1494 went into the wrong half of a long int.)
1495
1496 3. If the --colour option was given to pcregrep with -v (invert match), it
1497 did strange things, either producing crazy output, or crashing. It should,
1498 of course, ignore a request for colour when reporting lines that do not
1499 match.
1500
1501 4. Another pcregrep bug caused similar problems if --colour was specified with
1502 -M (multiline) and the pattern match finished with a line ending.
1503
1504 5. In pcregrep, when a pattern that ended with a literal newline sequence was
1505 matched in multiline mode, the following line was shown as part of the
1506 match. This seems wrong, so I have changed it.
1507
1508 6. Another pcregrep bug in multiline mode, when --colour was specified, caused
1509 the check for further matches in the same line (so they could be coloured)
1510 to overrun the end of the current line. If another match was found, it was
1511 incorrectly shown (and then shown again when found in the next line).
1512
1513 7. If pcregrep was compiled under Windows, there was a reference to the
1514 function pcregrep_exit() before it was defined. I am assuming this was
1515 the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
1516 reported by a user. I've moved the definition above the reference.
1517
1518
1519 Version 8.11 10-Dec-2010
1520 ------------------------
1521
1522 1. (*THEN) was not working properly if there were untried alternatives prior
1523 to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
1524 backtracked to try for "b" instead of moving to the next alternative branch
1525 at the same level (in this case, to look for "c"). The Perl documentation
1526 is clear that when (*THEN) is backtracked onto, it goes to the "next
1527 alternative in the innermost enclosing group".
1528
1529 2. (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
1530 such as (A(*COMMIT)B(*THEN)C|D) any failure after matching A should
1531 result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
1532 (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
1533 (*THEN).
1534
1535 3. If \s appeared in a character class, it removed the VT character from
1536 the class, even if it had been included by some previous item, for example
1537 in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
1538 of \s, but is part of the POSIX "space" class.)
1539
1540 4. A partial match never returns an empty string (because you can always
1541 match an empty string at the end of the subject); however the checking for
1542 an empty string was starting at the "start of match" point. This has been
1543 changed to the "earliest inspected character" point, because the returned
1544 data for a partial match starts at this character. This means that, for
1545 example, /(?<=abc)def/ gives a partial match for the subject "abc"
1546 (previously it gave "no match").
1547
1548 5. Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
1549 of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
1550 previously a full match would be given. However, setting PCRE_PARTIAL_HARD
1551 has an implication that the given string is incomplete (because a partial
1552 match is preferred over a full match). For this reason, these items now
1553 give a partial match in this situation. [Aside: previously, the one case
1554 /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
1555 match rather than a full match, which was wrong by the old rules, but is
1556 now correct.]
1557
1558 6. There was a bug in the handling of #-introduced comments, recognized when
1559 PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
1560 If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
1561 UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
1562 scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
1563 but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
1564 places in pcre_compile().
1565
1566 7. Related to (6) above, when pcre_compile() was skipping #-introduced
1567 comments when looking ahead for named forward references to subpatterns,
1568 the only newline sequence it recognized was NL. It now handles newlines
1569 according to the set newline convention.
1570
1571 8. SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
1572 former, but used strtoul(), whereas pcretest avoided strtoul() but did not
1573 cater for a lack of strerror(). These oversights have been fixed.
1574
1575 9. Added --match-limit and --recursion-limit to pcregrep.
1576
1577 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
1578
1579 11. When the -o option was used, pcregrep was setting a return code of 1, even
1580 when matches were found, and --line-buffered was not being honoured.
1581
1582 12. Added an optional parentheses number to the -o and --only-matching options
1583 of pcregrep.
1584
1585 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
1586 can match an empty string. The code to do it in pcretest and pcredemo
1587 needed fixing:
1588
1589 (a) When the newline convention was "crlf", pcretest got it wrong, skipping
1590 only one byte after an empty string match just before CRLF (this case
1591 just got forgotten; "any" and "anycrlf" were OK).
1592
1593 (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
1594 mode when an empty string match preceded an ASCII character followed by
1595 a non-ASCII character. (The code for advancing by one character rather
1596 than one byte was nonsense.)
1597
1598 (c) The pcredemo.c sample program did not have any code at all to handle
1599 the cases when CRLF is a valid newline sequence.
1600
1601 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
1602 as a starting offset was within the subject string. There is now a new
1603 error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
1604 negative or greater than the length of the string. In order to test this,
1605 pcretest is extended to allow the setting of negative starting offsets.
1606
1607 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
1608 starting offset points to the beginning of a UTF-8 character was
1609 unnecessarily clumsy. I tidied it up.
1610
1611 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
1612 bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
1613
1614 17. Nobody had reported that the --include_dir option, which was added in
1615 release 7.7 should have been called --include-dir (hyphen, not underscore)
1616 for compatibility with GNU grep. I have changed it to --include-dir, but
1617 left --include_dir as an undocumented synonym, and the same for
1618 --exclude-dir, though that is not available in GNU grep, at least as of
1619 release 2.5.4.
1620
1621 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
1622 characters from a string of bytes) have been redefined so as not to use
1623 loops, in order to improve performance in some environments. At the same
1624 time, I abstracted some of the common code into auxiliary macros to save
1625 repetition (this should not affect the compiled code).
1626
1627 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
1628 compile-time error is now given if \c is not followed by an ASCII
1629 character, that is, a byte less than 128. (In EBCDIC mode, the code is
1630 different, and any byte value is allowed.)
1631
1632 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
1633 START_OPTIMIZE option, which is now allowed at compile time - but just
1634 passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
1635 to pcregrep and other applications that have no direct access to PCRE
1636 options. The new /Y option in pcretest sets this option when calling
1637 pcre_compile().
1638
1639 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
1640 back references. Groups containing recursive back references were forced to
1641 be atomic by that change, but in the case of named groups, the amount of
1642 memory required was incorrectly computed, leading to "Failed: internal
1643 error: code overflow". This has been fixed.
1644
1645 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
1646 pcretest.c, to avoid build problems in some Borland environments.
1647
1648
1649 Version 8.10 25-Jun-2010
1650 ------------------------
1651
1652 1. Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
1653 THEN.
1654
1655 2. (*ACCEPT) was not working when inside an atomic group.
1656
1657 3. Inside a character class, \B is treated as a literal by default, but
1658 faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
1659 causes the error). The code is unchanged, but I tidied the documentation.
1660
1661 4. Inside a character class, PCRE always treated \R and \X as literals,
1662 whereas Perl faults them if its -w option is set. I have changed PCRE so
1663 that it faults them when PCRE_EXTRA is set.
1664
1665 5. Added support for \N, which always matches any character other than
1666 newline. (It is the same as "." when PCRE_DOTALL is not set.)
1667
1668 6. When compiling pcregrep with newer versions of gcc which may have
1669 FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
1670 declared with attribute warn_unused_result" were given. Just casting the
1671 result to (void) does not stop the warnings; a more elaborate fudge is
1672 needed. I've used a macro to implement this.
1673
1674 7. Minor change to pcretest.c to avoid a compiler warning.
1675
1676 8. Added four artifical Unicode properties to help with an option to make
1677 \s etc use properties (see next item). The new properties are: Xan
1678 (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
1679
1680 9. Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
1681 use Unicode properties. (*UCP) at the start of a pattern can be used to set
1682 this option. Modified pcretest to add /W to test this facility. Added
1683 REG_UCP to make it available via the POSIX interface.
1684
1685 10. Added --line-buffered to pcregrep.
1686
1687 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
1688 studied, and the match started with a letter with a code point greater than
1689 127 whose first byte was different to the first byte of the other case of
1690 the letter, the other case of this starting letter was not recognized
1691 (#976).
1692
1693 12. If a pattern that was studied started with a repeated Unicode property
1694 test, for example, \p{Nd}+, there was the theoretical possibility of
1695 setting up an incorrect bitmap of starting bytes, but fortunately it could
1696 not have actually happened in practice until change 8 above was made (it
1697 added property types that matched character-matching opcodes).
1698
1699 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
1700 possible starting bytes for non-anchored patterns.
1701
1702 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
1703 \R, and also a number of cases that involve Unicode properties, both
1704 explicit and implicit when PCRE_UCP is set.
1705
1706 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
1707 input, it could crash or give wrong results if characters with values
1708 greater than 0xc0 were present in the subject string. (Detail: it assumed
1709 UTF-8 input when processing these items.)
1710
1711 16. Added a lot of (int) casts to avoid compiler warnings in systems where
1712 size_t is 64-bit (#991).
1713
1714 17. Added a check for running out of memory when PCRE is compiled with
1715 --disable-stack-for-recursion (#990).
1716
1717 18. If the last data line in a file for pcretest does not have a newline on
1718 the end, a newline was missing in the output.
1719
1720 19. The default pcre_chartables.c file recognizes only ASCII characters (values
1721 less than 128) in its various bitmaps. However, there is a facility for
1722 generating tables according to the current locale when PCRE is compiled. It
1723 turns out that in some environments, 0x85 and 0xa0, which are Unicode space
1724 characters, are recognized by isspace() and therefore were getting set in
1725 these tables, and indeed these tables seem to approximate to ISO 8859. This
1726 caused a problem in UTF-8 mode when pcre_study() was used to create a list
1727 of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
1728 which of course cannot start UTF-8 characters. I have changed the code so
1729 that only real ASCII characters (less than 128) and the correct starting
1730 bytes for UTF-8 encodings are set for characters greater than 127 when in
1731 UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
1732 altogether.)
1733
1734 20. Added the /T option to pcretest so as to be able to run tests with non-
1735 standard character tables, thus making it possible to include the tests
1736 used for 19 above in the standard set of tests.
1737
1738 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
1739 reference to a subpattern the other side of a comment that contains an
1740 opening parenthesis caused either an internal compiling error, or a
1741 reference to the wrong subpattern.
1742
1743
1744 Version 8.02 19-Mar-2010
1745 ------------------------
1746
1747 1. The Unicode data tables have been updated to Unicode 5.2.0.
1748
1749 2. Added the option --libs-cpp to pcre-config, but only when C++ support is
1750 configured.
1751
1752 3. Updated the licensing terms in the pcregexp.pas file, as agreed with the
1753 original author of that file, following a query about its status.
1754
1755 4. On systems that do not have stdint.h (e.g. Solaris), check for and include
1756 inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
1757
1758 5. A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
1759 quantifier applied to a forward-referencing subroutine call, could compile
1760 incorrect code or give the error "internal error: previously-checked
1761 referenced subpattern not found".
1762
1763 6. Both MS Visual Studio and Symbian OS have problems with initializing
1764 variables to point to external functions. For these systems, therefore,
1765 pcre_malloc etc. are now initialized to local functions that call the
1766 relevant global functions.
1767
1768 7. There were two entries missing in the vectors called coptable and poptable
1769 in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
1770 I've fixed the data, and added a kludgy way of testing at compile time that
1771 the lengths are correct (equal to the number of opcodes).
1772
1773 8. Following on from 7, I added a similar kludge to check the length of the
1774 eint vector in pcreposix.c.
1775
1776 9. Error texts for pcre_compile() are held as one long string to avoid too
1777 much relocation at load time. To find a text, the string is searched,
1778 counting zeros. There was no check for running off the end of the string,
1779 which could happen if a new error number was added without updating the
1780 string.
1781
1782 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
1783
1784 11. \K was not working if it appeared in an atomic group or in a group that
1785 was called as a "subroutine", or in an assertion. Perl 5.11 documents that
1786 \K is "not well defined" if used in an assertion. PCRE now accepts it if
1787 the assertion is positive, but not if it is negative.
1788
1789 12. Change 11 fortuitously reduced the size of the stack frame used in the
1790 "match()" function of pcre_exec.c by one pointer. Forthcoming
1791 implementation of support for (*MARK) will need an extra pointer on the
1792 stack; I have reserved it now, so that the stack frame size does not
1793 decrease.
1794
1795 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
1796 item in branch that calls a recursion is a subroutine call - as in the
1797 second branch in the above example - was incorrectly given the compile-
1798 time error "recursive call could loop indefinitely" because pcre_compile()
1799 was not correctly checking the subroutine for matching a non-empty string.
1800
1801 14. The checks for overrunning compiling workspace could trigger after an
1802 overrun had occurred. This is a "should never occur" error, but it can be
1803 triggered by pathological patterns such as hundreds of nested parentheses.
1804 The checks now trigger 100 bytes before the end of the workspace.
1805
1806 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
1807
1808
1809 Version 8.01 19-Jan-2010
1810 ------------------------
1811
1812 1. If a pattern contained a conditional subpattern with only one branch (in
1813 particular, this includes all (*DEFINE) patterns), a call to pcre_study()
1814 computed the wrong minimum data length (which is of course zero for such
1815 subpatterns). This could cause incorrect "no match" results.
1816
1817 2. For patterns such as (?i)a(?-i)b|c where an option setting at the start of
1818 the pattern is reset in the first branch, pcre_compile() failed with
1819 "internal error: code overflow at offset...". This happened only when
1820 the reset was to the original external option setting. (An optimization
1821 abstracts leading options settings into an external setting, which was the
1822 cause of this.)
1823
1824 3. A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
1825 of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
1826 assertion pattern did not match (meaning that the assertion was true), it
1827 was incorrectly treated as false if the SKIP had been reached during the
1828 matching. This also applied to assertions used as conditions.
1829
1830 4. If an item that is not supported by pcre_dfa_exec() was encountered in an
1831 assertion subpattern, including such a pattern used as a condition,
1832 unpredictable results occurred, instead of the error return
1833 PCRE_ERROR_DFA_UITEM.
1834
1835 5. The C++ GlobalReplace function was not working like Perl for the special
1836 situation when an empty string is matched. It now does the fancy magic
1837 stuff that is necessary.
1838
1839 6. In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
1840 removed. (These were left over from very, very early versions of PCRE.)
1841
1842 7. Some cosmetic changes to the code to make life easier when compiling it
1843 as part of something else:
1844
1845 (a) Change DEBUG to PCRE_DEBUG.
1846
1847 (b) In pcre_compile(), rename the member of the "branch_chain" structure
1848 called "current" as "current_branch", to prevent a collision with the
1849 Linux macro when compiled as a kernel module.
1850
1851 (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
1852 prevent a collision with the Linux macro when compiled as a kernel
1853 module.
1854
1855 8. In pcre_compile() there are some checks for integer overflows that used to
1856 cast potentially large values to (double). This has been changed to that
1857 when building, a check for int64_t is made, and if it is found, it is used
1858 instead, thus avoiding the use of floating point arithmetic. (There is no
1859 other use of FP in PCRE.) If int64_t is not found, the fallback is to
1860 double.
1861
1862 9. Added two casts to avoid signed/unsigned warnings from VS Studio Express
1863 2005 (difference between two addresses compared to an unsigned value).
1864
1865 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
1866 custom one, because of the following reported problem in Windows:
1867
1868 - libbz2 uses the Pascal calling convention (WINAPI) for the functions
1869 under Win32.
1870 - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
1871 therefore missing the function definition.
1872 - The compiler thus generates a "C" signature for the test function.
1873 - The linker fails to find the "C" function.
1874 - PCRE fails to configure if asked to do so against libbz2.
1875
1876 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
1877 messages were output:
1878
1879 Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
1880 rerunning libtoolize, to keep the correct libtool macros in-tree.
1881 Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
1882
1883 I have done both of these things.
1884
1885 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
1886 most of the time, it *can* run out if it is given a pattern that contains a
1887 runaway infinite recursion. I updated the discussion in the pcrestack man
1888 page.
1889
1890 13. Now that we have gone to the x.xx style of version numbers, the minor
1891 version may start with zero. Using 08 or 09 is a bad idea because users
1892 might check the value of PCRE_MINOR in their code, and 08 or 09 may be
1893 interpreted as invalid octal numbers. I've updated the previous comment in
1894 configure.ac, and also added a check that gives an error if 08 or 09 are
1895 used.
1896
1897 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
1898 causing partial matching to fail when the end of the subject matched \W
1899 in a UTF-8 pattern where \W was quantified with a minimum of 3.
1900
1901 15. There were some discrepancies between the declarations in pcre_internal.h
1902 of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
1903 their definitions. The declarations used "const uschar *" and the
1904 definitions used USPTR. Even though USPTR is normally defined as "const
1905 unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
1906 reported that: "This difference in casting confuses some C++ compilers, for
1907 example, SunCC recognizes above declarations as different functions and
1908 generates broken code for hbpcre." I have changed the declarations to use
1909 USPTR.
1910
1911 16. GNU libtool is named differently on some systems. The autogen.sh script now
1912 tries several variants such as glibtoolize (MacOSX) and libtoolize1x
1913 (FreeBSD).
1914
1915 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
1916 (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
1917 comment: "Figure out how to create a longlong from a string: strtoll and
1918 equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
1919 instance, but it only takes 2 args instead of 3!"
1920
1921 18. A subtle bug concerned with back references has been fixed by a change of
1922 specification, with a corresponding code fix. A pattern such as
1923 ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
1924 refers, was giving matches when it shouldn't. For example, xa=xaaa would
1925 match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
1926 same bug. Such groups have to be quantified to be useful, or contained
1927 inside another quantified group. (If there's no repetition, the reference
1928 can never match.) The problem arises because, having left the group and
1929 moved on to the rest of the pattern, a later failure that backtracks into
1930 the group uses the captured value from the final iteration of the group
1931 rather than the correct earlier one. I have fixed this in PCRE by forcing
1932 any group that contains a reference to itself to be an atomic group; that
1933 is, there cannot be any backtracking into it once it has completed. This is
1934 similar to recursive and subroutine calls.
1935
1936
1937 Version 8.00 19-Oct-09
1938 ----------------------
1939
1940 1. The table for translating pcre_compile() error codes into POSIX error codes
1941 was out-of-date, and there was no check on the pcre_compile() error code
1942 being within the table. This could lead to an OK return being given in
1943 error.
1944
1945 2. Changed the call to open a subject file in pcregrep from fopen(pathname,
1946 "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
1947 in a Windows environment.
1948
1949 3. The pcregrep --count option prints the count for each file even when it is
1950 zero, as does GNU grep. However, pcregrep was also printing all files when
1951 --files-with-matches was added. Now, when both options are given, it prints
1952 counts only for those files that have at least one match. (GNU grep just
1953 prints the file name in this circumstance, but including the count seems
1954 more useful - otherwise, why use --count?) Also ensured that the
1955 combination -clh just lists non-zero counts, with no names.
1956
1957 4. The long form of the pcregrep -F option was incorrectly implemented as
1958 --fixed_strings instead of --fixed-strings. This is an incompatible change,
1959 but it seems right to fix it, and I didn't think it was worth preserving
1960 the old behaviour.
1961
1962 5. The command line items --regex=pattern and --regexp=pattern were not
1963 recognized by pcregrep, which required --regex pattern or --regexp pattern
1964 (with a space rather than an '='). The man page documented the '=' forms,
1965 which are compatible with GNU grep; these now work.
1966
1967 6. No libpcreposix.pc file was created for pkg-config; there was just
1968 libpcre.pc and libpcrecpp.pc. The omission has been rectified.
1969
1970 7. Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
1971 when UCP support is not needed, by modifying the Python script that
1972 generates it from Unicode data files. This should not matter if the module
1973 is correctly used as a library, but I received one complaint about 50K of
1974 unwanted data. My guess is that the person linked everything into his
1975 program rather than using a library. Anyway, it does no harm.
1976
1977 8. A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
1978 was a minimum greater than 1 for a wide character in a possessive
1979 repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
1980 which had an unlimited repeat of a nested, fixed maximum repeat of a wide
1981 character. Chaos in the form of incorrect output or a compiling loop could
1982 result.
1983
1984 9. The restrictions on what a pattern can contain when partial matching is
1985 requested for pcre_exec() have been removed. All patterns can now be
1986 partially matched by this function. In addition, if there are at least two
1987 slots in the offset vector, the offset of the earliest inspected character
1988 for the match and the offset of the end of the subject are set in them when
1989 PCRE_ERROR_PARTIAL is returned.
1990
1991 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
1992 synonymous with PCRE_PARTIAL, for backwards compatibility, and
1993 PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
1994 and may be more useful for multi-segment matching.
1995
1996 11. Partial matching with pcre_exec() is now more intuitive. A partial match
1997 used to be given if ever the end of the subject was reached; now it is
1998 given only if matching could not proceed because another character was
1999 needed. This makes a difference in some odd cases such as Z(*FAIL) with the
2000 string "Z", which now yields "no match" instead of "partial match". In the
2001 case of pcre_dfa_exec(), "no match" is given if every matching path for the
2002 final character ended with (*FAIL).
2003
2004 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
2005 if the pattern had a "must contain" character that was already found in the
2006 earlier partial match, unless partial matching was again requested. For
2007 example, with the pattern /dog.(body)?/, the "must contain" character is
2008 "g". If the first part-match was for the string "dog", restarting with
2009 "sbody" failed. This bug has been fixed.
2010
2011 13. The string returned by pcre_dfa_exec() after a partial match has been
2012 changed so that it starts at the first inspected character rather than the
2013 first character of the match. This makes a difference only if the pattern
2014 starts with a lookbehind assertion or \b or \B (\K is not supported by
2015 pcre_dfa_exec()). It's an incompatible change, but it makes the two
2016 matching functions compatible, and I think it's the right thing to do.
2017
2018 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
2019 so that the demonstration program is easily available in environments where
2020 PCRE has not been installed from source.
2021
2022 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
2023 libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
2024 library.
2025
2026 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
2027 It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
2028 is not the first non-POSIX option to be added. Clearly some people find
2029 these options useful.
2030
2031 17. If a caller to the POSIX matching function regexec() passes a non-zero
2032 value for nmatch with a NULL value for pmatch, the value of
2033 nmatch is forced to zero.
2034
2035 18. RunGrepTest did not have a test for the availability of the -u option of
2036 the diff command, as RunTest does. It now checks in the same way as
2037 RunTest, and also checks for the -b option.
2038
2039 19. If an odd number of negated classes containing just a single character
2040 interposed, within parentheses, between a forward reference to a named
2041 subpattern and the definition of the subpattern, compilation crashed with
2042 an internal error, complaining that it could not find the referenced
2043 subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
2044 [The bug was that it was starting one character too far in when skipping
2045 over the character class, thus treating the ] as data rather than
2046 terminating the class. This meant it could skip too much.]
2047
2048 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
2049 /g option in pcretest when the pattern contains \K, which makes it possible
2050 to have an empty string match not at the start, even when the pattern is
2051 anchored. Updated pcretest and pcredemo to use this option.
2052
2053 21. If the maximum number of capturing subpatterns in a recursion was greater
2054 than the maximum at the outer level, the higher number was returned, but
2055 with unset values at the outer level. The correct (outer level) value is
2056 now given.
2057
2058 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
2059 PCRE did not set those parentheses (unlike Perl). I have now found a way to
2060 make it do so. The string so far is captured, making this feature
2061 compatible with Perl.
2062
2063 23. The tests have been re-organized, adding tests 11 and 12, to make it
2064 possible to check the Perl 5.10 features against Perl 5.10.
2065
2066 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
2067 pattern matches a fixed length string. PCRE did not allow this; now it
2068 does. Neither allows recursion.
2069
2070 25. I finally figured out how to implement a request to provide the minimum
2071 length of subject string that was needed in order to match a given pattern.
2072 (It was back references and recursion that I had previously got hung up
2073 on.) This code has now been added to pcre_study(); it finds a lower bound
2074 to the length of subject needed. It is not necessarily the greatest lower
2075 bound, but using it to avoid searching strings that are too short does give
2076 some useful speed-ups. The value is available to calling programs via
2077 pcre_fullinfo().
2078
2079 26. While implementing 25, I discovered to my embarrassment that pcretest had
2080 not been passing the result of pcre_study() to pcre_dfa_exec(), so the
2081 study optimizations had never been tested with that matching function.
2082 Oops. What is worse, even when it was passed study data, there was a bug in
2083 pcre_dfa_exec() that meant it never actually used it. Double oops. There
2084 were also very few tests of studied patterns with pcre_dfa_exec().
2085
2086 27. If (?| is used to create subpatterns with duplicate numbers, they are now
2087 allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
2088 on the other side of the coin, they are no longer allowed to have different
2089 names, because these cannot be distinguished in PCRE, and this has caused
2090 confusion. (This is a difference from Perl.)
2091
2092 28. When duplicate subpattern names are present (necessarily with different
2093 numbers, as required by 27 above), and a test is made by name in a
2094 conditional pattern, either for a subpattern having been matched, or for
2095 recursion in such a pattern, all the associated numbered subpatterns are
2096 tested, and the overall condition is true if the condition is true for any
2097 one of them. This is the way Perl works, and is also more like the way
2098 testing by number works.
2099
2100
2101 Version 7.9 11-Apr-09
2102 ---------------------
2103
2104 1. When building with support for bzlib/zlib (pcregrep) and/or readline
2105 (pcretest), all targets were linked against these libraries. This included
2106 libpcre, libpcreposix, and libpcrecpp, even though they do not use these
2107 libraries. This caused unwanted dependencies to be created. This problem
2108 has been fixed, and now only pcregrep is linked with bzlib/zlib and only
2109 pcretest is linked with readline.
2110
2111 2. The "typedef int BOOL" in pcre_internal.h that was included inside the
2112 "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
2113 moved outside it again, because FALSE and TRUE are already defined in AIX,
2114 but BOOL is not.
2115
2116 3. The pcre_config() function was treating the PCRE_MATCH_LIMIT and
2117 PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
2118
2119 4. The pcregrep documentation said spaces were inserted as well as colons (or
2120 hyphens) following file names and line numbers when outputting matching
2121 lines. This is not true; no spaces are inserted. I have also clarified the
2122 wording for the --colour (or --color) option.
2123
2124 5. In pcregrep, when --colour was used with -o, the list of matching strings
2125 was not coloured; this is different to GNU grep, so I have changed it to be
2126 the same.
2127
2128 6. When --colo(u)r was used in pcregrep, only the first matching substring in
2129 each matching line was coloured. Now it goes on to look for further matches
2130 of any of the test patterns, which is the same behaviour as GNU grep.
2131
2132 7. A pattern that could match an empty string could cause pcregrep to loop; it
2133 doesn't make sense to accept an empty string match in pcregrep, so I have
2134 locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
2135 seems to be how GNU grep behaves. [But see later change 40 for release
2136 8.33.]
2137
2138 8. The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
2139 start or after a newline", because the conditional assertion was not being
2140 correctly handled. The rule now is that both the assertion and what follows
2141 in the first alternative must satisfy the test.
2142
2143 9. If auto-callout was enabled in a pattern with a conditional group whose
2144 condition was an assertion, PCRE could crash during matching, both with
2145 pcre_exec() and pcre_dfa_exec().
2146
2147 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
2148 used for matching.
2149
2150 11. Unicode property support in character classes was not working for
2151 characters (bytes) greater than 127 when not in UTF-8 mode.
2152
2153 12. Added the -M command line option to pcretest.
2154
2155 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
2156
2157 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
2158
2159 16. Added comments and documentation about mis-use of no_arg in the C++
2160 wrapper.
2161
2162 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
2163 from Martin Jerabek that uses macro names for all relevant character and
2164 string constants.
2165
2166 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
2167 SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
2168 SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
2169 these, but not everybody uses configure.
2170
2171 19. A conditional group that had only one branch was not being correctly
2172 recognized as an item that could match an empty string. This meant that an
2173 enclosing group might also not be so recognized, causing infinite looping
2174 (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
2175 with the subject "ab", where knowledge that the repeated group can match
2176 nothing is needed in order to break the loop.
2177
2178 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
2179 exec(), but without supplying a callout function, matching went wrong.
2180
2181 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
2182 leak if the size of the offset vector was greater than 30. When the vector
2183 is smaller, the saved offsets during recursion go onto a local stack
2184 vector, but for larger vectors malloc() is used. It was failing to free
2185 when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
2186 error, in fact).
2187
2188 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
2189 heapframe that is used only when UTF-8 support is enabled. This caused no
2190 problem, but was untidy.
2191
2192 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
2193 CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
2194 included within another project.
2195
2196 24. Steven Van Ingelgem's patches to add more options to the CMake support,
2197 slightly modified by me:
2198
2199 (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
2200 not building pcregrep.
2201
2202 (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
2203 if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
2204
2205 25. Forward references, both numeric and by name, in patterns that made use of
2206 duplicate group numbers, could behave incorrectly or give incorrect errors,
2207 because when scanning forward to find the reference group, PCRE was not
2208 taking into account the duplicate group numbers. A pattern such as
2209 ^X(?3)(a)(?|(b)|(q))(Y) is an example.
2210
2211 26. Changed a few more instances of "const unsigned char *" to USPTR, making
2212 the feature of a custom pointer more persuasive (as requested by a user).
2213
2214 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
2215 pcretest.c, inside #ifndefs, because it seems they are sometimes already
2216 pre-defined.
2217
2218 28. Added support for (*UTF8) at the start of a pattern.
2219
2220 29. Arrange for flags added by the "release type" setting in CMake to be shown
2221 in the configuration summary.
2222
2223
2224 Version 7.8 05-Sep-08
2225 ---------------------
2226
2227 1. Replaced UCP searching code with optimized version as implemented for Ad
2228 Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
2229 stage table and inline lookup instead of a function, giving speed ups of 2
2230 to 5 times on some simple patterns that I tested. Permission was given to
2231 distribute the MultiStage2.py script that generates the tables (it's not in
2232 the tarball, but is in the Subversion repository).
2233
2234 2. Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
2235 scripts.
2236
2237 3. Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
2238 a group with a zero qualifier. The result of the study could be incorrect,
2239 or the function might crash, depending on the pattern.
2240
2241 4. Caseless matching was not working for non-ASCII characters in back
2242 references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
2243 It now works when Unicode Property Support is available.
2244
2245 5. In pcretest, an escape such as \x{de} in the data was always generating
2246 a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
2247 non-UTF-8 mode. If the value is greater than 255, it gives a warning about
2248 truncation.
2249
2250 6. Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
2251
2252 7. Added two (int) casts to pcregrep when printing the difference of two
2253 pointers, in case they are 64-bit values.
2254
2255 8. Added comments about Mac OS X stack usage to the pcrestack man page and to
2256 test 2 if it fails.
2257
2258 9. Added PCRE_CALL_CONVENTION just before the names of all exported functions,
2259 and a #define of that name to empty if it is not externally set. This is to
2260 allow users of MSVC to set it if necessary.
2261
2262 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
2263 the convenience functions in the pcre_get.c source file.
2264
2265 11. An option change at the start of a pattern that had top-level alternatives
2266 could cause overwriting and/or a crash. This command provoked a crash in
2267 some environments:
2268
2269 printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
2270
2271 This potential security problem was recorded as CVE-2008-2371.
2272
2273 12. For a pattern where the match had to start at the beginning or immediately
2274 after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
2275 pcre_dfa_exec() could read past the end of the passed subject if there was
2276 no match. To help with detecting such bugs (e.g. with valgrind), I modified
2277 pcretest so that it places the subject at the end of its malloc-ed buffer.
2278
2279 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
2280 exec() might read past the end of the data buffer in UTF-8 mode.
2281
2282 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
2283 the data contained the byte 0x85 as part of a UTF-8 character within its
2284 first line. This applied both to normal and DFA matching.
2285
2286 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
2287 /^[^d]*?$/8 failed to match "abc".
2288
2289 16. Added a missing copyright notice to pcrecpp_internal.h.
2290
2291 17. Make it more clear in the documentation that values returned from
2292 pcre_exec() in ovector are byte offsets, not character counts.
2293
2294 18. Tidied a few places to stop certain compilers from issuing warnings.
2295
2296 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
2297 supplied by Stefan Weber. I made a further small update for 7.8 because
2298 there is a change of source arrangements: the pcre_searchfuncs.c module is
2299 replaced by pcre_ucd.c.
2300
2301
2302 Version 7.7 07-May-08
2303 ---------------------
2304
2305 1. Applied Craig's patch to sort out a long long problem: "If we can't convert
2306 a string to a long long, pretend we don't even have a long long." This is
2307 done by checking for the strtoq, strtoll, and _strtoi64 functions.
2308
2309 2. Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
2310 pre-7.6 versions, which defined a global no_arg variable instead of putting
2311 it in the RE class. (See also #8 below.)
2312
2313 3. Remove a line of dead code, identified by coverity and reported by Nuno
2314 Lopes.
2315
2316 4. Fixed two related pcregrep bugs involving -r with --include or --exclude:
2317
2318 (1) The include/exclude patterns were being applied to the whole pathnames
2319 of files, instead of just to the final components.
2320
2321 (2) If there was more than one level of directory, the subdirectories were
2322 skipped unless they satisfied the include/exclude conditions. This is
2323 inconsistent with GNU grep (and could even be seen as contrary to the
2324 pcregrep specification - which I improved to make it absolutely clear).
2325 The action now is always to scan all levels of directory, and just
2326 apply the include/exclude patterns to regular files.
2327
2328 5. Added the --include_dir and --exclude_dir patterns to pcregrep, and used
2329 --exclude_dir in the tests to avoid scanning .svn directories.
2330
2331 6. Applied Craig's patch to the QuoteMeta function so that it escapes the
2332 NUL character as backslash + 0 rather than backslash + NUL, because PCRE
2333 doesn't support NULs in patterns.
2334
2335 7. Added some missing "const"s to declarations of static tables in
2336 pcre_compile.c and pcre_dfa_exec.c.
2337
2338 8. Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
2339 caused by fix #2 above. (Subsequently also a second patch to fix the
2340 first patch. And a third patch - this was a messy problem.)
2341
2342 9. Applied Craig's patch to remove the use of push_back().
2343
2344 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
2345 matching function regexec().
2346
2347 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
2348 which, however, unlike Perl's \g{...}, are subroutine calls, not back
2349 references. PCRE supports relative numbers with this syntax (I don't think
2350 Oniguruma does).
2351
2352 12. Previously, a group with a zero repeat such as (...){0} was completely
2353 omitted from the compiled regex. However, this means that if the group
2354 was called as a subroutine from elsewhere in the pattern, things went wrong
2355 (an internal error was given). Such groups are now left in the compiled
2356 pattern, with a new opcode that causes them to be skipped at execution
2357 time.
2358
2359 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
2360 to the way PCRE behaves:
2361
2362 (a) A lone ] character is dis-allowed (Perl treats it as data).
2363
2364 (b) A back reference to an unmatched subpattern matches an empty string
2365 (Perl fails the current match path).
2366
2367 (c) A data ] in a character class must be notated as \] because if the
2368 first data character in a class is ], it defines an empty class. (In
2369 Perl it is not possible to have an empty class.) The empty class []
2370 never matches; it forces failure and is equivalent to (*FAIL) or (?!).
2371 The negative empty class [^] matches any one character, independently
2372 of the DOTALL setting.
2373
2374 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
2375 non-existent subpattern following a character class starting with ']' and
2376 containing () gave an internal compiling error instead of "reference to
2377 non-existent subpattern". Fortunately, when the pattern did exist, the
2378 compiled code was correct. (When scanning forwards to check for the
2379 existence of the subpattern, it was treating the data ']' as terminating
2380 the class, so got the count wrong. When actually compiling, the reference
2381 was subsequently set up correctly.)
2382
2383 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
2384 it was being rejected as not supported by pcre_dfa_exec(), even though
2385 other assertions are supported. I have made pcre_dfa_exec() support
2386 (*FAIL).
2387
2388 16. The implementation of 13c above involved the invention of a new opcode,
2389 OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
2390 cannot be changed at match time, I realized I could make a small
2391 improvement to matching performance by compiling OP_ALLANY instead of
2392 OP_ANY for "." when DOTALL was set, and then removing the runtime tests
2393 on the OP_ANY path.
2394
2395 17. Compiling pcretest on Windows with readline support failed without the
2396 following two fixes: (1) Make the unistd.h include conditional on
2397 HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
2398
2399 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
2400 ncurses library to be included for pcretest when ReadLine support is
2401 requested, but also to allow for it to be overridden. This patch came from
2402 Daniel Bergström.
2403
2404 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
2405 as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
2406 any errors with the current Unicode tables. Thanks to Peter Kankowski for
2407 spotting this.
2408
2409
2410 Version 7.6 28-Jan-08
2411 ---------------------
2412
2413 1. A character class containing a very large number of characters with
2414 codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
2415 overflow.
2416
2417 2. Patch to cut out the "long long" test in pcrecpp_unittest when
2418 HAVE_LONG_LONG is not defined.
2419
2420 3. Applied Christian Ehrlicher's patch to update the CMake build files to
2421 bring them up to date and include new features. This patch includes:
2422
2423 - Fixed PH's badly added libz and libbz2 support.
2424 - Fixed a problem with static linking.
2425 - Added pcredemo. [But later removed - see 7 below.]
2426 - Fixed dftables problem and added an option.
2427 - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
2428 HAVE_LONG_LONG.
2429 - Added readline support for pcretest.
2430 - Added an listing of the option settings after cmake has run.
2431
2432 4. A user submitted a patch to Makefile that makes it easy to create
2433 "pcre.dll" under mingw when using Configure/Make. I added stuff to
2434 Makefile.am that cause it to include this special target, without
2435 affecting anything else. Note that the same mingw target plus all
2436 the other distribution libraries and programs are now supported
2437 when configuring with CMake (see 6 below) instead of with
2438 Configure/Make.
2439
2440 5. Applied Craig's patch that moves no_arg into the RE class in the C++ code.
2441 This is an attempt to solve the reported problem "pcrecpp::no_arg is not
2442 exported in the Windows port". It has not yet been confirmed that the patch
2443 solves the problem, but it does no harm.
2444
2445 6. Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
2446 NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
2447 with CMake, and also correct the comment about stack recursion.
2448
2449 7. Remove the automatic building of pcredemo from the ./configure system and
2450 from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
2451 of a program that users should build themselves after PCRE is installed, so
2452 building it automatically is not really right. What is more, it gave
2453 trouble in some build environments.
2454
2455 8. Further tidies to CMakeLists.txt from Sheri and Christian.
2456
2457
2458 Version 7.5 10-Jan-08
2459 ---------------------
2460
2461 1. Applied a patch from Craig: "This patch makes it possible to 'ignore'
2462 values in parens when parsing an RE using the C++ wrapper."
2463
2464 2. Negative specials like \S did not work in character classes in UTF-8 mode.
2465 Characters greater than 255 were excluded from the class instead of being
2466 included.
2467
2468 3. The same bug as (2) above applied to negated POSIX classes such as
2469 [:^space:].
2470
2471 4. PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
2472 defined or documented. It seems to have been a typo for PCRE_STATIC, so
2473 I have changed it.
2474
2475 5. The construct (?&) was not diagnosed as a syntax error (it referenced the
2476 first named subpattern) and a construct such as (?&a) would reference the
2477 first named subpattern whose name started with "a" (in other words, the
2478 length check was missing). Both these problems are fixed. "Subpattern name
2479 expected" is now given for (?&) (a zero-length name), and this patch also
2480 makes it give the same error for \k'' (previously it complained that that
2481 was a reference to a non-existent subpattern).
2482
2483 6. The erroneous patterns (?+-a) and (?-+a) give different error messages;
2484 this is right because (?- can be followed by option settings as well as by
2485 digits. I have, however, made the messages clearer.
2486
2487 7. Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
2488 than the number used in the conditional) now cause a compile-time error.
2489 This is actually not compatible with Perl, which accepts such patterns, but
2490 treats the conditional as always being FALSE (as PCRE used to), but it
2491 seems to me that giving a diagnostic is better.
2492
2493 8. Change "alphameric" to the more common word "alphanumeric" in comments
2494 and messages.
2495
2496 9. Fix two occurrences of "backslash" in comments that should have been
2497 "backspace".
2498
2499 10. Remove two redundant lines of code that can never be obeyed (their function
2500 was moved elsewhere).
2501
2502 11. The program that makes PCRE's Unicode character property table had a bug
2503 which caused it to generate incorrect table entries for sequences of
2504 characters that have the same character type, but are in different scripts.
2505 It amalgamated them into a single range, with the script of the first of
2506 them. In other words, some characters were in the wrong script. There were
2507 thirteen such cases, affecting characters in the following ranges:
2508
2509 U+002b0 - U+002c1
2510 U+0060c - U+0060d
2511 U+0061e - U+00612
2512 U+0064b - U+0065e
2513 U+0074d - U+0076d
2514 U+01800 - U+01805
2515 U+01d00 - U+01d77
2516 U+01d9b - U+01dbf
2517 U+0200b - U+0200f
2518 U+030fc - U+030fe
2519 U+03260 - U+0327f
2520 U+0fb46 - U+0fbb1
2521 U+10450 - U+1049d
2522
2523 12. The -o option (show only the matching part of a line) for pcregrep was not
2524 compatible with GNU grep in that, if there was more than one match in a
2525 line, it showed only the first of them. It now behaves in the same way as
2526 GNU grep.
2527
2528 13. If the -o and -v options were combined for pcregrep, it printed a blank
2529 line for every non-matching line. GNU grep prints nothing, and pcregrep now
2530 does the same. The return code can be used to tell if there were any
2531 non-matching lines.
2532
2533 14. Added --file-offsets and --line-offsets to pcregrep.
2534
2535 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
2536 infinitely looping recursion. The bug was that positive lookaheads were not
2537 being skipped when checking for a possible empty match (negative lookaheads
2538 and both kinds of lookbehind were skipped).
2539
2540 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
2541 inclusion of <windows.h> to before rather than after the definition of
2542 INVALID_FILE_ATTRIBUTES (patch from David Byron).
2543
2544 17. Specifying a possessive quantifier with a specific limit for a Unicode
2545 character property caused pcre_compile() to compile bad code, which led at
2546 runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
2547 are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
2548 caused the error; without that there was no problem.
2549
2550 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
2551
2552 19. Added --enable-pcretest-libreadline.
2553
2554 20. In pcrecpp.cc, the variable 'count' was incremented twice in
2555 RE::GlobalReplace(). As a result, the number of replacements returned was
2556 double what it should be. I removed one of the increments, but Craig sent a
2557 later patch that removed the other one (the right fix) and added unit tests
2558 that check the return values (which was not done before).
2559
2560 21. Several CMake things:
2561
2562 (1) Arranged that, when cmake is used on Unix, the libraries end up with
2563 the names libpcre and libpcreposix, not just pcre and pcreposix.
2564
2565 (2) The above change means that pcretest and pcregrep are now correctly
2566 linked with the newly-built libraries, not previously installed ones.
2567
2568 (3) Added PCRE_SUPPORT_LIBREADLINE, PCRE_SUPPORT_LIBZ, PCRE_SUPPORT_LIBBZ2.
2569
2570 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
2571 crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
2572 UTF-8 newline character). The key issue is that the pattern starts .*;
2573 this means that the match must be either at the beginning, or after a
2574 newline. The bug was in the code for advancing after a failed match and
2575 checking that the new position followed a newline. It was not taking
2576 account of UTF-8 characters correctly.
2577
2578 23. PCRE was behaving differently from Perl in the way it recognized POSIX
2579 character classes. PCRE was not treating the sequence [:...:] as a
2580 character class unless the ... were all letters. Perl, however, seems to
2581 allow any characters between [: and :], though of course it rejects as
2582 unknown any "names" that contain non-letters, because all the known class
2583 names consist only of letters. Thus, Perl gives an error for [[:1234:]],
2584 for example, whereas PCRE did not - it did not recognize a POSIX character
2585 class. This seemed a bit dangerous, so the code has been changed to be
2586 closer to Perl. The behaviour is not identical to Perl, because PCRE will
2587 diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
2588 treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
2589 Perl does, and where it didn't before.
2590
2591 24. Rewrite so as to remove the single use of %n from pcregrep because in some
2592 Windows environments %n is disabled by default.
2593
2594
2595 Version 7.4 21-Sep-07
2596 ---------------------
2597
2598 1. Change 7.3/28 was implemented for classes by looking at the bitmap. This
2599 means that a class such as [\s] counted as "explicit reference to CR or
2600 LF". That isn't really right - the whole point of the change was to try to
2601 help when there was an actual mention of one of the two characters. So now
2602 the change happens only if \r or \n (or a literal CR or LF) character is
2603 encountered.
2604
2605 2. The 32-bit options word was also used for 6 internal flags, but the numbers
2606 of both had grown to the point where there were only 3 bits left.
2607 Fortunately, there was spare space in the data structure, and so I have
2608 moved the internal flags into a new 16-bit field to free up more option
2609 bits.
2610
2611 3. The appearance of (?J) at the start of a pattern set the DUPNAMES option,
2612 but did not set the internal JCHANGED flag - either of these is enough to
2613 control the way the "get" function works - but the PCRE_INFO_JCHANGED
2614 facility is supposed to tell if (?J) was ever used, so now (?J) at the
2615 start sets both bits.
2616
2617 4. Added options (at build time, compile time, exec time) to change \R from
2618 matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
2619
2620 5. doc/pcresyntax.html was missing from the distribution.
2621
2622 6. Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
2623 compatibility, even though it is no longer used.
2624
2625 7. Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
2626 strtoull to pcrecpp.cc to select the available functions in WIN32 when the
2627 windows.h file is present (where different names are used). [This was
2628 reversed later after testing - see 16 below.]
2629
2630 8. Changed all #include <config.h> to #include "config.h". There were also
2631 some further <pcre.h> cases that I changed to "pcre.h".
2632
2633 9. When pcregrep was used with the --colour option, it missed the line ending
2634 sequence off the lines that it output.
2635
2636 10. It was pointed out to me that arrays of string pointers cause lots of
2637 relocations when a shared library is dynamically loaded. A technique of
2638 using a single long string with a table of offsets can drastically reduce
2639 these. I have refactored PCRE in four places to do this. The result is
2640 dramatic:
2641
2642 Originally: 290
2643 After changing UCP table: 187
2644 After changing error message table: 43
2645 After changing table of "verbs" 36
2646 After changing table of Posix names 22
2647
2648 Thanks to the folks working on Gregex for glib for this insight.
2649
2650 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
2651 unicode-properties was also set.
2652
2653 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
2654
2655 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
2656 checked only for CRLF.
2657
2658 14. Added casts to pcretest.c to avoid compiler warnings.
2659
2660 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
2661
2662 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
2663 and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
2664 entirely. This removes changes made in 7 above.
2665
2666 17. The CMake files have been updated, and there is now more information about
2667 building with CMake in the NON-UNIX-USE document.
2668
2669
2670 Version 7.3 28-Aug-07
2671 ---------------------
2672
2673 1. In the rejigging of the build system that eventually resulted in 7.1, the
2674 line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
2675 brackets there is not right, since it causes compilers to look for an
2676 installed pcre.h, not the version that is in the source that is being
2677 compiled (which of course may be different). I have changed it back to:
2678
2679 #include "pcre.h"
2680
2681 I have a vague recollection that the change was concerned with compiling in
2682 different directories, but in the new build system, that is taken care of
2683 by the VPATH setting the Makefile.
2684
2685 2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
2686 when the subject happened to end in the byte 0x85 (e.g. if the last
2687 character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
2688 characters but of course it shouldn't be taken as a newline when it is part
2689 of another character. The bug was that, for an unlimited repeat of . in
2690 not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
2691 characters when looking for a newline.
2692
2693 3. A small performance improvement in the DOTALL UTF-8 mode .* case.
2694
2695 4. Debugging: adjusted the names of opcodes for different kinds of parentheses
2696 in debug output.
2697
2698 5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
2699 long printing in the pcrecpp unittest when running under MinGW.
2700
2701 6. ESC_K was left out of the EBCDIC table.
2702
2703 7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
2704 parentheses; I made it 1000, which seemed large enough. Unfortunately, the
2705 limit also applies to "virtual nesting" when a pattern is recursive, and in
2706 this case 1000 isn't so big. I have been able to remove this limit at the
2707 expense of backing off one optimization in certain circumstances. Normally,
2708 when pcre_exec() would call its internal match() function recursively and
2709 immediately return the result unconditionally, it uses a "tail recursion"
2710 feature to save stack. However, when a subpattern that can match an empty
2711 string has an unlimited repetition quantifier, it no longer makes this
2712 optimization. That gives it a stack frame in which to save the data for
2713 checking that an empty string has been matched. Previously this was taken
2714 from the 1000-entry workspace that had been reserved. So now there is no
2715 explicit limit, but more stack is used.
2716
2717 8. Applied Daniel's patches to solve problems with the import/export magic
2718 syntax that is required for Windows, and which was going wrong for the
2719 pcreposix and pcrecpp parts of the library. These were overlooked when this
2720 problem was solved for the main library.
2721
2722 9. There were some crude static tests to avoid integer overflow when computing
2723 the size of patterns that contain repeated groups with explicit upper
2724 limits. As the maximum quantifier is 65535, the maximum group length was
2725 set at 30,000 so that the product of these two numbers did not overflow a
2726 32-bit integer. However, it turns out that people want to use groups that
2727 are longer than 30,000 bytes (though not repeat them that many times).
2728 Change 7.0/17 (the refactoring of the way the pattern size is computed) has
2729 made it possible to implement the integer overflow checks in a much more
2730 dynamic way, which I have now done. The artificial limitation on group
2731 length has been removed - we now have only the limit on the total length of
2732 the compiled pattern, which depends on the LINK_SIZE setting.
2733
2734 10. Fixed a bug in the documentation for get/copy named substring when
2735 duplicate names are permitted. If none of the named substrings are set, the
2736 functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
2737 empty string.
2738
2739 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
2740 instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
2741 because the ] is interpreted as the first data character and the
2742 terminating ] is not found. PCRE has been made compatible with Perl in this
2743 regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
2744 cause memory overwriting.
2745
2746 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
2747 string has been matched (to stop an infinite loop). It was not recognizing
2748 a conditional subpattern that could match an empty string if that
2749 subpattern was within another subpattern. For example, it looped when
2750 trying to match (((?(1)X|))*) but it was OK with ((?(1)X|)*) where the
2751 condition was not nested. This bug has been fixed.
2752
2753 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
2754 past the start of the subject in the presence of bytes with the top bit
2755 set, for example "\x8aBCD".
2756
2757 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
2758 (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
2759
2760 14. Optimized (?!) to (*FAIL).
2761
2762 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
2763 This restricts code points to be within the range 0 to 0x10FFFF, excluding
2764 the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
2765 full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
2766 does: it's just the validity check that is more restrictive.
2767
2768 16. Inserted checks for integer overflows during escape sequence (backslash)
2769 processing, and also fixed erroneous offset values for syntax errors during
2770 backslash processing.
2771
2772 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
2773 for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
2774
2775 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
2776 caused an overrun.
2777
2778 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
2779 something other than just ASCII characters) inside a group that had an
2780 unlimited repeat caused a loop at compile time (while checking to see
2781 whether the group could match an empty string).
2782
2783 20. Debugging a pattern containing \p or \P could cause a crash. For example,
2784 [\P{Any}] did so. (Error in the code for printing property names.)
2785
2786 21. An orphan \E inside a character class could cause a crash.
2787
2788 22. A repeated capturing bracket such as (A)? could cause a wild memory
2789 reference during compilation.
2790
2791 23. There are several functions in pcre_compile() that scan along a compiled
2792 expression for various reasons (e.g. to see if it's fixed length for look
2793 behind). There were bugs in these functions when a repeated \p or \P was
2794 present in the pattern. These operators have additional parameters compared
2795 with \d, etc, and these were not being taken into account when moving along
2796 the compiled data. Specifically:
2797
2798 (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
2799 length.
2800
2801 (b) An item such as \pL+ within a repeated group could cause crashes or
2802 loops.
2803
2804 (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
2805 "reference to non-existent subpattern" error.
2806
2807 (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
2808
2809 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
2810 characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
2811
2812 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
2813
2814 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
2815 character were causing crashes (broken optimization).
2816
2817 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
2818 \p or \P) caused a compile-time loop.
2819
2820 28. More problems have arisen in unanchored patterns when CRLF is a valid line
2821 break. For example, the unstudied pattern [\r\n]A does not match the string
2822 "\r\nA" because change 7.0/46 below moves the current point on by two
2823 characters after failing to match at the start. However, the pattern \nA
2824 *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
2825 the same is true. There doesn't seem any very clean way out of this, but
2826 what I have chosen to do makes the common cases work: PCRE now takes note
2827 of whether there can be an explicit match for \r or \n anywhere in the
2828 pattern, and if so, 7.0/46 no longer applies. As part of this change,
2829 there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
2830 pattern has explicit CR or LF references.
2831
2832 29. Added (*CR) etc for changing newline setting at start of pattern.
2833
2834
2835 Version 7.2 19-Jun-07
2836 ---------------------
2837
2838 1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
2839 which is apparently normally available under Windows.
2840
2841 2. Re-jig the pcregrep tests with different newline settings in an attempt
2842 to make them independent of the local environment's newline setting.
2843
2844 3. Add code to configure.ac to remove -g from the CFLAGS default settings.
2845
2846 4. Some of the "internals" tests were previously cut out when the link size
2847 was not 2, because the output contained actual offsets. The recent new
2848 "Z" feature of pcretest means that these can be cut out, making the tests
2849 usable with all link sizes.
2850
2851 5. Implemented Stan Switzer's goto replacement for longjmp() when not using
2852 stack recursion. This gives a massive performance boost under BSD, but just
2853 a small improvement under Linux. However, it saves one field in the frame
2854 in all cases.
2855
2856 6. Added more features from the forthcoming Perl 5.10:
2857
2858 (a) (?-n) (where n is a string of digits) is a relative subroutine or
2859 recursion call. It refers to the nth most recently opened parentheses.
2860
2861 (b) (?+n) is also a relative subroutine call; it refers to the nth next
2862 to be opened parentheses.
2863
2864 (c) Conditions that refer to capturing parentheses can be specified
2865 relatively, for example, (?(-2)... or (?(+3)...
2866
2867 (d) \K resets the start of the current match so that everything before
2868 is not part of it.
2869
2870 (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
2871
2872 (f) \g{name} is another synonym - part of Perl 5.10's unification of
2873 reference syntax.
2874
2875 (g) (?| introduces a group in which the numbering of parentheses in each
2876 alternative starts with the same number.
2877
2878 (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
2879
2880 7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
2881 PCRE_INFO_JCHANGED.
2882
2883 8. A pattern such as (.*(.)?)* caused pcre_exec() to fail by either not
2884 terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
2885 for detecting groups that can match an empty string.
2886
2887 9. A pattern with a very large number of alternatives (more than several
2888 hundred) was running out of internal workspace during the pre-compile
2889 phase, where pcre_compile() figures out how much memory will be needed. A
2890 bit of new cunning has reduced the workspace needed for groups with
2891 alternatives. The 1000-alternative test pattern now uses 12 bytes of
2892 workspace instead of running out of the 4096 that are available.
2893
2894 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
2895
2896 11. Applied patch from Google to remove an optimization that didn't quite work.
2897 The report of the bug said:
2898
2899 pcrecpp::RE("a*").FullMatch("aaa") matches, while
2900 pcrecpp::RE("a*?").FullMatch("aaa") does not, and
2901 pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
2902
2903 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
2904 it matched the wrong number of bytes.
2905
2906
2907 Version 7.1 24-Apr-07
2908 ---------------------
2909
2910 1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
2911 that is more "standard", making use of automake and other Autotools. There
2912 is some re-arrangement of the files and adjustment of comments consequent
2913 on this.
2914
2915 2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
2916 for recursive directory scanning broke on some systems because the files
2917 are not scanned in any specific order and on different systems the order
2918 was different. A call to "sort" has been inserted into RunGrepTest for the
2919 approprate test as a short-term fix. In the longer term there may be an
2920 alternative.
2921
2922 3. I had an email from Eric Raymond about problems translating some of PCRE's
2923 man pages to HTML (despite the fact that I distribute HTML pages, some
2924 people do their own conversions for various reasons). The problems
2925 concerned the use of low-level troff macros .br and .in. I have therefore
2926 removed all such uses from the man pages (some were redundant, some could
2927 be replaced by .nf/.fi pairs). The 132html script that I use to generate
2928 HTML has been updated to handle .nf/.fi and to complain if it encounters
2929 .br or .in.
2930
2931 4. Updated comments in configure.ac that get placed in config.h.in and also
2932 arranged for config.h to be included in the distribution, with the name
2933 config.h.generic, for the benefit of those who have to compile without
2934 Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
2935
2936 5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
2937 Weber: (1) pcre_internal.h was missing some function renames; (2) updated
2938 makevp.bat for the current PCRE, using the additional files
2939 makevp_c.txt, makevp_l.txt, and pcregexp.pas.
2940
2941 6. A Windows user reported a minor discrepancy with test 2, which turned out
2942 to be caused by a trailing space on an input line that had got lost in his
2943 copy. The trailing space was an accident, so I've just removed it.
2944
2945 7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
2946 that is needed.
2947
2948 8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
2949 as "const" (a) because they are and (b) because it helps the PHP
2950 maintainers who have recently made a script to detect big data structures
2951 in the php code that should be moved to the .rodata section. I remembered
2952 to update Builducptable as well, so it won't revert if ucptable.h is ever
2953 re-created.
2954
2955 9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
2956 pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
2957 order to be able to cut out the UTF-8 tables in the latter when UTF-8
2958 support is not required. This saves 1.5-2K of code, which is important in
2959 some applications.
2960
2961 Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
2962 so as not to refer to the tables, even though these functions will never be
2963 called when UTF-8 support is disabled. Otherwise there are problems with a
2964 shared library.
2965
2966 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
2967
2968 (a) It was defining its arguments as char * instead of void *.
2969
2970 (b) It was assuming that all moves were upwards in memory; this was true
2971 a long time ago when I wrote it, but is no longer the case.
2972
2973 The emulated memove() is provided for those environments that have neither
2974 memmove() nor bcopy(). I didn't think anyone used it these days, but that
2975 is clearly not the case, as these two bugs were recently reported.
2976
2977 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
2978 and Detrail to create the HTML documentation, the .txt form of the man
2979 pages, and it removes trailing spaces from listed files. It also creates
2980 pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
2981 case, it wraps all the #defines with #ifndefs. This script should be run
2982 before "make dist".
2983
2984 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
2985 with Unicode property support.
2986
2987 (a) For a maximizing quantifier, if the two different cases of the
2988 character were of different lengths in their UTF-8 codings (there are
2989 some cases like this - I found 11), and the matching function had to
2990 back up over a mixture of the two cases, it incorrectly assumed they
2991 were both the same length.
2992
2993 (b) When PCRE was configured to use the heap rather than the stack for
2994 recursion during matching, it was not correctly preserving the data for
2995 the other case of a UTF-8 character when checking ahead for a match
2996 while processing a minimizing repeat. If the check also involved
2997 matching a wide character, but failed, corruption could cause an
2998 erroneous result when trying to check for a repeat of the original
2999 character.
3000
3001 13. Some tidying changes to the testing mechanism:
3002
3003 (a) The RunTest script now detects the internal link size and whether there
3004 is UTF-8 and UCP support by running ./pcretest -C instead of relying on
3005 values substituted by "configure". (The RunGrepTest script already did
3006 this for UTF-8.) The configure.ac script no longer substitutes the
3007 relevant variables.
3008
3009 (b) The debugging options /B and /D in pcretest show the compiled bytecode
3010 with length and offset values. This means that the output is different
3011 for different internal link sizes. Test 2 is skipped for link sizes
3012 other than 2 because of this, bypassing the problem. Unfortunately,
3013 there was also a test in test 3 (the locale tests) that used /B and
3014 failed for link sizes other than 2. Rather than cut the whole test out,
3015 I have added a new /Z option to pcretest that replaces the length and
3016 offset values with spaces. This is now used to make test 3 independent
3017 of link size. (Test 2 will be tidied up later.)
3018
3019 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
3020 segmentation fault instead of returning the appropriate error message.
3021
3022 15. In multiline mode when the newline sequence was set to "any", the pattern
3023 ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
3024 This doesn't seem right; it now treats the CRLF combination as the line
3025 ending, and so does not match in that case. It's only a pattern such as ^$
3026 that would hit this one: something like ^ABC$ would have failed after \r
3027 and then tried again after \r\n.
3028
3029 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
3030 in an attempt to make files that differ only in their line terminators
3031 compare equal. This works on Linux.
3032
3033 17. Under certain error circumstances pcregrep might try to free random memory
3034 as it exited. This is now fixed, thanks to valgrind.
3035
3036 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
3037 "abc\r\n\r\n", it found an unwanted second match after the second \r. This
3038 was because its rules for how to advance for /g after matching an empty
3039 string at the end of a line did not allow for this case. They now check for
3040 it specially.
3041
3042 20. pcretest is supposed to handle patterns and data of any length, by
3043 extending its buffers when necessary. It was getting this wrong when the
3044 buffer for a data line had to be extended.
3045
3046 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
3047 CRLF as a newline sequence.
3048
3049 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
3050 out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
3051 I have nevertheless tidied it up.
3052
3053 23. Added some casts to kill warnings from HP-UX ia64 compiler.
3054
3055 24. Added a man page for pcre-config.
3056
3057
3058 Version 7.0 19-Dec-06
3059 ---------------------
3060
3061 1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
3062 moving to gcc 4.1.1.
3063
3064 2. The -S option for pcretest uses setrlimit(); I had omitted to #include
3065 sys/time.h, which is documented as needed for this function. It doesn't
3066 seem to matter on Linux, but it showed up on some releases of OS X.
3067
3068 3. It seems that there are systems where bytes whose values are greater than
3069 127 match isprint() in the "C" locale. The "C" locale should be the
3070 default when a C program starts up. In most systems, only ASCII printing
3071 characters match isprint(). This difference caused the output from pcretest
3072 to vary, making some of the tests fail. I have changed pcretest so that:
3073
3074 (a) When it is outputting text in the compiled version of a pattern, bytes
3075 other than 32-126 are always shown as hex escapes.
3076
3077 (b) When it is outputting text that is a matched part of a subject string,
3078 it does the same, unless a different locale has been set for the match
3079 (using the /L modifier). In this case, it uses isprint() to decide.
3080
3081 4. Fixed a major bug that caused incorrect computation of the amount of memory
3082 required for a compiled pattern when options that changed within the
3083 pattern affected the logic of the preliminary scan that determines the
3084 length. The relevant options are -x, and -i in UTF-8 mode. The result was
3085 that the computed length was too small. The symptoms of this bug were
3086 either the PCRE error "internal error: code overflow" from pcre_compile(),
3087 or a glibc crash with a message such as "pcretest: free(): invalid next
3088 size (fast)". Examples of patterns that provoked this bug (shown in
3089 pcretest format) are:
3090
3091 /(?-x: )/x
3092 /(?x)(?-x: \s*#\s*)/
3093 /((?i)[\x{c0}])/8
3094 /(?i:[\x{c0}])/8
3095
3096 HOWEVER: Change 17 below makes this fix obsolete as the memory computation
3097 is now done differently.
3098
3099 5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
3100 wrapper classes; (b) implement a new function in the C++ scanner that is
3101 more efficient than the old way of doing things because it avoids levels of
3102 recursion in the regex matching; (c) add a paragraph to the documentation
3103 for the FullMatch() function.
3104
3105 6. The escape sequence \n was being treated as whatever was defined as
3106 "newline". Not only was this contrary to the documentation, which states
3107 that \n is character 10 (hex 0A), but it also went horribly wrong when
3108 "newline" was defined as CRLF. This has been fixed.
3109
3110 7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
3111 was being set to -1 for the "end of line" case (supposedly a value that no
3112 character can have). Though this value is never used (the check for end of
3113 line is "zero bytes in current character"), it caused compiler complaints.
3114 I've changed it to 0xffffffff.
3115
3116 8. In pcre_version.c, the version string was being built by a sequence of
3117 C macros that, in the event of PCRE_PRERELEASE being defined as an empty
3118 string (as it is for production releases) called a macro with an empty
3119 argument. The C standard says the result of this is undefined. The gcc
3120 compiler treats it as an empty string (which was what was wanted) but it is
3121 reported that Visual C gives an error. The source has been hacked around to
3122 avoid this problem.
3123
3124 9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
3125 builds of pcretest, and changed the call to _setmode() to use _O_BINARY
3126 instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
3127 of them did).
3128
3129 10. Originally, pcretest opened its input and output without "b"; then I was
3130 told that "b" was needed in some environments, so it was added for release
3131 5.0 to both the input and output. (It makes no difference on Unix-like
3132 systems.) Later I was told that it is wrong for the input on Windows. I've
3133 now abstracted the modes into two macros, to make it easier to fiddle with
3134 them, and removed "b" from the input mode under Windows.
3135
3136 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
3137
3138 12. Added -help and --help to pcretest as an official way of being reminded
3139 of the options.
3140
3141 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
3142 and pcrecpp.cc because they annoy compilers at high warning levels.
3143
3144 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
3145
3146 15. Fixed an occurrence of == in configure.ac that should have been = (shell
3147 scripts are not C programs :-) and which was not noticed because it works
3148 on Linux.
3149
3150 16. pcretest is supposed to handle any length of pattern and data line (as one
3151 line or as a continued sequence of lines) by extending its input buffer if
3152 necessary. This feature was broken for very long pattern lines, leading to
3153 a string of junk being passed to pcre_compile() if the pattern was longer
3154 than about 50K.
3155
3156 17. I have done a major re-factoring of the way pcre_compile() computes the
3157 amount of memory needed for a compiled pattern. Previously, there was code
3158 that made a preliminary scan of the pattern in order to do this. That was
3159 OK when PCRE was new, but as the facilities have expanded, it has become
3160 harder and harder to keep it in step with the real compile phase, and there
3161 have been a number of bugs (see for example, 4 above). I have now found a
3162 cunning way of running the real compile function in a "fake" mode that
3163 enables it to compute how much memory it would need, while actually only
3164 ever using a few hundred bytes of working memory and without too many
3165 tests of the mode. This should make future maintenance and development
3166 easier. A side effect of this work is that the limit of 200 on the nesting
3167 depth of parentheses has been removed (though this was never a serious
3168 limitation, I suspect). However, there is a downside: pcre_compile() now
3169 runs more slowly than before (30% or more, depending on the pattern). I
3170 hope this isn't a big issue. There is no effect on runtime performance.
3171
3172 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
3173 newline (only possible for the last line of a file) and it was a
3174 pattern that set a locale (followed by /Lsomething), pcretest crashed.
3175
3176 19. Added additional timing features to pcretest. (1) The -tm option now times
3177 matching only, not compiling. (2) Both -t and -tm can be followed, as a
3178 separate command line item, by a number that specifies the number of
3179 repeats to use when timing. The default is 50000; this gives better
3180 precision, but takes uncomfortably long for very large patterns.
3181
3182 20. Extended pcre_study() to be more clever in cases where a branch of a
3183 subpattern has no definite first character. For example, (a*|b*)[cd] would
3184 previously give no result from pcre_study(). Now it recognizes that the
3185 first character must be a, b, c, or d.
3186
3187 21. There was an incorrect error "recursive call could loop indefinitely" if
3188 a subpattern (or the entire pattern) that was being tested for matching an
3189 empty string contained only one non-empty item after a nested subpattern.
3190 For example, the pattern (?>\x{100}*)\d(?R) provoked this error
3191 incorrectly, because the \d was being skipped in the check.
3192
3193 22. The pcretest program now has a new pattern option /B and a command line
3194 option -b, which is equivalent to adding /B to every pattern. This causes
3195 it to show the compiled bytecode, without the additional information that
3196 -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
3197 is the same as /B/I).
3198
3199 23. A new optimization is now able automatically to treat some sequences such
3200 as a*b as a*+b. More specifically, if something simple (such as a character
3201 or a simple class like \d) has an unlimited quantifier, and is followed by
3202 something that cannot possibly match the quantified thing, the quantifier
3203 is automatically "possessified".
3204
3205 24. A recursive reference to a subpattern whose number was greater than 39
3206 went wrong under certain circumstances in UTF-8 mode. This bug could also
3207 have affected the operation of pcre_study().
3208
3209 25. Realized that a little bit of performance could be had by replacing
3210 (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
3211
3212 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
3213
3214 27. Possessive quantifiers such as a++ were previously implemented by turning
3215 them into atomic groups such as ($>a+). Now they have their own opcodes,
3216 which improves performance. This includes the automatically created ones
3217 from 23 above.
3218
3219 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
3220 lookahead was broken if it was not anchored. PCRE was mistakenly expecting
3221 the first matched character to be a colon. This applied both to named and
3222 numbered groups.
3223
3224 29. The ucpinternal.h header file was missing its idempotency #ifdef.
3225
3226 30. I was sent a "project" file called libpcre.a.dev which I understand makes
3227 building PCRE on Windows easier, so I have included it in the distribution.
3228
3229 31. There is now a check in pcretest against a ridiculously large number being
3230 returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
3231 loop, the loop is abandoned.
3232
3233 32. Forward references to subpatterns in conditions such as (?(2)...) where
3234 subpattern 2 is defined later cause pcre_compile() to search forwards in
3235 the pattern for the relevant set of parentheses. This search went wrong
3236 when there were unescaped parentheses in a character class, parentheses
3237 escaped with \Q...\E, or parentheses in a #-comment in /x mode.
3238
3239 33. "Subroutine" calls and backreferences were previously restricted to
3240 referencing subpatterns earlier in the regex. This restriction has now
3241 been removed.
3242
3243 34. Added a number of extra features that are going to be in Perl 5.10. On the
3244 whole, these are just syntactic alternatives for features that PCRE had
3245 previously implemented using the Python syntax or my own invention. The
3246 other formats are all retained for compatibility.
3247
3248 (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
3249 as (?P<name>...). The new forms, as well as being in Perl 5.10, are
3250 also .NET compatible.
3251
3252 (b) A recursion or subroutine call to a named group can now be defined as
3253 (?&name) as well as (?P>name).
3254
3255 (c) A backreference to a named group can now be defined as \k<name> or
3256 \k'name' as well as (?P=name). The new forms, as well as being in Perl
3257 5.10, are also .NET compatible.
3258
3259 (d) A conditional reference to a named group can now use the syntax
3260 (?(<name>) or (?('name') as well as (?(name).
3261
3262 (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
3263 groups (named and numbered) that are never evaluated inline, but can be
3264 called as "subroutines" from elsewhere. In effect, the DEFINE condition
3265 is always false. There may be only one alternative in such a group.
3266
3267 (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
3268 as the simple (?(R). The condition is true only if the most recent
3269 recursion is that of the given number or name. It does not search out
3270 through the entire recursion stack.
3271
3272 (g) The escape \gN or \g{N} has been added, where N is a positive or
3273 negative number, specifying an absolute or relative reference.
3274
3275 35. Tidied to get rid of some further signed/unsigned compiler warnings and
3276 some "unreachable code" warnings.
3277
3278 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
3279 things, this adds five new scripts.
3280
3281 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
3282 There were also incompatibilities regarding the handling of \Q..\E inside
3283 character classes, for example with patterns like [\Qa\E-\Qz\E] where the
3284 hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
3285
3286 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
3287 matches an empty string, and forcibly breaks the loop. There were bugs in
3288 this code in non-simple cases. For a pattern such as ^(a()*)* matched
3289 against aaaa the result was just "a" rather than "aaaa", for example. Two
3290 separate and independent bugs (that affected different cases) have been
3291 fixed.
3292
3293 39. Refactored the code to abolish the use of different opcodes for small
3294 capturing bracket numbers. This is a tidy that I avoided doing when I
3295 removed the limit on the number of capturing brackets for 3.5 back in 2001.
3296 The new approach is not only tidier, it makes it possible to reduce the
3297 memory needed to fix the previous bug (38).
3298
3299 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
3300 sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
3301 processing dot, circumflex, or dollar metacharacters, or #-comments in /x
3302 mode.
3303
3304 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
3305 report.
3306
3307 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
3308 copy construction and assignment in the C++ wrapper.
3309
3310 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
3311 couple of bugs that could have given wrong results in the "--newline=crlf"
3312 case.
3313
3314 44. Added a number of casts and did some reorganization of signed/unsigned int
3315 variables following suggestions from Dair Grant. Also renamed the variable
3316 "this" as "item" because it is a C++ keyword.
3317
3318 45. Arranged for dftables to add
3319
3320 #include "pcre_internal.h"
3321
3322 to pcre_chartables.c because without it, gcc 4.x may remove the array
3323 definition from the final binary if PCRE is built into a static library and
3324 dead code stripping is activated.
3325
3326 46. For an unanchored pattern, if a match attempt fails at the start of a
3327 newline sequence, and the newline setting is CRLF or ANY, and the next two
3328 characters are CRLF, advance by two characters instead of one.
3329
3330
3331 Version 6.7 04-Jul-06
3332 ---------------------
3333
3334 1. In order to handle tests when input lines are enormously long, pcretest has
3335 been re-factored so that it automatically extends its buffers when
3336 necessary. The code is crude, but this _is_ just a test program. The
3337 default size has been increased from 32K to 50K.
3338
3339 2. The code in pcre_study() was using the value of the re argument before
3340 testing it for NULL. (Of course, in any sensible call of the function, it
3341 won't be NULL.)
3342
3343 3. The memmove() emulation function in pcre_internal.h, which is used on
3344 systems that lack both memmove() and bcopy() - that is, hardly ever -
3345 was missing a "static" storage class specifier.
3346
3347 4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
3348 containing an extended class (one that cannot be represented by a bitmap
3349 because it contains high-valued characters or Unicode property items, e.g.
3350 [\pZ]). Almost always one would set UTF-8 mode when processing such a
3351 pattern, but PCRE should not loop if you do not (it no longer does).
3352 [Detail: two cases were found: (a) a repeated subpattern containing an
3353 extended class; (b) a recursive reference to a subpattern that followed a
3354 previous extended class. It wasn't skipping over the extended class
3355 correctly when UTF-8 mode was not set.]
3356
3357 5. A negated single-character class was not being recognized as fixed-length
3358 in lookbehind assertions such as (?<=[^f]), leading to an incorrect
3359 compile error "lookbehind assertion is not fixed length".
3360
3361 6. The RunPerlTest auxiliary script was showing an unexpected difference
3362 between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
3363 write a Perl script that can interpret lines of an input file either as
3364 byte characters or as UTF-8, which is what "perltest" was being required to
3365 do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
3366 can't do is switch easily at run time between having the "use utf8;" pragma
3367 or not. In the end, I fudged it by using the RunPerlTest script to insert
3368 "use utf8;" explicitly for the UTF-8 tests.
3369
3370 7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
3371 the end of the subject string, contrary to the documentation and to what
3372 Perl does. This was true of both matching functions. Now it matches only at
3373 the start of the subject and immediately after *internal* newlines.
3374
3375 8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
3376 a pointer to an int instead of a pointer to an unsigned long int. This
3377 caused problems on 64-bit systems.
3378
3379 9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
3380 instance of the 'standard' template library not being so standard".
3381
3382 10. There was no check on the number of named subpatterns nor the maximum
3383 length of a subpattern name. The product of these values is used to compute
3384 the size of the memory block for a compiled pattern. By supplying a very
3385 long subpattern name and a large number of named subpatterns, the size
3386 computation could be caused to overflow. This is now prevented by limiting
3387 the length of names to 32 characters, and the number of named subpatterns
3388 to 10,000.
3389
3390 11. Subpatterns that are repeated with specific counts have to be replicated in
3391 the compiled pattern. The size of memory for this was computed from the
3392 length of the subpattern and the repeat count. The latter is limited to
3393 65535, but there was no limit on the former, meaning that integer overflow
3394 could in principle occur. The compiled length of a repeated subpattern is
3395 now limited to 30,000 bytes in order to prevent this.
3396
3397 12. Added the optional facility to have named substrings with the same name.
3398
3399 13. Added the ability to use a named substring as a condition, using the
3400 Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
3401 are numbers (not recommended). Forward references are permitted.
3402
3403 14. Added forward references in named backreferences (if you see what I mean).
3404
3405 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
3406 pattern could run off the end of the subject. For example, the pattern
3407 "(?s)(.{1,5})"8 did this with the subject "ab".
3408
3409 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
3410 PCRE_CASELESS was set when matching characters that were quantified with ?
3411 or *.
3412
3413 17. A character class other than a single negated character that had a minimum
3414 but no maximum quantifier - for example [ab]{6,} - was not handled
3415 correctly by pce_dfa_exec(). It would match only one character.
3416
3417 18. A valid (though odd) pattern that looked like a POSIX character
3418 class but used an invalid character after [ (for example [[,abc,]]) caused
3419 pcre_compile() to give the error "Failed: internal error: code overflow" or
3420 in some cases to crash with a glibc free() error. This could even happen if
3421 the pattern terminated after [[ but there just happened to be a sequence of
3422 letters, a binary zero, and a closing ] in the memory that followed.
3423
3424 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
3425 over the years. Originally (before any Unicode support), just the bottom 8
3426 bits were taken. Thus, for example, \500 really meant \100. Nowadays the
3427 output from "man perlunicode" includes this:
3428
3429 The regular expression compiler produces polymorphic opcodes. That
3430 is, the pattern adapts to the data and automatically switches to
3431 the Unicode character scheme when presented with Unicode data--or
3432 instead uses a traditional byte scheme when presented with byte
3433 data.
3434
3435 Sadly, a wide octal escape does not cause a switch, and in a string with
3436 no other multibyte characters, these octal escapes are treated as before.
3437 Thus, in Perl, the pattern /\500/ actually matches \100 but the pattern
3438 /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
3439 Unicode string.
3440
3441 I have not perpetrated such confusion in PCRE. Up till now, it took just
3442 the bottom 8 bits, as in old Perl. I have now made octal escapes with
3443 values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
3444 translate to the appropriate multibyte character.
3445
3446 29. Applied some refactoring to reduce the number of warnings from Microsoft
3447 and Borland compilers. This has included removing the fudge introduced
3448 seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
3449 a warning about an unused variable.
3450
3451 21. PCRE has not included VT (character 0x0b) in the set of whitespace
3452 characters since release 4.0, because Perl (from release 5.004) does not.
3453 [Or at least, is documented not to: some releases seem to be in conflict
3454 with the documentation.] However, when a pattern was studied with
3455 pcre_study() and all its branches started with \s, PCRE still included VT
3456 as a possible starting character. Of course, this did no harm; it just
3457 caused an unnecessary match attempt.
3458
3459 22. Removed a now-redundant internal flag bit that recorded the fact that case
3460 dependency changed within the pattern. This was once needed for "required
3461 byte" processing, but is no longer used. This recovers a now-scarce options
3462 bit. Also moved the least significant internal flag bit to the most-
3463 significant bit of the word, which was not previously used (hangover from
3464 the days when it was an int rather than a uint) to free up another bit for
3465 the future.
3466
3467 23. Added support for CRLF line endings as well as CR and LF. As well as the
3468 default being selectable at build time, it can now be changed at runtime
3469 via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
3470 specify that it is scanning data with non-default line endings.
3471
3472 24. Changed the definition of CXXLINK to make it agree with the definition of
3473 LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
3474
3475 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
3476 recursions. This makes a big different to stack usage for some patterns.
3477
3478 26. If a subpattern containing a named recursion or subroutine reference such
3479 as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
3480 the space required for the compiled pattern went wrong and gave too small a
3481 value. Depending on the environment, this could lead to "Failed: internal
3482 error: code overflow at offset 49" or "glibc detected double free or
3483 corruption" errors.
3484
3485 27. Applied patches from Google (a) to support the new newline modes and (b) to
3486 advance over multibyte UTF-8 characters in GlobalReplace.
3487
3488 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
3489 difference for some implementation of PCRE in some Windows version.
3490
3491 29. Added some extra testing facilities to pcretest:
3492
3493 \q<number> in a data line sets the "match limit" value
3494 \Q<number> in a data line sets the "match recursion limt" value
3495 -S <number> sets the stack size, where <number> is in megabytes
3496
3497 The -S option isn't available for Windows.
3498
3499
3500 Version 6.6 06-Feb-06
3501 ---------------------
3502
3503 1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
3504 in pcreposix.h. I have copied the definition from pcre.h.
3505
3506 2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
3507 because pcre.h is no longer a built file.
3508
3509 3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
3510 not normally included in the compiled code.
3511
3512
3513 Version 6.5 01-Feb-06
3514 ---------------------
3515
3516 1. When using the partial match feature with pcre_dfa_exec(), it was not
3517 anchoring the second and subsequent partial matches at the new starting
3518 point. This could lead to incorrect results. For example, with the pattern
3519 /1234/, partially matching against "123" and then "a4" gave a match.
3520
3521 2. Changes to pcregrep:
3522
3523 (a) All non-match returns from pcre_exec() were being treated as failures
3524 to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
3525 error message is output. Some extra information is given for the
3526 PCRE_ERROR_MATCHLIMIT and PCRE_ERROR_RECURSIONLIMIT errors, which are
3527 probably the only errors that are likely to be caused by users (by
3528 specifying a regex that has nested indefinite repeats, for instance).
3529 If there are more than 20 of these errors, pcregrep is abandoned.
3530
3531 (b) A binary zero was treated as data while matching, but terminated the
3532 output line if it was written out. This has been fixed: binary zeroes
3533 are now no different to any other data bytes.
3534
3535 (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
3536 used to set a locale for matching. The --locale=xxxx long option has
3537 been added (no short equivalent) to specify a locale explicitly on the
3538 pcregrep command, overriding the environment variables.
3539
3540 (d) When -B was used with -n, some line numbers in the output were one less
3541 than they should have been.
3542
3543 (e) Added the -o (--only-matching) option.
3544
3545 (f) If -A or -C was used with -c (count only), some lines of context were
3546 accidentally printed for the final match.
3547
3548 (g) Added the -H (--with-filename) option.
3549
3550 (h) The combination of options -rh failed to suppress file names for files
3551 that were found from directory arguments.
3552
3553 (i) Added the -D (--devices) and -d (--directories) options.
3554
3555 (j) Added the -F (--fixed-strings) option.
3556
3557 (k) Allow "-" to be used as a file name for -f as well as for a data file.
3558
3559 (l) Added the --colo(u)r option.
3560
3561 (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
3562 is not present by default.
3563
3564 3. A nasty bug was discovered in the handling of recursive patterns, that is,
3565 items such as (?R) or (?1), when the recursion could match a number of
3566 alternatives. If it matched one of the alternatives, but subsequently,
3567 outside the recursion, there was a failure, the code tried to back up into
3568 the recursion. However, because of the way PCRE is implemented, this is not
3569 possible, and the result was an incorrect result from the match.
3570
3571 In order to prevent this happening, the specification of recursion has
3572 been changed so that all such subpatterns are automatically treated as
3573 atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
3574
3575 4. I had overlooked the fact that, in some locales, there are characters for
3576 which isalpha() is true but neither isupper() nor islower() are true. In
3577 the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
3578 and ordfeminine) are like this. This affected the treatment of \w and \W
3579 when they appeared in character classes, but not when they appeared outside
3580 a character class. The bit map for "word" characters is now created
3581 separately from the results of isalnum() instead of just taking it from the
3582 upper, lower, and digit maps. (Plus the underscore character, of course.)
3583
3584 5. The above bug also affected the handling of POSIX character classes such as
3585 [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
3586 permanent tables. Instead, the bit maps for such a class were previously
3587 created as the appropriate unions of the upper, lower, and digit bitmaps.
3588 Now they are created by subtraction from the [[:word:]] class, which has
3589 its own bitmap.
3590
3591 6. The [[:blank:]] character class matches horizontal, but not vertical space.
3592 It is created by subtracting the vertical space characters (\x09, \x0a,
3593 \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
3594 subtraction was done in the overall bitmap for a character class, meaning
3595 that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
3596 be recognized. This bug has been fixed.
3597
3598 7. Patches from the folks at Google:
3599
3600 (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
3601 real life, but is still worth protecting against".
3602
3603 (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
3604 regular expressions".
3605
3606 (c) pcre_scanner.cc: avoid use of std::count() because not all systems
3607 have it.
3608
3609 (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
3610 "configure" and the latter not, in order to fix a problem somebody had
3611 with compiling the Arg class on HP-UX.
3612
3613 (e) Improve the error-handling of the C++ wrapper a little bit.
3614
3615 (f) New tests for checking recursion limiting.
3616
3617 8. The pcre_memmove() function, which is used only if the environment does not
3618 have a standard memmove() function (and is therefore rarely compiled),
3619 contained two bugs: (a) use of int instead of size_t, and (b) it was not
3620 returning a result (though PCRE never actually uses the result).
3621
3622 9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
3623 large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
3624 returned instead of calling malloc() with an overflowing number that would
3625 most likely cause subsequent chaos.
3626
3627 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
3628
3629 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
3630 with this option is matched, the nmatch and pmatch options of regexec() are
3631 ignored.
3632
3633 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
3634 provided in case anyone wants to the the POSIX interface with UTF-8
3635 strings.
3636
3637 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
3638 C++ linking (needed for some HP-UX environments).
3639
3640 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
3641 (unused parameter) and in the pcre_printint() function (omitted "default"
3642 switch label when the default is to do nothing).
3643
3644 15. Added some code to make it possible, when PCRE is compiled as a C++
3645 library, to replace subject pointers for pcre_exec() with a smart pointer
3646 class, thus making it possible to process discontinuous strings.
3647
3648 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
3649 much the same function. They were added by different people who were trying
3650 to make PCRE easy to compile on non-Unix systems. It has been suggested
3651 that PCRE_EXPORT be abolished now that there is more automatic apparatus
3652 for compiling on Windows systems. I have therefore replaced it with
3653 PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
3654 defaults to "extern" for C or "extern C" for C++, which works fine on
3655 Unix-like systems. It is now possible to override the value of PCRE_DATA_
3656 SCOPE with something explicit in config.h. In addition:
3657
3658 (a) pcreposix.h still had just "extern" instead of either of these macros;
3659 I have replaced it with PCRE_DATA_SCOPE.
3660
3661 (b) Functions such as _pcre_xclass(), which are internal to the library,
3662 but external in the C sense, all had PCRE_EXPORT in their definitions.
3663 This is apparently wrong for the Windows case, so I have removed it.
3664 (It makes no difference on Unix-like systems.)
3665
3666 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
3667 of recursive calls to match(). This is different to MATCH_LIMIT because
3668 that limits the total number of calls to match(), not all of which increase
3669 the depth of recursion. Limiting the recursion depth limits the amount of
3670 stack (or heap if NO_RECURSE is set) that is used. The default can be set
3671 when PCRE is compiled, and changed at run time. A patch from Google adds
3672 this functionality to the C++ interface.
3673
3674 18. Changes to the handling of Unicode character properties:
3675
3676 (a) Updated the table to Unicode 4.1.0.
3677
3678 (b) Recognize characters that are not in the table as "Cn" (undefined).
3679
3680 (c) I revised the way the table is implemented to a much improved format
3681 which includes recognition of ranges. It now supports the ranges that
3682 are defined in UnicodeData.txt, and it also amalgamates other
3683 characters into ranges. This has reduced the number of entries in the
3684 table from around 16,000 to around 3,000, thus reducing its size
3685 considerably. I realized I did not need to use a tree structure after
3686 all - a binary chop search is just as efficient. Having reduced the
3687 number of entries, I extended their size from 6 bytes to 8 bytes to
3688 allow for more data.
3689
3690 (d) Added support for Unicode script names via properties such as \p{Han}.
3691
3692 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
3693 matching that character.
3694
3695 20. When matching a repeated Unicode property with a minimum greater than zero,
3696 (for example \pL{2,}), PCRE could look past the end of the subject if it
3697 reached it while seeking the minimum number of characters. This could
3698 happen only if some of the characters were more than one byte long, because
3699 there is a check for at least the minimum number of bytes.
3700
3701 21. Refactored the implementation of \p and \P so as to be more general, to
3702 allow for more different types of property in future. This has changed the
3703 compiled form incompatibly. Anybody with saved compiled patterns that use
3704 \p or \P will have to recompile them.
3705
3706 22. Added "Any" and "L&" to the supported property types.
3707
3708 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
3709 but give a compile time error if the value is greater than 0xff.
3710
3711 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
3712 accidentally not being installed or uninstalled.
3713
3714 25. The pcre.h file was built from pcre.h.in, but the only changes that were
3715 made were to insert the current release number. This seemed silly, because
3716 it made things harder for people building PCRE on systems that don't run
3717 "configure". I have turned pcre.h into a distributed file, no longer built
3718 by "configure", with the version identification directly included. There is
3719 no longer a pcre.h.in file.
3720
3721 However, this change necessitated a change to the pcre-config script as
3722 well. It is built from pcre-config.in, and one of the substitutions was the
3723 release number. I have updated configure.ac so that ./configure now finds
3724 the release number by grepping pcre.h.
3725
3726 26. Added the ability to run the tests under valgrind.
3727
3728
3729 Version 6.4 05-Sep-05
3730 ---------------------
3731
3732 1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
3733 "--" to be printed when multiple files were scanned, even when none of the
3734 -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
3735 consider it to be a bug, and have restored the previous behaviour.
3736
3737 2. A couple of code tidies to get rid of compiler warnings.
3738
3739 3. The pcretest program used to cheat by referring to symbols in the library
3740 whose names begin with _pcre_. These are internal symbols that are not
3741 really supposed to be visible externally, and in some environments it is
3742 possible to suppress them. The cheating is now confined to including
3743 certain files from the library's source, which is a bit cleaner.
3744
3745 4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
3746 file's purpose clearer.
3747
3748 5. Reorganized pcre_ucp_findchar().
3749
3750
3751 Version 6.3 15-Aug-05
3752 ---------------------
3753
3754 1. The file libpcre.pc.in did not have general read permission in the tarball.
3755
3756 2. There were some problems when building without C++ support:
3757
3758 (a) If C++ support was not built, "make install" and "make test" still
3759 tried to test it.
3760
3761 (b) There were problems when the value of CXX was explicitly set. Some
3762 changes have been made to try to fix these, and ...
3763
3764 (c) --disable-cpp can now be used to explicitly disable C++ support.
3765
3766 (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
3767 backslash in a target when C++ was disabled. This confuses some
3768 versions of "make", apparently. Using an intermediate variable solves
3769 this. (Same for CPP_LOBJ.)
3770
3771 3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
3772 (non-Windows) now includes $(CFLAGS) because these flags are sometimes
3773 necessary on certain architectures.
3774
3775 4. Added a setting of -export-symbols-regex to the link command to remove
3776 those symbols that are exported in the C sense, but actually are local
3777 within the library, and not documented. Their names all begin with
3778 "_pcre_". This is not a perfect job, because (a) we have to except some
3779 symbols that pcretest ("illegally") uses, and (b) the facility isn't always
3780 available (and never for static libraries). I have made a note to try to
3781 find a way round (a) in the future.
3782
3783
3784 Version 6.2 01-Aug-05
3785 ---------------------
3786
3787 1. There was no test for integer overflow of quantifier values. A construction
3788 such as {1111111111111111} would give undefined results. What is worse, if
3789 a minimum quantifier for a parenthesized subpattern overflowed and became
3790 negative, the calculation of the memory size went wrong. This could have
3791 led to memory overwriting.
3792
3793 2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
3794
3795 3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
3796 operating environments where this matters.
3797
3798 4. Applied Giuseppe Maxia's patch to add additional features for controlling
3799 PCRE options from within the C++ wrapper.
3800
3801 5. Named capturing subpatterns were not being correctly counted when a pattern
3802 was compiled. This caused two problems: (a) If there were more than 100
3803 such subpatterns, the calculation of the memory needed for the whole
3804 compiled pattern went wrong, leading to an overflow error. (b) Numerical
3805 back references of the form \12, where the number was greater than 9, were
3806 not recognized as back references, even though there were sufficient
3807 previous subpatterns.
3808
3809 6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
3810 versions of gcc, e.g. 2.95.4.
3811
3812
3813 Version 6.1 21-Jun-05
3814 ---------------------
3815
3816 1. There was one reference to the variable "posix" in pcretest.c that was not
3817 surrounded by "#if !defined NOPOSIX".
3818
3819 2. Make it possible to compile pcretest without DFA support, UTF8 support, or
3820 the cross-check on the old pcre_info() function, for the benefit of the
3821 cut-down version of PCRE that is currently imported into Exim.
3822
3823 3. A (silly) pattern starting with (?i)(?-i) caused an internal space
3824 allocation error. I've done the easy fix, which wastes 2 bytes for sensible
3825 patterns that start (?i) but I don't think that matters. The use of (?i) is
3826 just an example; this all applies to the other options as well.
3827
3828 4. Since libtool seems to echo the compile commands it is issuing, the output
3829 from "make" can be reduced a bit by putting "@" in front of each libtool
3830 compile command.
3831
3832 5. Patch from the folks at Google for configure.in to be a bit more thorough
3833 in checking for a suitable C++ installation before trying to compile the
3834 C++ stuff. This should fix a reported problem when a compiler was present,
3835 but no suitable headers.
3836
3837 6. The man pages all had just "PCRE" as their title. I have changed them to
3838 be the relevant file name. I have also arranged that these names are
3839 retained in the file doc/pcre.txt, which is a concatenation in text format
3840 of all the man pages except the little individual ones for each function.
3841
3842 7. The NON-UNIX-USE file had not been updated for the different set of source
3843 files that come with release 6. I also added a few comments about the C++
3844 wrapper.
3845
3846
3847 Version 6.0 07-Jun-05
3848 ---------------------
3849
3850 1. Some minor internal re-organization to help with my DFA experiments.
3851
3852 2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
3853 didn't matter for the library itself when fully configured, but did matter
3854 when compiling without UCP support, or within Exim, where the ucp files are
3855 not imported.
3856
3857 3. Refactoring of the library code to split up the various functions into
3858 different source modules. The addition of the new DFA matching code (see
3859 below) to a single monolithic source would have made it really too
3860 unwieldy, quite apart from causing all the code to be include in a
3861 statically linked application, when only some functions are used. This is
3862 relevant even without the DFA addition now that patterns can be compiled in
3863 one application and matched in another.
3864
3865 The downside of splitting up is that there have to be some external
3866 functions and data tables that are used internally in different modules of
3867 the library but which are not part of the API. These have all had their
3868 names changed to start with "_pcre_" so that they are unlikely to clash
3869 with other external names.
3870
3871 4. Added an alternate matching function, pcre_dfa_exec(), which matches using
3872 a different (DFA) algorithm. Although it is slower than the original
3873 function, it does have some advantages for certain types of matching
3874 problem.
3875
3876 5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
3877 including restarting after a partial match.
3878
3879 6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
3880 defined when compiling for Windows was sent to me. I have put it into the
3881 code, though I have no means of testing or verifying it.
3882
3883 7. Added the pcre_refcount() auxiliary function.
3884
3885 8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
3886 match before or at the first newline in the subject string. In pcretest,
3887 the /f option on a pattern can be used to set this.
3888
3889 9. A repeated \w when used in UTF-8 mode with characters greater than 256
3890 would behave wrongly. This has been present in PCRE since release 4.0.
3891
3892 10. A number of changes to the pcregrep command:
3893
3894 (a) Refactored how -x works; insert ^(...)$ instead of setting
3895 PCRE_ANCHORED and checking the length, in preparation for adding
3896 something similar for -w.
3897
3898 (b) Added the -w (match as a word) option.
3899
3900 (c) Refactored the way lines are read and buffered so as to have more
3901 than one at a time available.
3902
3903 (d) Implemented a pcregrep test script.
3904
3905 (e) Added the -M (multiline match) option. This allows patterns to match
3906 over several lines of the subject. The buffering ensures that at least
3907 8K, or the rest of the document (whichever is the shorter) is available
3908 for matching (and similarly the previous 8K for lookbehind assertions).
3909
3910 (f) Changed the --help output so that it now says
3911
3912 -w, --word-regex(p)
3913
3914 instead of two lines, one with "regex" and the other with "regexp"
3915 because that confused at least one person since the short forms are the
3916 same. (This required a bit of code, as the output is generated
3917 automatically from a table. It wasn't just a text change.)
3918
3919 (g) -- can be used to terminate pcregrep options if the next thing isn't an
3920 option but starts with a hyphen. Could be a pattern or a path name
3921 starting with a hyphen, for instance.
3922
3923 (h) "-" can be given as a file name to represent stdin.
3924
3925 (i) When file names are being printed, "(standard input)" is used for
3926 the standard input, for compatibility with GNU grep. Previously
3927 "<stdin>" was used.
3928
3929 (j) The option --label=xxx can be used to supply a name to be used for
3930 stdin when file names are being printed. There is no short form.
3931
3932 (k) Re-factored the options decoding logic because we are going to add
3933 two more options that take data. Such options can now be given in four
3934 different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
3935
3936 (l) Added the -A, -B, and -C options for requesting that lines of context
3937 around matches be printed.
3938
3939 (m) Added the -L option to print the names of files that do not contain
3940 any matching lines, that is, the complement of -l.
3941
3942 (n) The return code is 2 if any file cannot be opened, but pcregrep does
3943 continue to scan other files.
3944
3945 (o) The -s option was incorrectly implemented. For compatibility with other
3946 greps, it now suppresses the error message for a non-existent or non-
3947 accessible file (but not the return code). There is a new option called
3948 -q that suppresses the output of matching lines, which was what -s was
3949 previously doing.
3950
3951 (p) Added --include and --exclude options to specify files for inclusion
3952 and exclusion when recursing.
3953
3954 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
3955 Hopefully, it now does.
3956
3957 12. Missing cast in pcre_study().
3958
3959 13. Added an "uninstall" target to the makefile.
3960
3961 14. Replaced "extern" in the function prototypes in Makefile.in with
3962 "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
3963 world, but is set differently for Windows.
3964
3965 15. Added a second compiling function called pcre_compile2(). The only
3966 difference is that it has an extra argument, which is a pointer to an
3967 integer error code. When there is a compile-time failure, this is set
3968 non-zero, in addition to the error test pointer being set to point to an
3969 error message. The new argument may be NULL if no error number is required
3970 (but then you may as well call pcre_compile(), which is now just a
3971 wrapper). This facility is provided because some applications need a
3972 numeric error indication, but it has also enabled me to tidy up the way
3973 compile-time errors are handled in the POSIX wrapper.
3974
3975 16. Added VPATH=.libs to the makefile; this should help when building with one
3976 prefix path and installing with another. (Or so I'm told by someone who
3977 knows more about this stuff than I do.)
3978
3979 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
3980 passes PCRE_DOTALL to the pcre_compile() function, making the "." character
3981 match everything, including newlines. This is not POSIX-compatible, but
3982 somebody wanted the feature. From pcretest it can be activated by using
3983 both the P and the s flags.
3984
3985 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
3986
3987 19. libpcre.pc was being incorrectly installed as executable.
3988
3989 20. A couple of places in pcretest check for end-of-line by looking for '\n';
3990 it now also looks for '\r' so that it will work unmodified on Windows.
3991
3992 21. Added Google's contributed C++ wrapper to the distribution.
3993
3994 22. Added some untidy missing memory free() calls in pcretest, to keep
3995 Electric Fence happy when testing.
3996
3997
3998
3999 Version 5.0 13-Sep-04
4000 ---------------------
4001
4002 1. Internal change: literal characters are no longer packed up into items
4003 containing multiple characters in a single byte-string. Each character
4004 is now matched using a separate opcode. However, there may be more than one
4005 byte in the character in UTF-8 mode.
4006
4007 2. The pcre_callout_block structure has two new fields: pattern_position and
4008 next_item_length. These contain the offset in the pattern to the next match
4009 item, and its length, respectively.
4010
4011 3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
4012 insertion of callouts before each pattern item. Added the /C option to
4013 pcretest to make use of this.
4014
4015 4. On the advice of a Windows user, the lines
4016
4017 #if defined(_WIN32) || defined(WIN32)
4018 _setmode( _fileno( stdout ), 0x8000 );
4019 #endif /* defined(_WIN32) || defined(WIN32) */
4020
4021 have been added to the source of pcretest. This apparently does useful
4022 magic in relation to line terminators.
4023
4024 5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
4025 for the benefit of those environments where the "b" makes a difference.
4026
4027 6. The icc compiler has the same options as gcc, but "configure" doesn't seem
4028 to know about it. I have put a hack into configure.in that adds in code
4029 to set GCC=yes if CC=icc. This seems to end up at a point in the
4030 generated configure script that is early enough to affect the setting of
4031 compiler options, which is what is needed, but I have no means of testing
4032 whether it really works. (The user who reported this had patched the
4033 generated configure script, which of course I cannot do.)
4034
4035 LATER: After change 22 below (new libtool files), the configure script
4036 seems to know about icc (and also ecc). Therefore, I have commented out
4037 this hack in configure.in.
4038
4039 7. Added support for pkg-config (2 patches were sent in).
4040
4041 8. Negated POSIX character classes that used a combination of internal tables
4042 were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
4043 [[:^ascii]]. Typically, they would match almost any characters. The other
4044 POSIX classes were not broken in this way.
4045
4046 9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
4047 to find the match, as PCRE was deluded into thinking that the match had to
4048 start at the start point or following a newline. The same bug applied to
4049 patterns with negative forward assertions or any backward assertions
4050 preceding ".*" at the start, unless the pattern required a fixed first
4051 character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
4052
4053 10. In UTF-8 mode, when moving forwards in the subject after a failed match
4054 starting at the last subject character, bytes beyond the end of the subject
4055 string were read.
4056
4057 11. Renamed the variable "class" as "classbits" to make life easier for C++
4058 users. (Previously there was a macro definition, but it apparently wasn't
4059 enough.)
4060
4061 12. Added the new field "tables" to the extra data so that tables can be passed
4062 in at exec time, or the internal tables can be re-selected. This allows
4063 a compiled regex to be saved and re-used at a later time by a different
4064 program that might have everything at different addresses.
4065
4066 13. Modified the pcre-config script so that, when run on Solaris, it shows a
4067 -R library as well as a -L library.
4068
4069 14. The debugging options of pcretest (-d on the command line or D on a
4070 pattern) showed incorrect output for anything following an extended class
4071 that contained multibyte characters and which was followed by a quantifier.
4072
4073 15. Added optional support for general category Unicode character properties
4074 via the \p, \P, and \X escapes. Unicode property support implies UTF-8
4075 support. It adds about 90K to the size of the library. The meanings of the
4076 inbuilt class escapes such as \d and \s have NOT been changed.
4077
4078 16. Updated pcredemo.c to include calls to free() to release the memory for the
4079 compiled pattern.
4080
4081 17. The generated file chartables.c was being created in the source directory
4082 instead of in the building directory. This caused the build to fail if the
4083 source directory was different from the building directory, and was
4084 read-only.
4085
4086 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
4087 file. No doubt somebody will tell me if they don't make sense... Also added
4088 Dan Mooney's comments about building on OpenVMS.
4089
4090 19. Added support for partial matching via the PCRE_PARTIAL option for
4091 pcre_exec() and the \P data escape in pcretest.
4092
4093 20. Extended pcretest with 3 new pattern features:
4094
4095 (i) A pattern option of the form ">rest-of-line" causes pcretest to
4096 write the compiled pattern to the file whose name is "rest-of-line".
4097 This is a straight binary dump of the data, with the saved pointer to
4098 the character tables forced to be NULL. The study data, if any, is
4099 written too. After writing, pcretest reads a new pattern.
4100
4101 (ii) If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
4102 compiled pattern from the given file. There must not be any
4103 occurrences of "<" in the file name (pretty unlikely); if there are,
4104 pcretest will instead treat the initial "<" as a pattern delimiter.
4105 After reading in the pattern, pcretest goes on to read data lines as
4106 usual.
4107
4108 (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
4109 and 16-bit fields in a compiled pattern, to simulate a pattern that
4110 was compiled on a host of opposite endianness.
4111
4112 21. The pcre-exec() function can now cope with patterns that were compiled on
4113 hosts of opposite endianness, with this restriction:
4114
4115 As for any compiled expression that is saved and used later, the tables
4116 pointer field cannot be preserved; the extra_data field in the arguments
4117 to pcre_exec() should be used to pass in a tables address if a value
4118 other than the default internal tables were used at compile time.
4119
4120 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
4121 now diagnosed as an error. Previously, most of the time, a negative number
4122 would have been treated as zero, but if in addition "ovector" was passed as
4123 NULL, a crash could occur.
4124
4125 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
4126 new versions from the libtool 1.5 distribution (the last one is a copy of
4127 a file called libtool.m4). This seems to have fixed the need to patch
4128 "configure" to support Darwin 1.3 (which I used to do). However, I still
4129 had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
4130 workstation).
4131
4132 24. Changed the PCRE licence to be the more standard "BSD" licence.
4133
4134
4135 Version 4.5 01-Dec-03
4136 ---------------------
4137
4138 1. There has been some re-arrangement of the code for the match() function so
4139 that it can be compiled in a version that does not call itself recursively.
4140 Instead, it keeps those local variables that need separate instances for
4141 each "recursion" in a frame on the heap, and gets/frees frames whenever it
4142 needs to "recurse". Keeping track of where control must go is done by means
4143 of setjmp/longjmp. The whole thing is implemented by a set of macros that
4144 hide most of the details from the main code, and operates only if
4145 NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
4146 "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
4147 operating.
4148
4149 To make it easier for callers to provide specially tailored get/free
4150 functions for this usage, two new functions, pcre_stack_malloc, and
4151 pcre_stack_free, are used. They are always called in strict stacking order,
4152 and the size of block requested is always the same.
4153
4154 The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
4155 PCRE has been compiled to use the stack or the heap for recursion. The
4156 -C option of pcretest uses this to show which version is compiled.
4157
4158 A new data escape \S, is added to pcretest; it causes the amounts of store
4159 obtained and freed by both kinds of malloc/free at match time to be added
4160 to the output.
4161
4162 2. Changed the locale test to use "fr_FR" instead of "fr" because that's
4163 what's available on my current Linux desktop machine.
4164
4165 3. When matching a UTF-8 string, the test for a valid string at the start has
4166 been extended. If start_offset is not zero, PCRE now checks that it points
4167 to a byte that is the start of a UTF-8 character. If not, it returns
4168 PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
4169 this is necessary because there may be backward assertions in the pattern.
4170 When matching the same subject several times, it may save resources to use
4171 PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
4172
4173 4. The code for checking the validity of UTF-8 strings has been tightened so
4174 that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
4175 containing "overlong sequences".
4176
4177 5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
4178 I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
4179 should have been "&", but it just so happened that all the cases this let
4180 through by mistake were picked up later in the function.
4181
4182 6. I had used a variable called "isblank" - this is a C99 function, causing
4183 some compilers to warn. To avoid this, I renamed it (as "blankclass").
4184
4185 7. Cosmetic: (a) only output another newline at the end of pcretest if it is
4186 prompting; (b) run "./pcretest /dev/null" at the start of the test script
4187 so the version is shown; (c) stop "make test" echoing "./RunTest".
4188
4189 8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
4190
4191 9. The prototype for memmove() for systems that don't have it was using
4192 size_t, but the inclusion of the header that defines size_t was later. I've
4193 moved the #includes for the C headers earlier to avoid this.
4194
4195 10. Added some adjustments to the code to make it easier to compiler on certain
4196 special systems:
4197
4198 (a) Some "const" qualifiers were missing.
4199 (b) Added the macro EXPORT before all exported functions; by default this
4200 is defined to be empty.
4201 (c) Changed the dftables auxiliary program (that builds chartables.c) so
4202 that it reads its output file name as an argument instead of writing
4203 to the standard output and assuming this can be redirected.
4204
4205 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
4206 class containing characters with values greater than 255, PCRE compilation
4207 went into a loop.
4208
4209 12. A recursive reference to a subpattern that was within another subpattern
4210 that had a minimum quantifier of zero caused PCRE to crash. For example,
4211 (x(y(?2))z)? provoked this bug with a subject that got as far as the
4212 recursion. If the recursively-called subpattern itself had a zero repeat,
4213 that was OK.
4214
4215 13. In pcretest, the buffer for reading a data line was set at 30K, but the
4216 buffer into which it was copied (for escape processing) was still set at
4217 1024, so long lines caused crashes.
4218
4219 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
4220 "internal error: code overflow...". This applied to any character class
4221 that was followed by a possessive quantifier.
4222
4223 15. Modified the Makefile to add libpcre.la as a prerequisite for
4224 libpcreposix.la because I was told this is needed for a parallel build to
4225 work.
4226
4227 16. If a pattern that contained .* following optional items at the start was
4228 studied, the wrong optimizing data was generated, leading to matching
4229 errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
4230 matching string must start with a or b or c. The correct conclusion for
4231 this pattern is that a match can start with any character.
4232
4233
4234 Version 4.4 13-Aug-03
4235 ---------------------
4236
4237 1. In UTF-8 mode, a character class containing characters with values between
4238 127 and 255 was not handled correctly if the compiled pattern was studied.
4239 In fixing this, I have also improved the studying algorithm for such
4240 classes (slightly).
4241
4242 2. Three internal functions had redundant arguments passed to them. Removal
4243 might give a very teeny performance improvement.
4244
4245 3. Documentation bug: the value of the capture_top field in a callout is *one
4246 more than* the number of the hightest numbered captured substring.
4247
4248 4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
4249 in incorrectly linking with a previously installed version. They now link
4250 explicitly with libpcre.la.
4251
4252 5. configure.in no longer needs to recognize Cygwin specially.
4253
4254 6. A problem in pcre.in for Windows platforms is fixed.
4255
4256 7. If a pattern was successfully studied, and the -d (or /D) flag was given to
4257 pcretest, it used to include the size of the study block as part of its
4258 output. Unfortunately, the structure contains a field that has a different
4259 size on different hardware architectures. This meant that the tests that
4260 showed this size failed. As the block is currently always of a fixed size,
4261 this information isn't actually particularly useful in pcretest output, so
4262 I have just removed it.
4263
4264 8. Three pre-processor statements accidentally did not start in column 1.
4265 Sadly, there are *still* compilers around that complain, even though
4266 standard C has not required this for well over a decade. Sigh.
4267
4268 9. In pcretest, the code for checking callouts passed small integers in the
4269 callout_data field, which is a void * field. However, some picky compilers
4270 complained about the casts involved for this on 64-bit systems. Now
4271 pcretest passes the address of the small integer instead, which should get
4272 rid of the warnings.
4273
4274 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
4275 both compile and run time, and gives an error if an invalid UTF-8 sequence
4276 is found. There is a option for disabling this check in cases where the
4277 string is known to be correct and/or the maximum performance is wanted.
4278
4279 11. In response to a bug report, I changed one line in Makefile.in from
4280
4281 -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
4282 to
4283 -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
4284
4285 to look similar to other lines, but I have no way of telling whether this
4286 is the right thing to do, as I do not use Windows. No doubt I'll get told
4287 if it's wrong...
4288
4289
4290 Version 4.3 21-May-03
4291 ---------------------
4292
4293 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
4294 Makefile.
4295
4296 2. Some refactoring to improve the quality of the code:
4297
4298 (i) The utf8_table... variables are now declared "const".
4299
4300 (ii) The code for \cx, which used the "case flipping" table to upper case
4301 lower case letters, now just substracts 32. This is ASCII-specific,
4302 but the whole concept of \cx is ASCII-specific, so it seems
4303 reasonable.
4304
4305 (iii) PCRE was using its character types table to recognize decimal and
4306 hexadecimal digits in the pattern. This is silly, because it handles
4307 only 0-9, a-f, and A-F, but the character types table is locale-
4308 specific, which means strange things might happen. A private
4309 table is now used for this - though it costs 256 bytes, a table is
4310 much faster than multiple explicit tests. Of course, the standard
4311 character types table is still used for matching digits in subject
4312 strings against \d.
4313
4314 (iv) Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
4315 ending in _t are). So I've renamed it as ESC_tee.
4316
4317 3. The first argument for regexec() in the POSIX wrapper should have been
4318 defined as "const".
4319
4320 4. Changed pcretest to use malloc() for its buffers so that they can be
4321 Electric Fenced for debugging.
4322
4323 5. There were several places in the code where, in UTF-8 mode, PCRE would try
4324 to read one or more bytes before the start of the subject string. Often this
4325 had no effect on PCRE's behaviour, but in some circumstances it could
4326 provoke a segmentation fault.
4327
4328 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
4329 to try to read one or more bytes before the start of the subject string.
4330
4331 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
4332 UTF-8 support could misbehave in various ways if the subject string
4333 contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
4334 area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
4335 back over UTF-8 characters.)
4336
4337
4338 Version 4.2 14-Apr-03
4339 ---------------------
4340
4341 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
4342
4343 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
4344 [ON_WINDOWS]: new variable, "#" on non-Windows platforms
4345 [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
4346 [WIN_PREFIX]: new variable, "cyg" for Cygwin
4347 * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
4348 and BUILD_EXEEXT
4349 Note: automatic setting of the BUILD variables is not yet working
4350 set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
4351 compile-time but not at link-time
4352 [LINK]: use for linking executables only
4353 make different versions for Windows and non-Windows
4354 [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
4355 libraries
4356 [LINK_FOR_BUILD]: new variable
4357 [OBJEXT]: use throughout
4358 [EXEEXT]: use throughout
4359 <winshared>: new target
4360 <wininstall>: new target
4361 <dftables.o>: use native compiler
4362 <dftables>: use native linker
4363 <install>: handle Windows platform correctly
4364 <clean>: ditto
4365 <check>: ditto
4366 copy DLL to top builddir before testing
4367
4368 As part of these changes, -no-undefined was removed again. This was reported
4369 to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
4370 in any case.
4371
4372 3. Some tidies to get rid of compiler warnings:
4373
4374 . In the match_data structure, match_limit was an unsigned long int, whereas
4375 match_call_count was an int. I've made them both unsigned long ints.
4376
4377 . In pcretest the fact that a const uschar * doesn't automatically cast to
4378 a void * provoked a warning.
4379
4380 . Turning on some more compiler warnings threw up some "shadow" variables
4381 and a few more missing casts.
4382
4383 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
4384 option, a class that contained a single character with a value between 128
4385 and 255 (e.g. /[\xFF]/) caused PCRE to crash.
4386
4387 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
4388 option, a class that contained several characters, but with at least one
4389 whose value was between 128 and 255 caused PCRE to crash.
4390
4391
4392 Version 4.1 12-Mar-03
4393 ---------------------
4394
4395 1. Compiling with gcc -pedantic found a couple of places where casts were
4396 needed, and a string in dftables.c that was longer than standard compilers are
4397 required to support.
4398
4399 2. Compiling with Sun's compiler found a few more places where the code could
4400 be tidied up in order to avoid warnings.
4401
4402 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
4403 first of these names is deprecated in the latest Autoconf in favour of the name
4404 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
4405 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
4406 analogy I have changed it to CFLAGS_FOR_BUILD.
4407
4408 4. Added -no-undefined to the linking command in the Makefile, because this is
4409 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
4410 linking step for the pcreposix library.
4411
4412 5. PCRE was failing to diagnose the case of two named groups with the same
4413 name.
4414
4415 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
4416 literal character that is needed in the subject for a match, and scans along to
4417 ensure that it is present before embarking on the full matching process. This
4418 saves time in cases of nested unlimited repeats that are never going to match.
4419 Problem: the scan can take a lot of time if the subject is very long (e.g.
4420 megabytes), thus penalizing straightforward matches. It is now done only if the
4421 amount of subject to be scanned is less than 1000 bytes.
4422
4423 7. A lesser problem with the same optimization is that it was recording the
4424 first character of an anchored pattern as "needed", thus provoking a search
4425 right along the subject, even when the first match of the pattern was going to
4426 fail. The "needed" character is now not set for anchored patterns, unless it
4427 follows something in the pattern that is of non-fixed length. Thus, it still
4428 fulfils its original purpose of finding quick non-matches in cases of nested
4429 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
4430
4431
4432 Version 4.0 17-Feb-03
4433 ---------------------
4434
4435 1. If a comment in an extended regex that started immediately after a meta-item
4436 extended to the end of string, PCRE compiled incorrect data. This could lead to
4437 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
4438
4439 2. Moved to autoconf 2.53 and libtool 1.4.2.
4440
4441 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
4442 the special perltest8 script is no longer needed - all the tests can be run
4443 from a single perltest script.
4444
4445 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
4446 by \s. It has now been removed in PCRE. This means it isn't recognized as
4447 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
4448 class [:space:] *does* include VT, thereby creating a mess.
4449
4450 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
4451 space and tab.
4452
4453 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
4454 its new features into the main test script, reducing the number of scripts.
4455
4456 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
4457 were backward compatible, and made the (?i) apply to the whole pattern, as if
4458 /i were given. Now it behaves more logically, and applies the option setting
4459 only to what follows. PCRE has been changed to follow suit. However, if it
4460 finds options settings right at the start of the pattern, it extracts them into
4461 the global options, as before. Thus, they show up in the info data.
4462
4463 8. Added support for the \Q...\E escape sequence. Characters in between are
4464 treated as literals. This is slightly different from Perl in that $ and @ are
4465 also handled as literals inside the quotes. In Perl, they will cause variable
4466 interpolation. Note the following examples:
4467
4468 Pattern PCRE matches Perl matches
4469
4470 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
4471 \Qabc\$xyz\E abc\$xyz abc\$xyz
4472 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
4473
4474 For compatibility with Perl, \Q...\E sequences are recognized inside character
4475 classes as well as outside them.
4476
4477 9. Re-organized 3 code statements in pcretest to avoid "overflow in
4478 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
4479 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
4480 signed/unsigned warnings.
4481
4482 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
4483 option for pcretest, so I've replaced it by a simple function that does just
4484 that job.
4485
4486 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
4487 "pcregrep -".
4488
4489 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
4490 Java package. This provides some syntactic sugar for simple cases of what my
4491 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
4492 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
4493 item, you can use this simplified notation. Note that only makes sense with
4494 greedy quantifiers. Consequently, the use of the possessive quantifier forces
4495 greediness, whatever the setting of the PCRE_UNGREEDY option.
4496
4497 13. A change of greediness default within a pattern was not taking effect at
4498 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
4499 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
4500 was abstracted outside.
4501
4502 14. PCRE now supports the \G assertion. It is true when the current matching
4503 position is at the start point of the match. This differs from \A when the
4504 starting offset is non-zero. Used with the /g option of pcretest (or similar
4505 code), it works in the same way as it does for Perl's /g option. If all
4506 alternatives of a regex begin with \G, the expression is anchored to the start
4507 match position, and the "anchored" flag is set in the compiled expression.
4508
4509 15. Some bugs concerning the handling of certain option changes within patterns
4510 have been fixed. These applied to options other than (?ims). For example,
4511 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
4512 been the other way round. Some of this was related to change 7 above.
4513
4514 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
4515 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
4516 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
4517 POSIX classes only within a class (e.g. /[[:alpha:]]/).
4518
4519 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
4520 mode. Unlike ".", it always matches newline, whatever the setting of
4521 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
4522 assertions. Perl allows it, but it doesn't (in general) work because it can't
4523 calculate the length of the lookbehind. At least, that's the case for Perl
4524 5.8.0 - I've been told they are going to document that it doesn't work in
4525 future.
4526
4527 18. Added an error diagnosis for escapes that PCRE does not support: these are
4528 \L, \l, \N, \P, \p, \U, \u, and \X.
4529
4530 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
4531 reading past the end of the pattern in cases such as /[abcd/.
4532
4533 20. PCRE was getting more memory than necessary for patterns with classes that
4534 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
4535
4536 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
4537 compiling PCRE for use with Virtual Pascal.
4538
4539 22. Small fix to the Makefile to make it work properly if the build is done
4540 outside the source tree.
4541
4542 23. Added a new extension: a condition to go with recursion. If a conditional
4543 subpattern starts with (?(R) the "true" branch is used if recursion has
4544 happened, whereas the "false" branch is used only at the top level.
4545
4546 24. When there was a very long string of literal characters (over 255 bytes
4547 without UTF support, over 250 bytes with UTF support), the computation of how
4548 much memory was required could be incorrect, leading to segfaults or other
4549 strange effects.
4550
4551 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
4552 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
4553 there was a subsequent back reference to those brackets. This meant that, for
4554 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
4555 possible to check for precisely this case. All we can do is abandon the
4556 optimization if .* occurs inside capturing brackets when there are any back
4557 references whatsoever. (See below for a better fix that came later.)
4558
4559 26. The handling of the optimization for finding the first character of a
4560 non-anchored pattern, and for finding a character that is required later in the
4561 match were failing in some cases. This didn't break the matching; it just
4562 failed to optimize when it could. The way this is done has been re-implemented.
4563
4564 27. Fixed typo in error message for invalid (?R item (it said "(?p").
4565
4566 28. Added a new feature that provides some of the functionality that Perl
4567 provides with (?{...}). The facility is termed a "callout". The way it is done
4568 in PCRE is for the caller to provide an optional function, by setting
4569 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
4570 global variable. By default it is unset, which disables all calling out. To get
4571 the function called, the regex must include (?C) at appropriate points. This
4572 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
4573 This provides a means of identifying different callout points. When PCRE
4574 reaches such a point in the regex, if pcre_callout has been set, the external
4575 function is called. It is provided with data in a structure called
4576 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
4577 matching continues; if it returns a non-zero value, the match at the current
4578 point fails. However, backtracking will occur if possible. [This was changed
4579 later and other features added - see item 49 below.]
4580
4581 29. pcretest is upgraded to test the callout functionality. It provides a
4582 callout function that displays information. By default, it shows the start of
4583 the match and the current position in the text. There are some new data escapes
4584 to vary what happens:
4585
4586 \C+ in addition, show current contents of captured substrings
4587 \C- do not supply a callout function
4588 \C!n return 1 when callout number n is reached
4589 \C!n!m return 1 when callout number n is reached for the mth time
4590
4591 30. If pcregrep was called with the -l option and just a single file name, it
4592 output "<stdin>" if a match was found, instead of the file name.
4593
4594 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
4595 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
4596 pcre_exec(). This saves a malloc/free per call. The default value of
4597 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
4598 when configuring.
4599
4600 32. The default maximum size of a compiled pattern is 64K. There have been a
4601 few cases of people hitting this limit. The code now uses macros to handle the
4602 storing of links as offsets within the compiled pattern. It defaults to 2-byte
4603 links, but this can be changed to 3 or 4 bytes by --with-link-size when
4604 configuring. Tests 2 and 5 work only with 2-byte links because they output
4605 debugging information about compiled patterns.
4606
4607 33. Internal code re-arrangements:
4608
4609 (a) Moved the debugging function for printing out a compiled regex into
4610 its own source file (printint.c) and used #include to pull it into
4611 pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
4612 separate copies.
4613
4614 (b) Defined the list of op-code names for debugging as a macro in
4615 internal.h so that it is next to the definition of the opcodes.
4616
4617 (c) Defined a table of op-code lengths for simpler skipping along compiled
4618 code. This is again a macro in internal.h so that it is next to the
4619 definition of the opcodes.
4620
4621 34. Added support for recursive calls to individual subpatterns, along the
4622 lines of Robin Houston's patch (but implemented somewhat differently).
4623
4624 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
4625 allow it to read and process whole directories in Win32. This code was
4626 contributed by Lionel Fourquaux; it has not been tested by me.
4627
4628 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
4629 used to name a group. Names consist of alphanumerics and underscores, and must
4630 be unique. Back references use the syntax (?P=name) and recursive calls use
4631 (?P>name) which is a PCRE extension to the Python extension. Groups still have
4632 numbers. The function pcre_fullinfo() can be used after compilation to extract
4633 a name/number map. There are three relevant calls:
4634
4635 PCRE_INFO_NAMEENTRYSIZE yields the size of each entry in the map
4636 PCRE_INFO_NAMECOUNT yields the number of entries
4637 PCRE_INFO_NAMETABLE yields a pointer to the map.
4638
4639 The map is a vector of fixed-size entries. The size of each entry depends on
4640 the length of the longest name used. The first two bytes of each entry are the
4641 group number, most significant byte first. There follows the corresponding
4642 name, zero terminated. The names are in alphabetical order.
4643
4644 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
4645 case instead of 255. Making it the same both with and without UTF-8 support
4646 means that the same test output works with both.
4647
4648 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
4649 calling malloc() with a zero argument.
4650
4651 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
4652 optimization. I've improved things by keeping a bitmap of backreferences with
4653 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
4654 fact referenced, the optimization can be applied. It is unlikely that a
4655 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
4656 the match to follow \n) will appear inside brackets with a number greater than
4657 31, but if it does, any back reference > 31 suppresses the optimization.
4658
4659 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
4660 of disabling numbered capturing parentheses. Any opening parenthesis that is
4661 not followed by ? behaves as if it were followed by ?: but named parentheses
4662 can still be used for capturing (and they will acquire numbers in the usual
4663 way).
4664
4665 41. Redesigned the return codes from the match() function into yes/no/error so
4666 that errors can be passed back from deep inside the nested calls. A malloc
4667 failure while inside a recursive subpattern call now causes the
4668 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
4669
4670 42. It is now possible to set a limit on the number of times the match()
4671 function is called in a call to pcre_exec(). This facility makes it possible to
4672 limit the amount of recursion and backtracking, though not in a directly
4673 obvious way, because the match() function is used in a number of different
4674 circumstances. The count starts from zero for each position in the subject
4675 string (for non-anchored patterns). The default limit is, for compatibility, a
4676 large number, namely 10 000 000. You can change this in two ways:
4677
4678 (a) When configuring PCRE before making, you can use --with-match-limit=n
4679 to set a default value for the compiled library.
4680
4681 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
4682 a different value is set. See 45 below.
4683
4684 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
4685
4686 43. Added a new function pcre_config(int, void *) to enable run-time extraction
4687 of things that can be changed at compile time. The first argument specifies
4688 what is wanted and the second points to where the information is to be placed.
4689 The current list of available information is:
4690
4691 PCRE_CONFIG_UTF8
4692
4693 The output is an integer that is set to one if UTF-8 support is available;
4694 otherwise it is set to zero.
4695
4696 PCRE_CONFIG_NEWLINE
4697
4698 The output is an integer that it set to the value of the code that is used for
4699 newline. It is either LF (10) or CR (13).
4700
4701 PCRE_CONFIG_LINK_SIZE
4702
4703 The output is an integer that contains the number of bytes used for internal
4704 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
4705
4706 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
4707
4708 The output is an integer that contains the threshold above which the POSIX
4709 interface uses malloc() for output vectors. See item 31 above.
4710
4711 PCRE_CONFIG_MATCH_LIMIT
4712
4713 The output is an unsigned integer that contains the default limit of the number
4714 of match() calls in a pcre_exec() execution. See 42 above.
4715
4716 44. pcretest has been upgraded by the addition of the -C option. This causes it
4717 to extract all the available output from the new pcre_config() function, and to
4718 output it. The program then exits immediately.
4719
4720 45. A need has arisen to pass over additional data with calls to pcre_exec() in
4721 order to support additional features. One way would have been to define
4722 pcre_exec2() (for example) with extra arguments, but this would not have been
4723 extensible, and would also have required all calls to the original function to
4724 be mapped to the new one. Instead, I have chosen to extend the mechanism that
4725 is used for passing in "extra" data from pcre_study().
4726
4727 The pcre_extra structure is now exposed and defined in pcre.h. It currently
4728 contains the following fields:
4729
4730 flags a bitmap indicating which of the following fields are set
4731 study_data opaque data from pcre_study()
4732 match_limit a way of specifying a limit on match() calls for a specific
4733 call to pcre_exec()
4734 callout_data data for callouts (see 49 below)
4735
4736 The flag bits are also defined in pcre.h, and are
4737
4738 PCRE_EXTRA_STUDY_DATA
4739 PCRE_EXTRA_MATCH_LIMIT
4740 PCRE_EXTRA_CALLOUT_DATA
4741
4742 The pcre_study() function now returns one of these new pcre_extra blocks, with
4743 the actual study data pointed to by the study_data field, and the
4744 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
4745 before. That is, this change is entirely upwards-compatible and requires no
4746 change to existing code.
4747
4748 If you want to pass in additional data to pcre_exec(), you can either place it
4749 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
4750 block.
4751
4752 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
4753 data string contains the escape sequence \M, pcretest calls pcre_exec() several
4754 times with different match limits, until it finds the minimum value needed for
4755 pcre_exec() to complete. The value is then output. This can be instructive; for
4756 most simple matches the number is quite small, but for pathological cases it
4757 gets very large very quickly.
4758
4759 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
4760 returns the size of the data block pointed to by the study_data field in a
4761 pcre_extra block, that is, the value that was passed as the argument to
4762 pcre_malloc() when PCRE was getting memory in which to place the information
4763 created by pcre_study(). The fourth argument should point to a size_t variable.
4764 pcretest has been extended so that this information is shown after a successful
4765 pcre_study() call when information about the compiled regex is being displayed.
4766
4767 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
4768 because what follows is always an absolute path. (Later: it turns out that this
4769 is more than cosmetic for MinGW, because it doesn't like empty path
4770 components.)
4771
4772 49. Some changes have been made to the callout feature (see 28 above):
4773
4774 (i) A callout function now has three choices for what it returns:
4775
4776 0 => success, carry on matching
4777 > 0 => failure at this point, but backtrack if possible
4778 < 0 => serious error, return this value from pcre_exec()
4779
4780 Negative values should normally be chosen from the set of PCRE_ERROR_xxx
4781 values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
4782 "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
4783 use by callout functions. It will never be used by PCRE itself.
4784
4785 (ii) The pcre_extra structure (see 45 above) has a void * field called
4786 callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
4787 pcre_callout_block structure has a field of the same name. The contents of
4788 the field passed in the pcre_extra structure are passed to the callout
4789 function in the corresponding field in the callout block. This makes it
4790 easier to use the same callout-containing regex from multiple threads. For
4791 testing, the pcretest program has a new data escape
4792
4793 \C*n pass the number n (may be negative) as callout_data
4794
4795 If the callout function in pcretest receives a non-zero value as
4796 callout_data, it returns that value.
4797
4798 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
4799 there were some redundant $(CFLAGS) in commands that are now specified as
4800 $(LINK), which already includes $(CFLAGS).
4801
4802 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
4803 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
4804 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
4805 one-byte characters throughout. Note that case-insensitive matching applies
4806 only to characters whose values are less than 256. PCRE doesn't support the
4807 notion of cases for higher-valued characters.
4808
4809 (i) A character class whose characters are all within 0-255 is handled as
4810 a bit map, and the map is inverted for negative classes. Previously, a
4811 character > 255 always failed to match such a class; however it should
4812 match if the class was a negative one (e.g. [^ab]). This has been fixed.
4813
4814 (ii) A negated character class with a single character < 255 is coded as
4815 "not this character" (OP_NOT). This wasn't working properly when the test
4816 character was multibyte, either singly or repeated.
4817
4818 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
4819 mode, for example: \x{100}{2,3}.
4820
4821 (iv) The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
4822 singly or repeated) now correctly test multibyte characters. However,
4823 PCRE doesn't recognize any characters with values greater than 255 as
4824 digits, spaces, or word characters. Such characters always match \D, \S,
4825 and \W, and never match \d, \s, or \w.
4826
4827 (v) Classes may now contain characters and character ranges with values
4828 greater than 255. For example: [ab\x{100}-\x{400}].
4829
4830 (vi) pcregrep now has a --utf-8 option (synonym -u) which makes it call
4831 PCRE in UTF-8 mode.
4832
4833 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
4834 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
4835 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
4836 value.)
4837
4838 53. The single man page has become too large. I have therefore split it up into
4839 a number of separate man pages. These also give rise to individual HTML pages;
4840 these are now put in a separate directory, and there is an index.html page that
4841 lists them all. Some hyperlinking between the pages has been installed.
4842
4843 54. Added convenience functions for handling named capturing parentheses.
4844
4845 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
4846 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
4847 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
4848 are faulted.
4849
4850 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
4851 calling configure. These values are used when compiling the dftables.c program
4852 which is run to generate the source of the default character tables. They
4853 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
4854 you will need to set these values.
4855
4856 57. Updated the building process for Windows DLL, as provided by Fred Cox.
4857
4858
4859 Version 3.9 02-Jan-02
4860 ---------------------
4861
4862 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
4863
4864 2. If --disable-static was given, the building process failed when trying to
4865 build pcretest and pcregrep. (For some reason it was using libtool to compile
4866 them, which is not right, as they aren't part of the library.)
4867
4868
4869 Version 3.8 18-Dec-01
4870 ---------------------
4871
4872 1. The experimental UTF-8 code was completely screwed up. It was packing the
4873 bytes in the wrong order. How dumb can you get?
4874
4875
4876 Version 3.7 29-Oct-01
4877 ---------------------
4878
4879 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
4880 This caused pcretest, when used on the test data, to segfault. Unfortunately,
4881 this didn't happen under Solaris 8, where I normally test things.
4882
4883 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
4884 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
4885 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
4886 this fix an hour or so after the initial 3.7 release.)
4887
4888
4889 Version 3.6 23-Oct-01
4890 ---------------------
4891
4892 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
4893 offsets passed as NULL with zero offset count.
4894
4895 2. The config.guess and config.sub files had not been updated when I moved to
4896 the latest autoconf.
4897
4898
4899 Version 3.5 15-Aug-01
4900 ---------------------
4901
4902 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
4903 had been forgotten.
4904
4905 2. By using declared but undefined structures, we can avoid using "void"
4906 definitions in pcre.h while keeping the internal definitions of the structures
4907 private.
4908
4909 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
4910 user point of view, this means that both static and shared libraries are built
4911 by default, but this can be individually controlled. More of the work of
4912 handling this static/shared cases is now inside libtool instead of PCRE's make
4913 file.
4914
4915 4. The pcretest utility is now installed along with pcregrep because it is
4916 useful for users (to test regexs) and by doing this, it automatically gets
4917 relinked by libtool. The documentation has been turned into a man page, so
4918 there are now .1, .txt, and .html versions in /doc.
4919
4920 5. Upgrades to pcregrep:
4921 (i) Added long-form option names like gnu grep.
4922 (ii) Added --help to list all options with an explanatory phrase.
4923 (iii) Added -r, --recursive to recurse into sub-directories.
4924 (iv) Added -f, --file to read patterns from a file.
4925
4926 6. pcre_exec() was referring to its "code" argument before testing that
4927 argument for NULL (and giving an error if it was NULL).
4928
4929 7. Upgraded Makefile.in to allow for compiling in a different directory from
4930 the source directory.
4931
4932 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
4933 options bits, the pointer it was passed was to an int instead of to an unsigned
4934 long int. This mattered only on 64-bit systems.
4935
4936 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
4937 generated) instead of pcre.in, which it its source. Also made the same change
4938 in several of the .c files.
4939
4940 10. A new release of gcc defines printf() as a macro, which broke pcretest
4941 because it had an ifdef in the middle of a string argument for printf(). Fixed
4942 by using separate calls to printf().
4943
4944 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
4945 script, to force use of CR or LF instead of \n in the source. On non-Unix
4946 systems, the value can be set in config.h.
4947
4948 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
4949 absolute limit. Changed the text of the error message to make this clear, and
4950 likewise updated the man page.
4951
4952 13. The limit of 99 on the number of capturing subpatterns has been removed.
4953 The new limit is 65535, which I hope will not be a "real" limit.
4954
4955
4956 Version 3.4 22-Aug-00
4957 ---------------------
4958
4959 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
4960
4961 2. Diagnose condition (?(0) as an error instead of crashing on matching.
4962
4963
4964 Version 3.3 01-Aug-00
4965 ---------------------
4966
4967 1. If an octal character was given, but the value was greater than \377, it
4968 was not getting masked to the least significant bits, as documented. This could
4969 lead to crashes in some systems.
4970
4971 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
4972 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
4973
4974 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
4975 These just pass their arguments on to (pcre_free)(), but they are provided
4976 because some uses of PCRE bind it to non-C systems that can call its functions,
4977 but cannot call free() or pcre_free() directly.
4978
4979 4. Add "make test" as a synonym for "make check". Corrected some comments in
4980 the Makefile.
4981
4982 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
4983 Makefile.
4984
4985 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
4986 command called pgrep for grepping around the active processes.
4987
4988 7. Added the beginnings of support for UTF-8 character strings.
4989
4990 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
4991 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
4992 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
4993 out for the ar command.)
4994
4995
4996 Version 3.2 12-May-00
4997 ---------------------
4998
4999 This is purely a bug fixing release.
5000
5001 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
5002 of ZA. This was just one example of several cases that could provoke this bug,
5003 which was introduced by change 9 of version 2.00. The code for breaking
5004 infinite loops after an iteration that matches an empty string was't working
5005 correctly.
5006
5007 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
5008 when matched against abbab (for example). After matching an empty string, it
5009 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
5010 caused it to match further down the string than it should.
5011
5012 3. The code contained an inclusion of sys/types.h. It isn't clear why this
5013 was there because it doesn't seem to be needed, and it causes trouble on some
5014 systems, as it is not a Standard C header. It has been removed.
5015
5016 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
5017 were reported on the Macintosh. The changes were from
5018
5019 while ((c = *(++ptr)) != 0 && c != '\n');
5020 to
5021 while ((c = *(++ptr)) != 0 && c != '\n') ;
5022
5023 Totally extraordinary, but if that's what it takes...
5024
5025 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
5026 available. Added HAVE_BCOPY and an autoconf test for it; if neither
5027 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
5028 assumes the way PCRE uses memmove() (always moving upwards).
5029
5030 6. PCRE is being used in one environment where strchr() is not available. There
5031 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
5032 faster code anyway.
5033
5034
5035 Version 3.1 09-Feb-00
5036 ---------------------
5037
5038 The only change in this release is the fixing of some bugs in Makefile.in for
5039 the "install" target:
5040
5041 (1) It was failing to install pcreposix.h.
5042
5043 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
5044
5045
5046 Version 3.0 01-Feb-00
5047 ---------------------
5048
5049 1. Add support for the /+ modifier to perltest (to output $` like it does in
5050 pcretest).
5051
5052 2. Add support for the /g modifier to perltest.
5053
5054 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
5055 matches null strings.
5056
5057 4. Fix perltest so that it doesn't do unwanted things when fed an empty
5058 pattern. Perl treats empty patterns specially - it reuses the most recent
5059 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
5060 effect.
5061
5062 5. The POSIX interface was broken in that it was just handing over the POSIX
5063 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
5064 required a bigger vector, with some working space on the end. This means that
5065 the POSIX wrapper now has to get and free some memory, and copy the results.
5066
5067 6. Added some simple autoconf support, placing the test data and the
5068 documentation in separate directories, re-organizing some of the
5069 information files, and making it build pcre-config (a GNU standard). Also added
5070 libtool support for building PCRE as a shared library, which is now the
5071 default.
5072
5073 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
5074 09 are not valid octal constants. Single digits will be used for minor values
5075 less than 10.
5076
5077 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
5078 existing programs that set these in the POSIX interface can use PCRE without
5079 modification.
5080
5081 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
5082 return all that pcre_info() returns, plus additional data. The pcre_info()
5083 function is retained for compatibility, but is considered to be obsolete.
5084
5085 10. Added experimental recursion feature (?R) to handle one common case that
5086 Perl 5.6 will be able to do with (?p{...}).
5087
5088 11. Added support for POSIX character classes like [:alpha:], which Perl is
5089 adopting.
5090
5091
5092 Version 2.08 31-Aug-99
5093 ----------------------
5094
5095 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
5096 trying to match at the startoffset position, but instead was moving forward to
5097 the next newline as if a previous match had failed.
5098
5099 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
5100 and could get into a loop if a null string was matched other than at the start
5101 of the subject.
5102
5103 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
5104 be distinguished at compile time, and for completeness also added PCRE_DATE.
5105
5106 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
5107 in GnuWin32 environments.
5108
5109
5110 Version 2.07 29-Jul-99
5111 ----------------------
5112
5113 1. The documentation is now supplied in plain text form and HTML as well as in
5114 the form of man page sources.
5115
5116 2. C++ compilers don't like assigning (void *) values to other pointer types.
5117 In particular this affects malloc(). Although there is no problem in Standard
5118 C, I've put in casts to keep C++ compilers happy.
5119
5120 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
5121 should be (const char *).
5122
5123 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
5124 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
5125 However, I haven't made this a standard facility. The documentation doesn't
5126 mention it, and the Makefile doesn't support it.
5127
5128 5. The Makefile now contains an "install" target, with editable destinations at
5129 the top of the file. The pcretest program is not installed.
5130
5131 6. pgrep -V now gives the PCRE version number and date.
5132
5133 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
5134 causing the entire string to be ignored, instead of just the last character.
5135
5136 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
5137 non-matching string, it can take a very, very long time, even for strings of
5138 quite modest length, because of the nested recursion. PCRE now does better in
5139 some of these cases. It does this by remembering the last required literal
5140 character in the pattern, and pre-searching the subject to ensure it is present
5141 before running the real match. In other words, it applies a heuristic to detect
5142 some types of certain failure quickly, and in the above example, if presented
5143 with a string that has no trailing " it gives "no match" very quickly.
5144
5145 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
5146 other alternatives are tried instead.
5147
5148
5149 Version 2.06 09-Jun-99
5150 ----------------------
5151
5152 1. Change pcretest's output for amount of store used to show just the code
5153 space, because the remainder (the data block) varies in size between 32-bit and
5154 64-bit systems.
5155
5156 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
5157 start matching at. This allows lookbehinds to work when searching for multiple
5158 occurrences in a string.
5159
5160 3. Added additional options to pcretest for testing multiple occurrences:
5161
5162 /+ outputs the rest of the string that follows a match
5163 /g loops for multiple occurrences, using the new startoffset argument
5164 /G loops for multiple occurrences by passing an incremented pointer
5165
5166 4. PCRE wasn't doing the "first character" optimization for patterns starting
5167 with \b or \B, though it was doing it for other lookbehind assertions. That is,
5168 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
5169 the letter 'x'. On long subject strings, this gives a significant speed-up.
5170
5171
5172 Version 2.05 21-Apr-99
5173 ----------------------
5174
5175 1. Changed the type of magic_number from int to long int so that it works
5176 properly on 16-bit systems.
5177
5178 2. Fixed a bug which caused patterns starting with .* not to work correctly
5179 when the subject string contained newline characters. PCRE was assuming
5180 anchoring for such patterns in all cases, which is not correct because .* will
5181 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
5182 DOTALL is set at top level; otherwise it knows that patterns starting with .*
5183 must be retried after every newline in the subject.
5184
5185
5186 Version 2.04 18-Feb-99
5187 ----------------------
5188
5189 1. For parenthesized subpatterns with repeats whose minimum was zero, the
5190 computation of the store needed to hold the pattern was incorrect (too large).
5191 If such patterns were nested a few deep, this could multiply and become a real
5192 problem.
5193
5194 2. Added /M option to pcretest to show the memory requirement of a specific
5195 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
5196
5197 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
5198 compiled in such a way that the backtracking after subsequent failure was
5199 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
5200 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
5201
5202
5203 Version 2.03 02-Feb-99
5204 ----------------------
5205
5206 1. Fixed typo and small mistake in man page.
5207
5208 2. Added 4th condition (GPL supersedes if conflict) and created separate
5209 LICENCE file containing the conditions.
5210
5211 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
5212 Perl, that is the internal \ allows the delimiter to be included in the
5213 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
5214 the final delimiter, add \ to the end of the pattern (to test the error).
5215
5216 4. Added the convenience functions for extracting substrings after a successful
5217 match. Updated pcretest to make it able to test these functions.
5218
5219
5220 Version 2.02 14-Jan-99
5221 ----------------------
5222
5223 1. Initialized the working variables associated with each extraction so that
5224 their saving and restoring doesn't refer to uninitialized store.
5225
5226 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
5227 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
5228 fix the problem.
5229
5230 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
5231 calls, and wasn't printing the correct value for compiling calls. Increased the
5232 default value of LOOPREPEAT, and the number of significant figures in the
5233 times.
5234
5235 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
5236
5237 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
5238 a building problem on Windows NT with a FAT file system.
5239
5240
5241 Version 2.01 21-Oct-98
5242 ----------------------
5243
5244 1. Changed the API for pcre_compile() to allow for the provision of a pointer
5245 to character tables built by pcre_maketables() in the current locale. If NULL
5246 is passed, the default tables are used.
5247
5248
5249 Version 2.00 24-Sep-98
5250 ----------------------
5251
5252 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
5253 it any more.
5254
5255 2. Allow quantification of (?>) groups, and make it work correctly.
5256
5257 3. The first character computation wasn't working for (?>) groups.
5258
5259 4. Correct the implementation of \Z (it is permitted to match on the \n at the
5260 end of the subject) and add 5.005's \z, which really does match only at the
5261 very end of the subject.
5262
5263 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
5264
5265 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
5266 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
5267 localized options. All options to pcre_study() were also removed.
5268
5269 7. Add other new features from 5.005:
5270
5271 $(?<= positive lookbehind
5272 $(?<! negative lookbehind
5273 (?imsx-imsx) added the unsetting capability
5274 such a setting is global if at outer level; local otherwise
5275 (?imsx-imsx:) non-capturing groups with option setting
5276 (?(cond)re|re) conditional pattern matching
5277
5278 A backreference to itself in a repeated group matches the previous
5279 captured string.
5280
5281 8. General tidying up of studying (both automatic and via "study")
5282 consequential on the addition of new assertions.
5283
5284 9. As in 5.005, unlimited repeated groups that could match an empty substring
5285 are no longer faulted at compile time. Instead, the loop is forcibly broken at
5286 runtime if any iteration does actually match an empty substring.
5287
5288 10. Include the RunTest script in the distribution.
5289
5290 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
5291 discrepancies, some of which were old and were also with respect to 5.004. They
5292 have now been fixed.
5293
5294
5295 Version 1.09 28-Apr-98
5296 ----------------------
5297
5298 1. A negated single character class followed by a quantifier with a minimum
5299 value of one (e.g. [^x]{1,6} ) was not compiled correctly. This could lead to
5300 program crashes, or just wrong answers. This did not apply to negated classes
5301 containing more than one character, or to minima other than one.
5302
5303
5304 Version 1.08 27-Mar-98
5305 ----------------------
5306
5307 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
5308
5309 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
5310 latter must appear before anything that relies on it in the pattern.
5311
5312
5313 Version 1.07 16-Feb-98
5314 ----------------------
5315
5316 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
5317 repeat of a potentially empty string).
5318
5319
5320 Version 1.06 23-Jan-98
5321 ----------------------
5322
5323 1. Added Markus Oberhumer's little patches for C++.
5324
5325 2. Literal strings longer than 255 characters were broken.
5326
5327
5328 Version 1.05 23-Dec-97
5329 ----------------------
5330
5331 1. Negated character classes containing more than one character were failing if
5332 PCRE_CASELESS was set at run time.
5333
5334
5335 Version 1.04 19-Dec-97
5336 ----------------------
5337
5338 1. Corrected the man page, where some "const" qualifiers had been omitted.
5339
5340 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
5341 input syntax.
5342
5343 3. Fixed memory leak which occurred when a regex with back references was
5344 matched with an offsets vector that wasn't big enough. The temporary memory
5345 that is used in this case wasn't being freed if the match failed.
5346
5347 4. Tidied pcretest to ensure it frees memory that it gets.
5348
5349 5. Temporary memory was being obtained in the case where the passed offsets
5350 vector was exactly big enough.
5351
5352 6. Corrected definition of offsetof() from change 5 below.
5353
5354 7. I had screwed up change 6 below and broken the rules for the use of
5355 setjmp(). Now fixed.
5356
5357
5358 Version 1.03 18-Dec-97
5359 ----------------------
5360
5361 1. A erroneous regex with a missing opening parenthesis was correctly
5362 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
5363 on some systems.
5364
5365 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
5366 it was reported that one broken compiler failed on the former because "code" is
5367 also an independent variable.
5368
5369 3. The erroneous regex a[]b caused an array overrun reference.
5370
5371 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
5372 fail on data ending with that character. (It was going on too far, and checking
5373 the next character, typically a binary zero.) This was specific to the
5374 optimized code for single-character negative classes.
5375
5376 5. Added a contributed patch from the TIN world which does the following:
5377
5378 + Add an undef for memmove, in case the the system defines a macro for it.
5379
5380 + Add a definition of offsetof(), in case there isn't one. (I don't know
5381 the reason behind this - offsetof() is part of the ANSI standard - but
5382 it does no harm).
5383
5384 + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
5385 most of the places where whitespace preceded '#'. I have given up and
5386 allowed the remaining 2 cases to be at the margin.
5387
5388 + Rename some variables in pcre to eliminate shadowing. This seems very
5389 pedantic, but does no harm, of course.
5390
5391 6. Moved the call to setjmp() into its own function, to get rid of warnings
5392 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
5393
5394 7. Constructs such as \d{8,} were compiling into the equivalent of
5395 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
5396 outcome, but in this particular case used more store than had been allocated,
5397 which caused the bug to be discovered because it threw up an internal error.
5398
5399 8. The debugging code in both pcre and pcretest for outputting the compiled
5400 form of a regex was going wrong in the case of back references followed by
5401 curly-bracketed repeats.
5402
5403
5404 Version 1.02 12-Dec-97
5405 ----------------------
5406
5407 1. Typos in pcre.3 and comments in the source fixed.
5408
5409 2. Applied a contributed patch to get rid of places where it used to remove
5410 'const' from variables, and fixed some signed/unsigned and uninitialized
5411 variable warnings.
5412
5413 3. Added the "runtest" target to Makefile.
5414
5415 4. Set default compiler flag to -O2 rather than just -O.
5416
5417
5418 Version 1.01 19-Nov-97
5419 ----------------------
5420
5421 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
5422 like /([ab]*)*/, that is, for classes with more than one character in them.
5423
5424 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
5425 as /((?>a*))*/ (a PCRE_EXTRA facility).
5426
5427
5428 Version 1.00 18-Nov-97
5429 ----------------------
5430
5431 1. Added compile-time macros to support systems such as SunOS4 which don't have
5432 memmove() or strerror() but have other things that can be used instead.
5433
5434 2. Arranged that "make clean" removes the executables.
5435
5436
5437 Version 0.99 27-Oct-97
5438 ----------------------
5439
5440 1. Fixed bug in code for optimizing classes with only one character. It was
5441 initializing a 32-byte map regardless, which could cause it to run off the end
5442 of the memory it had got.
5443
5444 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
5445
5446
5447 Version 0.98 22-Oct-97
5448 ----------------------
5449
5450 1. Fixed bug in code for handling temporary memory usage when there are more
5451 back references than supplied space in the ovector. This could cause segfaults.
5452
5453
5454 Version 0.97 21-Oct-97
5455 ----------------------
5456
5457 1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
5458
5459 2. Optimized negated single characters not to use a bit map.
5460
5461 3. Brought error texts together as macro definitions; clarified some of them;
5462 fixed one that was wrong - it said "range out of order" when it meant "invalid
5463 escape sequence".
5464
5465 4. Changed some char * arguments to const char *.
5466
5467 5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
5468
5469 6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
5470 pcretest.
5471
5472
5473 Version 0.96 16-Oct-97
5474 ----------------------
5475
5476 1. Added a simple "pgrep" utility to the distribution.
5477
5478 2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
5479 unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
5480 where "ddd" means "one or more decimal digits".
5481
5482 3. Fixed serious bug. If a pattern had a back reference, but the call to
5483 pcre_exec() didn't supply a large enough ovector to record the related
5484 identifying subpattern, the match always failed. PCRE now remembers the number
5485 of the largest back reference, and gets some temporary memory in which to save
5486 the offsets during matching if necessary, in order to ensure that
5487 backreferences always work.
5488
5489 4. Increased the compatibility with Perl in a number of ways:
5490
5491 (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
5492 to request this handling. The option can be set at compile or exec time.
5493
5494 (b) $ matches before a terminating newline by default; an option
5495 PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
5496 mode). The option can be set at compile or exec time.
5497
5498 (c) The handling of \ followed by a digit other than 0 is now supposed to be
5499 the same as Perl's. If the decimal number it represents is less than 10
5500 or there aren't that many previous left capturing parentheses, an octal
5501 escape is read. Inside a character class, it's always an octal escape,
5502 even if it is a single digit.
5503
5504 (d) An escaped but undefined alphabetic character is taken as a literal,
5505 unless PCRE_EXTRA is set. Currently this just reserves the remaining
5506 escapes.
5507
5508 (e) {0} is now permitted. (The previous item is removed from the compiled
5509 pattern).
5510
5511 5. Changed all the names of code files so that the basic parts are no longer
5512 than 10 characters, and abolished the teeny "globals.c" file.
5513
5514 6. Changed the handling of character classes; they are now done with a 32-byte
5515 bit map always.
5516
5517 7. Added the -d and /D options to pcretest to make it possible to look at the
5518 internals of compilation without having to recompile pcre.
5519
5520
5521 Version 0.95 23-Sep-97
5522 ----------------------
5523
5524 1. Fixed bug in pre-pass concerning escaped "normal" characters such as \x5c or
5525 \x20 at the start of a run of normal characters. These were being treated as
5526 real characters, instead of the source characters being re-checked.
5527
5528
5529 Version 0.94 18-Sep-97
5530 ----------------------
5531
5532 1. The functions are now thread-safe, with the caveat that the global variables
5533 containing pointers to malloc() and free() or alternative functions are the
5534 same for all threads.
5535
5536 2. Get pcre_study() to generate a bitmap of initial characters for non-
5537 anchored patterns when this is possible, and use it if passed to pcre_exec().
5538
5539
5540 Version 0.93 15-Sep-97
5541 ----------------------
5542
5543 1. /(b)|(:+)/ was computing an incorrect first character.
5544
5545 2. Add pcre_study() to the API and the passing of pcre_extra to pcre_exec(),
5546 but not actually doing anything yet.
5547
5548 3. Treat "-" characters in classes that cannot be part of ranges as literals,
5549 as Perl does (e.g. [-az] or [az-]).
5550
5551 4. Set the anchored flag if a branch starts with .* or .*? because that tests
5552 all possible positions.
5553
5554 5. Split up into different modules to avoid including unneeded functions in a
5555 compiled binary. However, compile and exec are still in one module. The "study"
5556 function is split off.
5557
5558 6. The character tables are now in a separate module whose source is generated
5559 by an auxiliary program - but can then be edited by hand if required. There are
5560 now no calls to isalnum(), isspace(), isdigit(), isxdigit(), tolower() or
5561 toupper() in the code.
5562
5563 7. Turn the malloc/free funtions variables into pcre_malloc and pcre_free and
5564 make them global. Abolish the function for setting them, as the caller can now
5565 set them directly.
5566
5567
5568 Version 0.92 11-Sep-97
5569 ----------------------
5570
5571 1. A repeat with a fixed maximum and a minimum of 1 for an ordinary character
5572 (e.g. /a{1,3}/) was broken (I mis-optimized it).
5573
5574 2. Caseless matching was not working in character classes if the characters in
5575 the pattern were in upper case.
5576
5577 3. Make ranges like [W-c] work in the same way as Perl for caseless matching.
5578
5579 4. Make PCRE_ANCHORED public and accept as a compile option.
5580
5581 5. Add an options word to pcre_exec() and accept PCRE_ANCHORED and
5582 PCRE_CASELESS at run time. Add escapes \A and \I to pcretest to cause it to
5583 pass them.
5584
5585 6. Give an error if bad option bits passed at compile or run time.
5586
5587 7. Add PCRE_MULTILINE at compile and exec time, and (?m) as well. Add \M to
5588 pcretest to cause it to pass that flag.
5589
5590 8. Add pcre_info(), to get the number of identifying subpatterns, the stored
5591 options, and the first character, if set.
5592
5593 9. Recognize C+ or C{n,m} where n >= 1 as providing a fixed starting character.
5594
5595
5596 Version 0.91 10-Sep-97
5597 ----------------------
5598
5599 1. PCRE was failing to diagnose unlimited repeats of subpatterns that could
5600 match the empty string as in /(a*)*/. It was looping and ultimately crashing.
5601
5602 2. PCRE was looping on encountering an indefinitely repeated back reference to
5603 a subpattern that had matched an empty string, e.g. /(a|)\1*/. It now does what
5604 Perl does - treats the match as successful.
5605
5606 ****

Properties

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

  ViewVC Help
Powered by ViewVC 1.1.5