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

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1513 - (show annotations) (download)
Wed Nov 19 20:57:13 2014 UTC (3 days, 7 hours ago) by ph10
File size: 269139 byte(s)
Fix zero-repeat assertion condition bug.

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

Properties

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

webmaster@exim.org
ViewVC Help
Powered by ViewVC 1.1.12