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

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1529 - (show annotations) (download)
Mon Mar 2 17:09:58 2015 UTC (46 hours, 3 minutes ago) by ph10
File size: 273146 byte(s)
Fix bad compilation for patterns like /((?+1)(\1))/ with forward reference 
subroutine and recursive back reference within the same group.

1 ChangeLog for PCRE
2 ------------------
3
4 Version 8.37 xx-xxx-2015
5 ------------------------
6
7 1. When an (*ACCEPT) is triggered inside capturing parentheses, it arranges
8 for those parentheses to be closed with whatever has been captured so far.
9 However, it was failing to mark any other groups between the hightest
10 capture so far and the currrent group as "unset". Thus, the ovector for
11 those groups contained whatever was previously there. An example is the
12 pattern /(x)|((*ACCEPT))/ when matched against "abcd".
13
14 2. If an assertion condition was quantified with a minimum of zero (an odd
15 thing to do, but it happened), SIGSEGV or other misbehaviour could occur.
16
17 3. If a pattern in pcretest input had the P (POSIX) modifier followed by an
18 unrecognized modifier, a crash could occur.
19
20 4. An attempt to do global matching in pcretest with a zero-length ovector
21 caused a crash.
22
23 5. Fixed a memory leak during matching that could occur for a subpattern
24 subroutine call (recursive or otherwise) if the number of captured groups
25 that had to be saved was greater than ten.
26
27 6. Catch a bad opcode during auto-possessification after compiling a bad UTF
28 string with NO_UTF_CHECK. This is a tidyup, not a bug fix, as passing bad
29 UTF with NO_UTF_CHECK is documented as having an undefined outcome.
30
31 7. A UTF pattern containing a "not" match of a non-ASCII character and a
32 subroutine reference could loop at compile time. Example: /[^\xff]((?1))/.
33
34 8. When a pattern is compiled, it remembers the highest back reference so that
35 when matching, if the ovector is too small, extra memory can be obtained to
36 use instead. A conditional subpattern whose condition is a check on a
37 capture having happened, such as, for example in the pattern
38 /^(?:(a)|b)(?(1)A|B)/, is another kind of back reference, but it was not
39 setting the highest backreference number. This mattered only if pcre_exec()
40 was called with an ovector that was too small to hold the capture, and there
41 was no other kind of back reference (a situation which is probably quite
42 rare). The effect of the bug was that the condition was always treated as
43 FALSE when the capture could not be consulted, leading to a incorrect
44 behaviour by pcre2_match(). This bug has been fixed.
45
46 9. A reference to a duplicated named group (either a back reference or a test
47 for being set in a conditional) that occurred in a part of the pattern where
48 PCRE_DUPNAMES was not set caused the amount of memory needed for the pattern
49 to be incorrectly calculated, leading to overwriting.
50
51 10. A mutually recursive set of back references such as (\2)(\1) caused a
52 segfault at study time (while trying to find the minimum matching length).
53 The infinite loop is now broken (with the minimum length unset, that is,
54 zero).
55
56 11. If an assertion that was used as a condition was quantified with a minimum
57 of zero, matching went wrong. In particular, if the whole group had
58 unlimited repetition and could match an empty string, a segfault was
59 likely. The pattern (?(?=0)?)+ is an example that caused this. Perl allows
60 assertions to be quantified, but not if they are being used as conditions,
61 so the above pattern is faulted by Perl. PCRE has now been changed so that
62 it also rejects such patterns.
63
64 12. A possessive capturing group such as (a)*+ with a minimum repeat of zero
65 failed to allow the zero-repeat case if pcre2_exec() was called with an
66 ovector too small to capture the group.
67
68 13. Fixed two bugs in pcretest that were discovered by fuzzing and reported by
69 Red Hat Product Security:
70
71 (a) A crash if /K and /F were both set with the option to save the compiled
72 pattern.
73
74 (b) Another crash if the option to print captured substrings in a callout
75 was combined with setting a null ovector, for example \O\C+ as a subject
76 string.
77
78 14. A pattern such as "((?2){0,1999}())?", which has a group containing a
79 forward reference repeated a large (but limited) number of times within a
80 repeated outer group that has a zero minimum quantifier, caused incorrect
81 code to be compiled, leading to the error "internal error:
82 previously-checked referenced subpattern not found" when an incorrect
83 memory address was read. This bug was reported as "heap overflow",
84 discovered by Kai Lu of Fortinet's FortiGuard Labs.
85
86 23. A pattern such as "((?+1)(\1))/" containing a forward reference subroutine
87 call within a group that also contained a recursive back reference caused
88 incorrect code to be compiled. This bug was reported as "heap overflow",
89 discovered by Kai Lu of Fortinet's FortiGuard Labs.
90
91
92 Version 8.36 26-September-2014
93 ------------------------------
94
95 1. Got rid of some compiler warnings in the C++ modules that were shown up by
96 -Wmissing-field-initializers and -Wunused-parameter.
97
98 2. The tests for quantifiers being too big (greater than 65535) were being
99 applied after reading the number, and stupidly assuming that integer
100 overflow would give a negative number. The tests are now applied as the
101 numbers are read.
102
103 3. Tidy code in pcre_exec.c where two branches that used to be different are
104 now the same.
105
106 4. The JIT compiler did not generate match limit checks for certain
107 bracketed expressions with quantifiers. This may lead to exponential
108 backtracking, instead of returning with PCRE_ERROR_MATCHLIMIT. This
109 issue should be resolved now.
110
111 5. Fixed an issue, which occures when nested alternatives are optimized
112 with table jumps.
113
114 6. Inserted two casts and changed some ints to size_t in the light of some
115 reported 64-bit compiler warnings (Bugzilla 1477).
116
117 7. Fixed a bug concerned with zero-minimum possessive groups that could match
118 an empty string, which sometimes were behaving incorrectly in the
119 interpreter (though correctly in the JIT matcher). This pcretest input is
120 an example:
121
122 '\A(?:[^"]++|"(?:[^"]*+|"")*+")++'
123 NON QUOTED "QUOT""ED" AFTER "NOT MATCHED
124
125 the interpreter was reporting a match of 'NON QUOTED ' only, whereas the
126 JIT matcher and Perl both matched 'NON QUOTED "QUOT""ED" AFTER '. The test
127 for an empty string was breaking the inner loop and carrying on at a lower
128 level, when possessive repeated groups should always return to a higher
129 level as they have no backtrack points in them. The empty string test now
130 occurs at the outer level.
131
132 8. Fixed a bug that was incorrectly auto-possessifying \w+ in the pattern
133 ^\w+(?>\s*)(?<=\w) which caused it not to match "test test".
134
135 9. Give a compile-time error for \o{} (as Perl does) and for \x{} (which Perl
136 doesn't).
137
138 10. Change 8.34/15 introduced a bug that caused the amount of memory needed
139 to hold a pattern to be incorrectly computed (too small) when there were
140 named back references to duplicated names. This could cause "internal
141 error: code overflow" or "double free or corruption" or other memory
142 handling errors.
143
144 11. When named subpatterns had the same prefixes, back references could be
145 confused. For example, in this pattern:
146
147 /(?P<Name>a)?(?P<Name2>b)?(?(<Name>)c|d)*l/
148
149 the reference to 'Name' was incorrectly treated as a reference to a
150 duplicate name.
151
152 12. A pattern such as /^s?c/mi8 where the optional character has more than
153 one "other case" was incorrectly compiled such that it would only try to
154 match starting at "c".
155
156 13. When a pattern starting with \s was studied, VT was not included in the
157 list of possible starting characters; this should have been part of the
158 8.34/18 patch.
159
160 14. If a character class started [\Qx]... where x is any character, the class
161 was incorrectly terminated at the ].
162
163 15. If a pattern that started with a caseless match for a character with more
164 than one "other case" was studied, PCRE did not set up the starting code
165 unit bit map for the list of possible characters. Now it does. This is an
166 optimization improvement, not a bug fix.
167
168 16. The Unicode data tables have been updated to Unicode 7.0.0.
169
170 17. Fixed a number of memory leaks in pcregrep.
171
172 18. Avoid a compiler warning (from some compilers) for a function call with
173 a cast that removes "const" from an lvalue by using an intermediate
174 variable (to which the compiler does not object).
175
176 19. Incorrect code was compiled if a group that contained an internal recursive
177 back reference was optional (had quantifier with a minimum of zero). This
178 example compiled incorrect code: /(((a\2)|(a*)\g<-1>))*/ and other examples
179 caused segmentation faults because of stack overflows at compile time.
180
181 20. A pattern such as /((?(R)a|(?1)))+/, which contains a recursion within a
182 group that is quantified with an indefinite repeat, caused a compile-time
183 loop which used up all the system stack and provoked a segmentation fault.
184 This was not the same bug as 19 above.
185
186 21. Add PCRECPP_EXP_DECL declaration to operator<< in pcre_stringpiece.h.
187 Patch by Mike Frysinger.
188
189
190 Version 8.35 04-April-2014
191 --------------------------
192
193 1. A new flag is set, when property checks are present in an XCLASS.
194 When this flag is not set, PCRE can perform certain optimizations
195 such as studying these XCLASS-es.
196
197 2. The auto-possessification of character sets were improved: a normal
198 and an extended character set can be compared now. Furthermore
199 the JIT compiler optimizes more character set checks.
200
201 3. Got rid of some compiler warnings for potentially uninitialized variables
202 that show up only when compiled with -O2.
203
204 4. A pattern such as (?=ab\K) that uses \K in an assertion can set the start
205 of a match later then the end of the match. The pcretest program was not
206 handling the case sensibly - it was outputting from the start to the next
207 binary zero. It now reports this situation in a message, and outputs the
208 text from the end to the start.
209
210 5. Fast forward search is improved in JIT. Instead of the first three
211 characters, any three characters with fixed position can be searched.
212 Search order: first, last, middle.
213
214 6. Improve character range checks in JIT. Characters are read by an inprecise
215 function now, which returns with an unknown value if the character code is
216 above a certain threshold (e.g: 256). The only limitation is that the value
217 must be bigger than the threshold as well. This function is useful when
218 the characters above the threshold are handled in the same way.
219
220 7. The macros whose names start with RAWUCHAR are placeholders for a future
221 mode in which only the bottom 21 bits of 32-bit data items are used. To
222 make this more memorable for those maintaining the code, the names have
223 been changed to start with UCHAR21, and an extensive comment has been added
224 to their definition.
225
226 8. Add missing (new) files sljitNativeTILEGX.c and sljitNativeTILEGX-encoder.c
227 to the export list in Makefile.am (they were accidentally omitted from the
228 8.34 tarball).
229
230 9. The informational output from pcretest used the phrase "starting byte set"
231 which is inappropriate for the 16-bit and 32-bit libraries. As the output
232 for "first char" and "need char" really means "non-UTF-char", I've changed
233 "byte" to "char", and slightly reworded the output. The documentation about
234 these values has also been (I hope) clarified.
235
236 10. Another JIT related optimization: use table jumps for selecting the correct
237 backtracking path, when more than four alternatives are present inside a
238 bracket.
239
240 11. Empty match is not possible, when the minimum length is greater than zero,
241 and there is no \K in the pattern. JIT should avoid empty match checks in
242 such cases.
243
244 12. In a caseless character class with UCP support, when a character with more
245 than one alternative case was not the first character of a range, not all
246 the alternative cases were added to the class. For example, s and \x{17f}
247 are both alternative cases for S: the class [RST] was handled correctly,
248 but [R-T] was not.
249
250 13. The configure.ac file always checked for pthread support when JIT was
251 enabled. This is not used in Windows, so I have put this test inside a
252 check for the presence of windows.h (which was already tested for).
253
254 14. Improve pattern prefix search by a simplified Boyer-Moore algorithm in JIT.
255 The algorithm provides a way to skip certain starting offsets, and usually
256 faster than linear prefix searches.
257
258 15. Change 13 for 8.20 updated RunTest to check for the 'fr' locale as well
259 as for 'fr_FR' and 'french'. For some reason, however, it then used the
260 Windows-specific input and output files, which have 'french' screwed in.
261 So this could never have worked. One of the problems with locales is that
262 they aren't always the same. I have now updated RunTest so that it checks
263 the output of the locale test (test 3) against three different output
264 files, and it allows the test to pass if any one of them matches. With luck
265 this should make the test pass on some versions of Solaris where it was
266 failing. Because of the uncertainty, the script did not used to stop if
267 test 3 failed; it now does. If further versions of a French locale ever
268 come to light, they can now easily be added.
269
270 16. If --with-pcregrep-bufsize was given a non-integer value such as "50K",
271 there was a message during ./configure, but it did not stop. This now
272 provokes an error. The invalid example in README has been corrected.
273 If a value less than the minimum is given, the minimum value has always
274 been used, but now a warning is given.
275
276 17. If --enable-bsr-anycrlf was set, the special 16/32-bit test failed. This
277 was a bug in the test system, which is now fixed. Also, the list of various
278 configurations that are tested for each release did not have one with both
279 16/32 bits and --enable-bar-anycrlf. It now does.
280
281 18. pcretest was missing "-C bsr" for displaying the \R default setting.
282
283 19. Little endian PowerPC systems are supported now by the JIT compiler.
284
285 20. The fast forward newline mechanism could enter to an infinite loop on
286 certain invalid UTF-8 input. Although we don't support these cases
287 this issue can be fixed by a performance optimization.
288
289 21. Change 33 of 8.34 is not sufficient to ensure stack safety because it does
290 not take account if existing stack usage. There is now a new global
291 variable called pcre_stack_guard that can be set to point to an external
292 function to check stack availability. It is called at the start of
293 processing every parenthesized group.
294
295 22. A typo in the code meant that in ungreedy mode the max/min qualifier
296 behaved like a min-possessive qualifier, and, for example, /a{1,3}b/U did
297 not match "ab".
298
299 23. When UTF was disabled, the JIT program reported some incorrect compile
300 errors. These messages are silenced now.
301
302 24. Experimental support for ARM-64 and MIPS-64 has been added to the JIT
303 compiler.
304
305 25. Change all the temporary files used in RunGrepTest to be different to those
306 used by RunTest so that the tests can be run simultaneously, for example by
307 "make -j check".
308
309
310 Version 8.34 15-December-2013
311 -----------------------------
312
313 1. Add pcre[16|32]_jit_free_unused_memory to forcibly free unused JIT
314 executable memory. Patch inspired by Carsten Klein.
315
316 2. ./configure --enable-coverage defined SUPPORT_GCOV in config.h, although
317 this macro is never tested and has no effect, because the work to support
318 coverage involves only compiling and linking options and special targets in
319 the Makefile. The comment in config.h implied that defining the macro would
320 enable coverage support, which is totally false. There was also support for
321 setting this macro in the CMake files (my fault, I just copied it from
322 configure). SUPPORT_GCOV has now been removed.
323
324 3. Make a small performance improvement in strlen16() and strlen32() in
325 pcretest.
326
327 4. Change 36 for 8.33 left some unreachable statements in pcre_exec.c,
328 detected by the Solaris compiler (gcc doesn't seem to be able to diagnose
329 these cases). There was also one in pcretest.c.
330
331 5. Cleaned up a "may be uninitialized" compiler warning in pcre_exec.c.
332
333 6. In UTF mode, the code for checking whether a group could match an empty
334 string (which is used for indefinitely repeated groups to allow for
335 breaking an infinite loop) was broken when the group contained a repeated
336 negated single-character class with a character that occupied more than one
337 data item and had a minimum repetition of zero (for example, [^\x{100}]* in
338 UTF-8 mode). The effect was undefined: the group might or might not be
339 deemed as matching an empty string, or the program might have crashed.
340
341 7. The code for checking whether a group could match an empty string was not
342 recognizing that \h, \H, \v, \V, and \R must match a character.
343
344 8. Implemented PCRE_INFO_MATCH_EMPTY, which yields 1 if the pattern can match
345 an empty string. If it can, pcretest shows this in its information output.
346
347 9. Fixed two related bugs that applied to Unicode extended grapheme clusters
348 that were repeated with a maximizing qualifier (e.g. \X* or \X{2,5}) when
349 matched by pcre_exec() without using JIT:
350
351 (a) If the rest of the pattern did not match after a maximal run of
352 grapheme clusters, the code for backing up to try with fewer of them
353 did not always back up over a full grapheme when characters that do not
354 have the modifier quality were involved, e.g. Hangul syllables.
355
356 (b) If the match point in a subject started with modifier character, and
357 there was no match, the code could incorrectly back up beyond the match
358 point, and potentially beyond the first character in the subject,
359 leading to a segfault or an incorrect match result.
360
361 10. A conditional group with an assertion condition could lead to PCRE
362 recording an incorrect first data item for a match if no other first data
363 item was recorded. For example, the pattern (?(?=ab)ab) recorded "a" as a
364 first data item, and therefore matched "ca" after "c" instead of at the
365 start.
366
367 11. Change 40 for 8.33 (allowing pcregrep to find empty strings) showed up a
368 bug that caused the command "echo a | ./pcregrep -M '|a'" to loop.
369
370 12. The source of pcregrep now includes z/OS-specific code so that it can be
371 compiled for z/OS as part of the special z/OS distribution.
372
373 13. Added the -T and -TM options to pcretest.
374
375 14. The code in pcre_compile.c for creating the table of named capturing groups
376 has been refactored. Instead of creating the table dynamically during the
377 actual compiling pass, the information is remembered during the pre-compile
378 pass (on the stack unless there are more than 20 named groups, in which
379 case malloc() is used) and the whole table is created before the actual
380 compile happens. This has simplified the code (it is now nearly 150 lines
381 shorter) and prepared the way for better handling of references to groups
382 with duplicate names.
383
384 15. A back reference to a named subpattern when there is more than one of the
385 same name now checks them in the order in which they appear in the pattern.
386 The first one that is set is used for the reference. Previously only the
387 first one was inspected. This change makes PCRE more compatible with Perl.
388
389 16. Unicode character properties were updated from Unicode 6.3.0.
390
391 17. The compile-time code for auto-possessification has been refactored, based
392 on a patch by Zoltan Herczeg. It now happens after instead of during
393 compilation. The code is cleaner, and more cases are handled. The option
394 PCRE_NO_AUTO_POSSESS is added for testing purposes, and the -O and /O
395 options in pcretest are provided to set it. It can also be set by
396 (*NO_AUTO_POSSESS) at the start of a pattern.
397
398 18. The character VT has been added to the default ("C" locale) set of
399 characters that match \s and are generally treated as white space,
400 following this same change in Perl 5.18. There is now no difference between
401 "Perl space" and "POSIX space". Whether VT is treated as white space in
402 other locales depends on the locale.
403
404 19. The code for checking named groups as conditions, either for being set or
405 for being recursed, has been refactored (this is related to 14 and 15
406 above). Processing unduplicated named groups should now be as fast at
407 numerical groups, and processing duplicated groups should be faster than
408 before.
409
410 20. Two patches to the CMake build system, by Alexander Barkov:
411
412 (1) Replace the "source" command by "." in CMakeLists.txt because
413 "source" is a bash-ism.
414
415 (2) Add missing HAVE_STDINT_H and HAVE_INTTYPES_H to config-cmake.h.in;
416 without these the CMake build does not work on Solaris.
417
418 21. Perl has changed its handling of \8 and \9. If there is no previously
419 encountered capturing group of those numbers, they are treated as the
420 literal characters 8 and 9 instead of a binary zero followed by the
421 literals. PCRE now does the same.
422
423 22. Following Perl, added \o{} to specify codepoints in octal, making it
424 possible to specify values greater than 0777 and also making them
425 unambiguous.
426
427 23. Perl now gives an error for missing closing braces after \x{... instead of
428 treating the string as literal. PCRE now does the same.
429
430 24. RunTest used to grumble if an inappropriate test was selected explicitly,
431 but just skip it when running all tests. This make it awkward to run ranges
432 of tests when one of them was inappropriate. Now it just skips any
433 inappropriate tests, as it always did when running all tests.
434
435 25. If PCRE_AUTO_CALLOUT and PCRE_UCP were set for a pattern that contained
436 character types such as \d or \w, too many callouts were inserted, and the
437 data that they returned was rubbish.
438
439 26. In UCP mode, \s was not matching two of the characters that Perl matches,
440 namely NEL (U+0085) and MONGOLIAN VOWEL SEPARATOR (U+180E), though they
441 were matched by \h. The code has now been refactored so that the lists of
442 the horizontal and vertical whitespace characters used for \h and \v (which
443 are defined only in one place) are now also used for \s.
444
445 27. Add JIT support for the 64 bit TileGX architecture.
446 Patch by Jiong Wang (Tilera Corporation).
447
448 28. Possessive quantifiers for classes (both explicit and automatically
449 generated) now use special opcodes instead of wrapping in ONCE brackets.
450
451 29. Whereas an item such as A{4}+ ignored the possessivenes of the quantifier
452 (because it's meaningless), this was not happening when PCRE_CASELESS was
453 set. Not wrong, but inefficient.
454
455 30. Updated perltest.pl to add /u (force Unicode mode) when /W (use Unicode
456 properties for \w, \d, etc) is present in a test regex. Otherwise if the
457 test contains no characters greater than 255, Perl doesn't realise it
458 should be using Unicode semantics.
459
460 31. Upgraded the handling of the POSIX classes [:graph:], [:print:], and
461 [:punct:] when PCRE_UCP is set so as to include the same characters as Perl
462 does in Unicode mode.
463
464 32. Added the "forbid" facility to pcretest so that putting tests into the
465 wrong test files can sometimes be quickly detected.
466
467 33. There is now a limit (default 250) on the depth of nesting of parentheses.
468 This limit is imposed to control the amount of system stack used at compile
469 time. It can be changed at build time by --with-parens-nest-limit=xxx or
470 the equivalent in CMake.
471
472 34. Character classes such as [A-\d] or [a-[:digit:]] now cause compile-time
473 errors. Perl warns for these when in warning mode, but PCRE has no facility
474 for giving warnings.
475
476 35. Change 34 for 8.13 allowed quantifiers on assertions, because Perl does.
477 However, this was not working for (?!) because it is optimized to (*FAIL),
478 for which PCRE does not allow quantifiers. The optimization is now disabled
479 when a quantifier follows (?!). I can't see any use for this, but it makes
480 things uniform.
481
482 36. Perl no longer allows group names to start with digits, so I have made this
483 change also in PCRE. It simplifies the code a bit.
484
485 37. In extended mode, Perl ignores spaces before a + that indicates a
486 possessive quantifier. PCRE allowed a space before the quantifier, but not
487 before the possessive +. It now does.
488
489 38. The use of \K (reset reported match start) within a repeated possessive
490 group such as (a\Kb)*+ was not working.
491
492 40. Document that the same character tables must be used at compile time and
493 run time, and that the facility to pass tables to pcre_exec() and
494 pcre_dfa_exec() is for use only with saved/restored patterns.
495
496 41. Applied Jeff Trawick's patch CMakeLists.txt, which "provides two new
497 features for Builds with MSVC:
498
499 1. Support pcre.rc and/or pcreposix.rc (as is already done for MinGW
500 builds). The .rc files can be used to set FileDescription and many other
501 attributes.
502
503 2. Add an option (-DINSTALL_MSVC_PDB) to enable installation of .pdb files.
504 This allows higher-level build scripts which want .pdb files to avoid
505 hard-coding the exact files needed."
506
507 42. Added support for [[:<:]] and [[:>:]] as used in the BSD POSIX library to
508 mean "start of word" and "end of word", respectively, as a transition aid.
509
510 43. A minimizing repeat of a class containing codepoints greater than 255 in
511 non-UTF 16-bit or 32-bit modes caused an internal error when PCRE was
512 compiled to use the heap for recursion.
513
514 44. Got rid of some compiler warnings for unused variables when UTF but not UCP
515 is configured.
516
517
518 Version 8.33 28-May-2013
519 ------------------------
520
521 1. Added 'U' to some constants that are compared to unsigned integers, to
522 avoid compiler signed/unsigned warnings. Added (int) casts to unsigned
523 variables that are added to signed variables, to ensure the result is
524 signed and can be negated.
525
526 2. Applied patch by Daniel Richard G for quashing MSVC warnings to the
527 CMake config files.
528
529 3. Revise the creation of config.h.generic so that all boolean macros are
530 #undefined, whereas non-boolean macros are #ifndef/#endif-ed. This makes
531 overriding via -D on the command line possible.
532
533 4. Changing the definition of the variable "op" in pcre_exec.c from pcre_uchar
534 to unsigned int is reported to make a quite noticeable speed difference in
535 a specific Windows environment. Testing on Linux did also appear to show
536 some benefit (and it is clearly not harmful). Also fixed the definition of
537 Xop which should be unsigned.
538
539 5. Related to (4), changing the definition of the intermediate variable cc
540 in repeated character loops from pcre_uchar to pcre_uint32 also gave speed
541 improvements.
542
543 6. Fix forward search in JIT when link size is 3 or greater. Also removed some
544 unnecessary spaces.
545
546 7. Adjust autogen.sh and configure.ac to lose warnings given by automake 1.12
547 and later.
548
549 8. Fix two buffer over read issues in 16 and 32 bit modes. Affects JIT only.
550
551 9. Optimizing fast_forward_start_bits in JIT.
552
553 10. Adding support for callouts in JIT, and fixing some issues revealed
554 during this work. Namely:
555
556 (a) Unoptimized capturing brackets incorrectly reset on backtrack.
557
558 (b) Minimum length was not checked before the matching is started.
559
560 11. The value of capture_last that is passed to callouts was incorrect in some
561 cases when there was a capture on one path that was subsequently abandoned
562 after a backtrack. Also, the capture_last value is now reset after a
563 recursion, since all captures are also reset in this case.
564
565 12. The interpreter no longer returns the "too many substrings" error in the
566 case when an overflowing capture is in a branch that is subsequently
567 abandoned after a backtrack.
568
569 13. In the pathological case when an offset vector of size 2 is used, pcretest
570 now prints out the matched string after a yield of 0 or 1.
571
572 14. Inlining subpatterns in recursions, when certain conditions are fulfilled.
573 Only supported by the JIT compiler at the moment.
574
575 15. JIT compiler now supports 32 bit Macs thanks to Lawrence Velazquez.
576
577 16. Partial matches now set offsets[2] to the "bumpalong" value, that is, the
578 offset of the starting point of the matching process, provided the offsets
579 vector is large enough.
580
581 17. The \A escape now records a lookbehind value of 1, though its execution
582 does not actually inspect the previous character. This is to ensure that,
583 in partial multi-segment matching, at least one character from the old
584 segment is retained when a new segment is processed. Otherwise, if there
585 are no lookbehinds in the pattern, \A might match incorrectly at the start
586 of a new segment.
587
588 18. Added some #ifdef __VMS code into pcretest.c to help VMS implementations.
589
590 19. Redefined some pcre_uchar variables in pcre_exec.c as pcre_uint32; this
591 gives some modest performance improvement in 8-bit mode.
592
593 20. Added the PCRE-specific property \p{Xuc} for matching characters that can
594 be expressed in certain programming languages using Universal Character
595 Names.
596
597 21. Unicode validation has been updated in the light of Unicode Corrigendum #9,
598 which points out that "non characters" are not "characters that may not
599 appear in Unicode strings" but rather "characters that are reserved for
600 internal use and have only local meaning".
601
602 22. When a pattern was compiled with automatic callouts (PCRE_AUTO_CALLOUT) and
603 there was a conditional group that depended on an assertion, if the
604 assertion was false, the callout that immediately followed the alternation
605 in the condition was skipped when pcre_exec() was used for matching.
606
607 23. Allow an explicit callout to be inserted before an assertion that is the
608 condition for a conditional group, for compatibility with automatic
609 callouts, which always insert a callout at this point.
610
611 24. In 8.31, (*COMMIT) was confined to within a recursive subpattern. Perl also
612 confines (*SKIP) and (*PRUNE) in the same way, and this has now been done.
613
614 25. (*PRUNE) is now supported by the JIT compiler.
615
616 26. Fix infinite loop when /(?<=(*SKIP)ac)a/ is matched against aa.
617
618 27. Fix the case where there are two or more SKIPs with arguments that may be
619 ignored.
620
621 28. (*SKIP) is now supported by the JIT compiler.
622
623 29. (*THEN) is now supported by the JIT compiler.
624
625 30. Update RunTest with additional test selector options.
626
627 31. The way PCRE handles backtracking verbs has been changed in two ways.
628
629 (1) Previously, in something like (*COMMIT)(*SKIP), COMMIT would override
630 SKIP. Now, PCRE acts on whichever backtracking verb is reached first by
631 backtracking. In some cases this makes it more Perl-compatible, but Perl's
632 rather obscure rules do not always do the same thing.
633
634 (2) Previously, backtracking verbs were confined within assertions. This is
635 no longer the case for positive assertions, except for (*ACCEPT). Again,
636 this sometimes improves Perl compatibility, and sometimes does not.
637
638 32. A number of tests that were in test 2 because Perl did things differently
639 have been moved to test 1, because either Perl or PCRE has changed, and
640 these tests are now compatible.
641
642 32. Backtracking control verbs are now handled in the same way in JIT and
643 interpreter.
644
645 33. An opening parenthesis in a MARK/PRUNE/SKIP/THEN name in a pattern that
646 contained a forward subroutine reference caused a compile error.
647
648 34. Auto-detect and optimize limited repetitions in JIT.
649
650 35. Implement PCRE_NEVER_UTF to lock out the use of UTF, in particular,
651 blocking (*UTF) etc.
652
653 36. In the interpreter, maximizing pattern repetitions for characters and
654 character types now use tail recursion, which reduces stack usage.
655
656 37. The value of the max lookbehind was not correctly preserved if a compiled
657 and saved regex was reloaded on a host of different endianness.
658
659 38. Implemented (*LIMIT_MATCH) and (*LIMIT_RECURSION). As part of the extension
660 of the compiled pattern block, expand the flags field from 16 to 32 bits
661 because it was almost full.
662
663 39. Try madvise first before posix_madvise.
664
665 40. Change 7 for PCRE 7.9 made it impossible for pcregrep to find empty lines
666 with a pattern such as ^$. It has taken 4 years for anybody to notice! The
667 original change locked out all matches of empty strings. This has been
668 changed so that one match of an empty string per line is recognized.
669 Subsequent searches on the same line (for colouring or for --only-matching,
670 for example) do not recognize empty strings.
671
672 41. Applied a user patch to fix a number of spelling mistakes in comments.
673
674 42. Data lines longer than 65536 caused pcretest to crash.
675
676 43. Clarified the data type for length and startoffset arguments for pcre_exec
677 and pcre_dfa_exec in the function-specific man pages, where they were
678 explicitly stated to be in bytes, never having been updated. I also added
679 some clarification to the pcreapi man page.
680
681 44. A call to pcre_dfa_exec() with an output vector size less than 2 caused
682 a segmentation fault.
683
684
685 Version 8.32 30-November-2012
686 -----------------------------
687
688 1. Improved JIT compiler optimizations for first character search and single
689 character iterators.
690
691 2. Supporting IBM XL C compilers for PPC architectures in the JIT compiler.
692 Patch by Daniel Richard G.
693
694 3. Single character iterator optimizations in the JIT compiler.
695
696 4. Improved JIT compiler optimizations for character ranges.
697
698 5. Rename the "leave" variable names to "quit" to improve WinCE compatibility.
699 Reported by Giuseppe D'Angelo.
700
701 6. The PCRE_STARTLINE bit, indicating that a match can occur only at the start
702 of a line, was being set incorrectly in cases where .* appeared inside
703 atomic brackets at the start of a pattern, or where there was a subsequent
704 *PRUNE or *SKIP.
705
706 7. Improved instruction cache flush for POWER/PowerPC.
707 Patch by Daniel Richard G.
708
709 8. Fixed a number of issues in pcregrep, making it more compatible with GNU
710 grep:
711
712 (a) There is now no limit to the number of patterns to be matched.
713
714 (b) An error is given if a pattern is too long.
715
716 (c) Multiple uses of --exclude, --exclude-dir, --include, and --include-dir
717 are now supported.
718
719 (d) --exclude-from and --include-from (multiple use) have been added.
720
721 (e) Exclusions and inclusions now apply to all files and directories, not
722 just to those obtained from scanning a directory recursively.
723
724 (f) Multiple uses of -f and --file-list are now supported.
725
726 (g) In a Windows environment, the default for -d has been changed from
727 "read" (the GNU grep default) to "skip", because otherwise the presence
728 of a directory in the file list provokes an error.
729
730 (h) The documentation has been revised and clarified in places.
731
732 9. Improve the matching speed of capturing brackets.
733
734 10. Changed the meaning of \X so that it now matches a Unicode extended
735 grapheme cluster.
736
737 11. Patch by Daniel Richard G to the autoconf files to add a macro for sorting
738 out POSIX threads when JIT support is configured.
739
740 12. Added support for PCRE_STUDY_EXTRA_NEEDED.
741
742 13. In the POSIX wrapper regcomp() function, setting re_nsub field in the preg
743 structure could go wrong in environments where size_t is not the same size
744 as int.
745
746 14. Applied user-supplied patch to pcrecpp.cc to allow PCRE_NO_UTF8_CHECK to be
747 set.
748
749 15. The EBCDIC support had decayed; later updates to the code had included
750 explicit references to (e.g.) \x0a instead of CHAR_LF. There has been a
751 general tidy up of EBCDIC-related issues, and the documentation was also
752 not quite right. There is now a test that can be run on ASCII systems to
753 check some of the EBCDIC-related things (but is it not a full test).
754
755 16. The new PCRE_STUDY_EXTRA_NEEDED option is now used by pcregrep, resulting
756 in a small tidy to the code.
757
758 17. Fix JIT tests when UTF is disabled and both 8 and 16 bit mode are enabled.
759
760 18. If the --only-matching (-o) option in pcregrep is specified multiple
761 times, each one causes appropriate output. For example, -o1 -o2 outputs the
762 substrings matched by the 1st and 2nd capturing parentheses. A separating
763 string can be specified by --om-separator (default empty).
764
765 19. Improving the first n character searches.
766
767 20. Turn case lists for horizontal and vertical white space into macros so that
768 they are defined only once.
769
770 21. This set of changes together give more compatible Unicode case-folding
771 behaviour for characters that have more than one other case when UCP
772 support is available.
773
774 (a) The Unicode property table now has offsets into a new table of sets of
775 three or more characters that are case-equivalent. The MultiStage2.py
776 script that generates these tables (the pcre_ucd.c file) now scans
777 CaseFolding.txt instead of UnicodeData.txt for character case
778 information.
779
780 (b) The code for adding characters or ranges of characters to a character
781 class has been abstracted into a generalized function that also handles
782 case-independence. In UTF-mode with UCP support, this uses the new data
783 to handle characters with more than one other case.
784
785 (c) A bug that is fixed as a result of (b) is that codepoints less than 256
786 whose other case is greater than 256 are now correctly matched
787 caselessly. Previously, the high codepoint matched the low one, but not
788 vice versa.
789
790 (d) The processing of \h, \H, \v, and \ in character classes now makes use
791 of the new class addition function, using character lists defined as
792 macros alongside the case definitions of 20 above.
793
794 (e) Caseless back references now work with characters that have more than
795 one other case.
796
797 (f) General caseless matching of characters with more than one other case
798 is supported.
799
800 22. Unicode character properties were updated from Unicode 6.2.0
801
802 23. Improved CMake support under Windows. Patch by Daniel Richard G.
803
804 24. Add support for 32-bit character strings, and UTF-32
805
806 25. Major JIT compiler update (code refactoring and bugfixing).
807 Experimental Sparc 32 support is added.
808
809 26. Applied a modified version of Daniel Richard G's patch to create
810 pcre.h.generic and config.h.generic by "make" instead of in the
811 PrepareRelease script.
812
813 27. Added a definition for CHAR_NULL (helpful for the z/OS port), and use it in
814 pcre_compile.c when checking for a zero character.
815
816 28. Introducing a native interface for JIT. Through this interface, the compiled
817 machine code can be directly executed. The purpose of this interface is to
818 provide fast pattern matching, so several sanity checks are not performed.
819 However, feature tests are still performed. The new interface provides
820 1.4x speedup compared to the old one.
821
822 29. If pcre_exec() or pcre_dfa_exec() was called with a negative value for
823 the subject string length, the error given was PCRE_ERROR_BADOFFSET, which
824 was confusing. There is now a new error PCRE_ERROR_BADLENGTH for this case.
825
826 30. In 8-bit UTF-8 mode, pcretest failed to give an error for data codepoints
827 greater than 0x7fffffff (which cannot be represented in UTF-8, even under
828 the "old" RFC 2279). Instead, it ended up passing a negative length to
829 pcre_exec().
830
831 31. Add support for GCC's visibility feature to hide internal functions.
832
833 32. Running "pcretest -C pcre8" or "pcretest -C pcre16" gave a spurious error
834 "unknown -C option" after outputting 0 or 1.
835
836 33. There is now support for generating a code coverage report for the test
837 suite in environments where gcc is the compiler and lcov is installed. This
838 is mainly for the benefit of the developers.
839
840 34. If PCRE is built with --enable-valgrind, certain memory regions are marked
841 unaddressable using valgrind annotations, allowing valgrind to detect
842 invalid memory accesses. This is mainly for the benefit of the developers.
843
844 25. (*UTF) can now be used to start a pattern in any of the three libraries.
845
846 26. Give configure error if --enable-cpp but no C++ compiler found.
847
848
849 Version 8.31 06-July-2012
850 -------------------------
851
852 1. Fixing a wrong JIT test case and some compiler warnings.
853
854 2. Removed a bashism from the RunTest script.
855
856 3. Add a cast to pcre_exec.c to fix the warning "unary minus operator applied
857 to unsigned type, result still unsigned" that was given by an MS compiler
858 on encountering the code "-sizeof(xxx)".
859
860 4. Partial matching support is added to the JIT compiler.
861
862 5. Fixed several bugs concerned with partial matching of items that consist
863 of more than one character:
864
865 (a) /^(..)\1/ did not partially match "aba" because checking references was
866 done on an "all or nothing" basis. This also applied to repeated
867 references.
868
869 (b) \R did not give a hard partial match if \r was found at the end of the
870 subject.
871
872 (c) \X did not give a hard partial match after matching one or more
873 characters at the end of the subject.
874
875 (d) When newline was set to CRLF, a pattern such as /a$/ did not recognize
876 a partial match for the string "\r".
877
878 (e) When newline was set to CRLF, the metacharacter "." did not recognize
879 a partial match for a CR character at the end of the subject string.
880
881 6. If JIT is requested using /S++ or -s++ (instead of just /S+ or -s+) when
882 running pcretest, the text "(JIT)" added to the output whenever JIT is
883 actually used to run the match.
884
885 7. Individual JIT compile options can be set in pcretest by following -s+[+]
886 or /S+[+] with a digit between 1 and 7.
887
888 8. OP_NOT now supports any UTF character not just single-byte ones.
889
890 9. (*MARK) control verb is now supported by the JIT compiler.
891
892 10. The command "./RunTest list" lists the available tests without actually
893 running any of them. (Because I keep forgetting what they all are.)
894
895 11. Add PCRE_INFO_MAXLOOKBEHIND.
896
897 12. Applied a (slightly modified) user-supplied patch that improves performance
898 when the heap is used for recursion (compiled with --disable-stack-for-
899 recursion). Instead of malloc and free for each heap frame each time a
900 logical recursion happens, frames are retained on a chain and re-used where
901 possible. This sometimes gives as much as 30% improvement.
902
903 13. As documented, (*COMMIT) is now confined to within a recursive subpattern
904 call.
905
906 14. As documented, (*COMMIT) is now confined to within a positive assertion.
907
908 15. It is now possible to link pcretest with libedit as an alternative to
909 libreadline.
910
911 16. (*COMMIT) control verb is now supported by the JIT compiler.
912
913 17. The Unicode data tables have been updated to Unicode 6.1.0.
914
915 18. Added --file-list option to pcregrep.
916
917 19. Added binary file support to pcregrep, including the -a, --binary-files,
918 -I, and --text options.
919
920 20. The madvise function is renamed for posix_madvise for QNX compatibility
921 reasons. Fixed by Giuseppe D'Angelo.
922
923 21. Fixed a bug for backward assertions with REVERSE 0 in the JIT compiler.
924
925 22. Changed the option for creating symbolic links for 16-bit man pages from
926 -s to -sf so that re-installing does not cause issues.
927
928 23. Support PCRE_NO_START_OPTIMIZE in JIT as (*MARK) support requires it.
929
930 24. Fixed a very old bug in pcretest that caused errors with restarted DFA
931 matches in certain environments (the workspace was not being correctly
932 retained). Also added to pcre_dfa_exec() a simple plausibility check on
933 some of the workspace data at the beginning of a restart.
934
935 25. \s*\R was auto-possessifying the \s* when it should not, whereas \S*\R
936 was not doing so when it should - probably a typo introduced by SVN 528
937 (change 8.10/14).
938
939 26. When PCRE_UCP was not set, \w+\x{c4} was incorrectly auto-possessifying the
940 \w+ when the character tables indicated that \x{c4} was a word character.
941 There were several related cases, all because the tests for doing a table
942 lookup were testing for characters less than 127 instead of 255.
943
944 27. If a pattern contains capturing parentheses that are not used in a match,
945 their slots in the ovector are set to -1. For those that are higher than
946 any matched groups, this happens at the end of processing. In the case when
947 there were back references that the ovector was too small to contain
948 (causing temporary malloc'd memory to be used during matching), and the
949 highest capturing number was not used, memory off the end of the ovector
950 was incorrectly being set to -1. (It was using the size of the temporary
951 memory instead of the true size.)
952
953 28. To catch bugs like 27 using valgrind, when pcretest is asked to specify an
954 ovector size, it uses memory at the end of the block that it has got.
955
956 29. Check for an overlong MARK name and give an error at compile time. The
957 limit is 255 for the 8-bit library and 65535 for the 16-bit library.
958
959 30. JIT compiler update.
960
961 31. JIT is now supported on jailbroken iOS devices. Thanks for Ruiger
962 Rill for the patch.
963
964 32. Put spaces around SLJIT_PRINT_D in the JIT compiler. Required by CXX11.
965
966 33. Variable renamings in the PCRE-JIT compiler. No functionality change.
967
968 34. Fixed typos in pcregrep: in two places there was SUPPORT_LIBZ2 instead of
969 SUPPORT_LIBBZ2. This caused a build problem when bzip2 but not gzip (zlib)
970 was enabled.
971
972 35. Improve JIT code generation for greedy plus quantifier.
973
974 36. When /((?:a?)*)*c/ or /((?>a?)*)*c/ was matched against "aac", it set group
975 1 to "aa" instead of to an empty string. The bug affected repeated groups
976 that could potentially match an empty string.
977
978 37. Optimizing single character iterators in JIT.
979
980 38. Wide characters specified with \uxxxx in JavaScript mode are now subject to
981 the same checks as \x{...} characters in non-JavaScript mode. Specifically,
982 codepoints that are too big for the mode are faulted, and in a UTF mode,
983 disallowed codepoints are also faulted.
984
985 39. If PCRE was compiled with UTF support, in three places in the DFA
986 matcher there was code that should only have been obeyed in UTF mode, but
987 was being obeyed unconditionally. In 8-bit mode this could cause incorrect
988 processing when bytes with values greater than 127 were present. In 16-bit
989 mode the bug would be provoked by values in the range 0xfc00 to 0xdc00. In
990 both cases the values are those that cannot be the first data item in a UTF
991 character. The three items that might have provoked this were recursions,
992 possessively repeated groups, and atomic groups.
993
994 40. Ensure that libpcre is explicitly listed in the link commands for pcretest
995 and pcregrep, because some OS require shared objects to be explicitly
996 passed to ld, causing the link step to fail if they are not.
997
998 41. There were two incorrect #ifdefs in pcre_study.c, meaning that, in 16-bit
999 mode, patterns that started with \h* or \R* might be incorrectly matched.
1000
1001
1002 Version 8.30 04-February-2012
1003 -----------------------------
1004
1005 1. Renamed "isnumber" as "is_a_number" because in some Mac environments this
1006 name is defined in ctype.h.
1007
1008 2. Fixed a bug in fixed-length calculation for lookbehinds that would show up
1009 only in quite long subpatterns.
1010
1011 3. Removed the function pcre_info(), which has been obsolete and deprecated
1012 since it was replaced by pcre_fullinfo() in February 2000.
1013
1014 4. For a non-anchored pattern, if (*SKIP) was given with a name that did not
1015 match a (*MARK), and the match failed at the start of the subject, a
1016 reference to memory before the start of the subject could occur. This bug
1017 was introduced by fix 17 of release 8.21.
1018
1019 5. A reference to an unset group with zero minimum repetition was giving
1020 totally wrong answers (in non-JavaScript-compatibility mode). For example,
1021 /(another)?(\1?)test/ matched against "hello world test". This bug was
1022 introduced in release 8.13.
1023
1024 6. Add support for 16-bit character strings (a large amount of work involving
1025 many changes and refactorings).
1026
1027 7. RunGrepTest failed on msys because \r\n was replaced by whitespace when the
1028 command "pattern=`printf 'xxx\r\njkl'`" was run. The pattern is now taken
1029 from a file.
1030
1031 8. Ovector size of 2 is also supported by JIT based pcre_exec (the ovector size
1032 rounding is not applied in this particular case).
1033
1034 9. The invalid Unicode surrogate codepoints U+D800 to U+DFFF are now rejected
1035 if they appear, or are escaped, in patterns.
1036
1037 10. Get rid of a number of -Wunused-but-set-variable warnings.
1038
1039 11. The pattern /(?=(*:x))(q|)/ matches an empty string, and returns the mark
1040 "x". The similar pattern /(?=(*:x))((*:y)q|)/ did not return a mark at all.
1041 Oddly, Perl behaves the same way. PCRE has been fixed so that this pattern
1042 also returns the mark "x". This bug applied to capturing parentheses,
1043 non-capturing parentheses, and atomic parentheses. It also applied to some
1044 assertions.
1045
1046 12. Stephen Kelly's patch to CMakeLists.txt allows it to parse the version
1047 information out of configure.ac instead of relying on pcre.h.generic, which
1048 is not stored in the repository.
1049
1050 13. Applied Dmitry V. Levin's patch for a more portable method for linking with
1051 -lreadline.
1052
1053 14. ZH added PCRE_CONFIG_JITTARGET; added its output to pcretest -C.
1054
1055 15. Applied Graycode's patch to put the top-level frame on the stack rather
1056 than the heap when not using the stack for recursion. This gives a
1057 performance improvement in many cases when recursion is not deep.
1058
1059 16. Experimental code added to "pcretest -C" to output the stack frame size.
1060
1061
1062 Version 8.21 12-Dec-2011
1063 ------------------------
1064
1065 1. Updating the JIT compiler.
1066
1067 2. JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
1068 are added as well.
1069
1070 3. Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
1071 PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
1072 calling _pcre_jit_exec. Some extra comments are added.
1073
1074 4. (*MARK) settings inside atomic groups that do not contain any capturing
1075 parentheses, for example, (?>a(*:m)), were not being passed out. This bug
1076 was introduced by change 18 for 8.20.
1077
1078 5. Supporting of \x, \U and \u in JavaScript compatibility mode based on the
1079 ECMA-262 standard.
1080
1081 6. Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
1082 erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
1083 This bug was probably introduced by change 9 of 8.13.
1084
1085 7. While fixing 6 above, I noticed that a number of other items were being
1086 incorrectly rejected as "not fixed length". This arose partly because newer
1087 opcodes had not been added to the fixed-length checking code. I have (a)
1088 corrected the bug and added tests for these items, and (b) arranged for an
1089 error to occur if an unknown opcode is encountered while checking for fixed
1090 length instead of just assuming "not fixed length". The items that were
1091 rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
1092 (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
1093 repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
1094
1095 8. A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
1096 being incorrectly compiled and would have given unpredicatble results.
1097
1098 9. A possessively repeated subpattern with minimum repeat count greater than
1099 one behaved incorrectly. For example, (A){2,}+ behaved as if it was
1100 (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
1101 the first (A) could occur when it should not.
1102
1103 10. Add a cast and remove a redundant test from the code.
1104
1105 11. JIT should use pcre_malloc/pcre_free for allocation.
1106
1107 12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
1108 best practice nowadays, and helps with cross-compiling. (If the exec_prefix
1109 is anything other than /usr, -L is still shown).
1110
1111 13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
1112
1113 14. Perl does not support \N without a following name in a [] class; PCRE now
1114 also gives an error.
1115
1116 15. If a forward reference was repeated with an upper limit of around 2000,
1117 it caused the error "internal error: overran compiling workspace". The
1118 maximum number of forward references (including repeats) was limited by the
1119 internal workspace, and dependent on the LINK_SIZE. The code has been
1120 rewritten so that the workspace expands (via pcre_malloc) if necessary, and
1121 the default depends on LINK_SIZE. There is a new upper limit (for safety)
1122 of around 200,000 forward references. While doing this, I also speeded up
1123 the filling in of repeated forward references.
1124
1125 16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
1126 incorrectly expecting the subject to contain another "a" after the start.
1127
1128 17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
1129 in the match, the SKIP should be ignored. This was not happening; instead
1130 the SKIP was being treated as NOMATCH. For patterns such as
1131 /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
1132 tested.
1133
1134 18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
1135 now much more compatible with Perl, in particular in cases where the result
1136 is a non-match for a non-anchored pattern. For example, if
1137 /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
1138 "m", where previously it did not return a name. A side effect of this
1139 change is that for partial matches, the last encountered mark name is
1140 returned, as for non matches. A number of tests that were previously not
1141 Perl-compatible have been moved into the Perl-compatible test files. The
1142 refactoring has had the pleasing side effect of removing one argument from
1143 the match() function, thus reducing its stack requirements.
1144
1145 19. If the /S+ option was used in pcretest to study a pattern using JIT,
1146 subsequent uses of /S (without +) incorrectly behaved like /S+.
1147
1148 21. Retrieve executable code size support for the JIT compiler and fixing
1149 some warnings.
1150
1151 22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
1152 not work when the shorter character appeared right at the end of the
1153 subject string.
1154
1155 23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
1156 systems.
1157
1158 24. Added PCRE_INFO_JITSIZE to pass on the value from (21) above, and also
1159 output it when the /M option is used in pcretest.
1160
1161 25. The CheckMan script was not being included in the distribution. Also, added
1162 an explicit "perl" to run Perl scripts from the PrepareRelease script
1163 because this is reportedly needed in Windows.
1164
1165 26. If study data was being save in a file and studying had not found a set of
1166 "starts with" bytes for the pattern, the data written to the file (though
1167 never used) was taken from uninitialized memory and so caused valgrind to
1168 complain.
1169
1170 27. Updated RunTest.bat as provided by Sheri Pierce.
1171
1172 28. Fixed a possible uninitialized memory bug in pcre_jit_compile.c.
1173
1174 29. Computation of memory usage for the table of capturing group names was
1175 giving an unnecessarily large value.
1176
1177
1178 Version 8.20 21-Oct-2011
1179 ------------------------
1180
1181 1. Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
1182 a POSIX class. After further experiments with Perl, which convinced me that
1183 Perl has bugs and confusions, a closing square bracket is no longer allowed
1184 in a POSIX name. This bug also affected patterns with classes that started
1185 with full stops.
1186
1187 2. If a pattern such as /(a)b|ac/ is matched against "ac", there is no
1188 captured substring, but while checking the failing first alternative,
1189 substring 1 is temporarily captured. If the output vector supplied to
1190 pcre_exec() was not big enough for this capture, the yield of the function
1191 was still zero ("insufficient space for captured substrings"). This cannot
1192 be totally fixed without adding another stack variable, which seems a lot
1193 of expense for a edge case. However, I have improved the situation in cases
1194 such as /(a)(b)x|abc/ matched against "abc", where the return code
1195 indicates that fewer than the maximum number of slots in the ovector have
1196 been set.
1197
1198 3. Related to (2) above: when there are more back references in a pattern than
1199 slots in the output vector, pcre_exec() uses temporary memory during
1200 matching, and copies in the captures as far as possible afterwards. It was
1201 using the entire output vector, but this conflicts with the specification
1202 that only 2/3 is used for passing back captured substrings. Now it uses
1203 only the first 2/3, for compatibility. This is, of course, another edge
1204 case.
1205
1206 4. Zoltan Herczeg's just-in-time compiler support has been integrated into the
1207 main code base, and can be used by building with --enable-jit. When this is
1208 done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
1209 runtime --no-jit option is given.
1210
1211 5. When the number of matches in a pcre_dfa_exec() run exactly filled the
1212 ovector, the return from the function was zero, implying that there were
1213 other matches that did not fit. The correct "exactly full" value is now
1214 returned.
1215
1216 6. If a subpattern that was called recursively or as a subroutine contained
1217 (*PRUNE) or any other control that caused it to give a non-standard return,
1218 invalid errors such as "Error -26 (nested recursion at the same subject
1219 position)" or even infinite loops could occur.
1220
1221 7. If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
1222 computing the minimum length on reaching *ACCEPT, and so ended up with the
1223 wrong value of 1 rather than 0. Further investigation indicates that
1224 computing a minimum subject length in the presence of *ACCEPT is difficult
1225 (think back references, subroutine calls), and so I have changed the code
1226 so that no minimum is registered for a pattern that contains *ACCEPT.
1227
1228 8. If (*THEN) was present in the first (true) branch of a conditional group,
1229 it was not handled as intended. [But see 16 below.]
1230
1231 9. Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
1232 Sheri Pierce.
1233
1234 10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
1235 the first byte in a match must be "a".
1236
1237 11. Change 17 for 8.13 increased the recursion depth for patterns like
1238 /a(?:.)*?a/ drastically. I've improved things by remembering whether a
1239 pattern contains any instances of (*THEN). If it does not, the old
1240 optimizations are restored. It would be nice to do this on a per-group
1241 basis, but at the moment that is not feasible.
1242
1243 12. In some environments, the output of pcretest -C is CRLF terminated. This
1244 broke RunTest's code that checks for the link size. A single white space
1245 character after the value is now allowed for.
1246
1247 13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
1248 For "fr", it uses the Windows-specific input and output files.
1249
1250 14. If (*THEN) appeared in a group that was called recursively or as a
1251 subroutine, it did not work as intended. [But see next item.]
1252
1253 15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
1254 pattern fragments (but not containing any | characters). If A and B are
1255 matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
1256 was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
1257 D. In other words, Perl considers parentheses that do not contain any |
1258 characters to be part of a surrounding alternative, whereas PCRE was
1259 treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
1260 differently. PCRE now behaves in the same way as Perl, except in the case
1261 of subroutine/recursion calls such as (?1) which have in any case always
1262 been different (but PCRE had them first :-).
1263
1264 16. Related to 15 above: Perl does not treat the | in a conditional group as
1265 creating alternatives. Such a group is treated in the same way as an
1266 ordinary group without any | characters when processing (*THEN). PCRE has
1267 been changed to match Perl's behaviour.
1268
1269 17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
1270 RunGrepTest script failed.
1271
1272 18. Change 22 for version 13 caused atomic groups to use more stack. This is
1273 inevitable for groups that contain captures, but it can lead to a lot of
1274 stack use in large patterns. The old behaviour has been restored for atomic
1275 groups that do not contain any capturing parentheses.
1276
1277 19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
1278 suppress the check for a minimum subject length at run time. (If it was
1279 given to pcre_exec() or pcre_dfa_exec() it did work.)
1280
1281 20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
1282 fail to work when decoding hex characters in data strings in EBCDIC
1283 environments.
1284
1285 21. It appears that in at least one Mac OS environment, the isxdigit() function
1286 is implemented as a macro that evaluates to its argument more than once,
1287 contravening the C 90 Standard (I haven't checked a later standard). There
1288 was an instance in pcretest which caused it to go wrong when processing
1289 \x{...} escapes in subject strings. The has been rewritten to avoid using
1290 things like p++ in the argument of isxdigit().
1291
1292
1293 Version 8.13 16-Aug-2011
1294 ------------------------
1295
1296 1. The Unicode data tables have been updated to Unicode 6.0.0.
1297
1298 2. Two minor typos in pcre_internal.h have been fixed.
1299
1300 3. Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
1301 pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
1302 in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
1303
1304 4. There were a number of related bugs in the code for matching backrefences
1305 caselessly in UTF-8 mode when codes for the characters concerned were
1306 different numbers of bytes. For example, U+023A and U+2C65 are an upper
1307 and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
1308 (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
1309 code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
1310 2-byte code at the end of the subject (it thought there wasn't enough data
1311 left).
1312
1313 5. Comprehensive information about what went wrong is now returned by
1314 pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
1315 as the output vector has at least 2 elements. The offset of the start of
1316 the failing character and a reason code are placed in the vector.
1317
1318 6. When the UTF-8 string check fails for pcre_compile(), the offset that is
1319 now returned is for the first byte of the failing character, instead of the
1320 last byte inspected. This is an incompatible change, but I hope it is small
1321 enough not to be a problem. It makes the returned offset consistent with
1322 pcre_exec() and pcre_dfa_exec().
1323
1324 7. pcretest now gives a text phrase as well as the error number when
1325 pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
1326 failure, the offset and reason code are output.
1327
1328 8. When \R was used with a maximizing quantifier it failed to skip backwards
1329 over a \r\n pair if the subsequent match failed. Instead, it just skipped
1330 back over a single character (\n). This seems wrong (because it treated the
1331 two characters as a single entity when going forwards), conflicts with the
1332 documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
1333 behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
1334 has been changed.
1335
1336 9. Some internal refactoring has changed the processing so that the handling
1337 of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
1338 time (the PCRE_DOTALL option was changed this way some time ago: version
1339 7.7 change 16). This has made it possible to abolish the OP_OPT op code,
1340 which was always a bit of a fudge. It also means that there is one less
1341 argument for the match() function, which reduces its stack requirements
1342 slightly. This change also fixes an incompatibility with Perl: the pattern
1343 (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
1344
1345 10. More internal refactoring has drastically reduced the number of recursive
1346 calls to match() for possessively repeated groups such as (abc)++ when
1347 using pcre_exec().
1348
1349 11. While implementing 10, a number of bugs in the handling of groups were
1350 discovered and fixed:
1351
1352 (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
1353 (a|)*(?1) gave a compile-time internal error.
1354 ((a|)+)+ did not notice that the outer group could match an empty string.
1355 (^a|^)+ was not marked as anchored.
1356 (.*a|.*)+ was not marked as matching at start or after a newline.
1357
1358 12. Yet more internal refactoring has removed another argument from the match()
1359 function. Special calls to this function are now indicated by setting a
1360 value in a variable in the "match data" data block.
1361
1362 13. Be more explicit in pcre_study() instead of relying on "default" for
1363 opcodes that mean there is no starting character; this means that when new
1364 ones are added and accidentally left out of pcre_study(), testing should
1365 pick them up.
1366
1367 14. The -s option of pcretest has been documented for ages as being an old
1368 synonym of -m (show memory usage). I have changed it to mean "force study
1369 for every regex", that is, assume /S for every regex. This is similar to -i
1370 and -d etc. It's slightly incompatible, but I'm hoping nobody is still
1371 using it. It makes it easier to run collections of tests with and without
1372 study enabled, and thereby test pcre_study() more easily. All the standard
1373 tests are now run with and without -s (but some patterns can be marked as
1374 "never study" - see 20 below).
1375
1376 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
1377 restoration of the capturing data to the outer values was not happening
1378 correctly.
1379
1380 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
1381 empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
1382 pattern had matched an empty string, and so incorrectly returned a no
1383 match.
1384
1385 17. There was optimizing code for the last branch of non-capturing parentheses,
1386 and also for the obeyed branch of a conditional subexpression, which used
1387 tail recursion to cut down on stack usage. Unfortunately, now that there is
1388 the possibility of (*THEN) occurring in these branches, tail recursion is
1389 no longer possible because the return has to be checked for (*THEN). These
1390 two optimizations have therefore been removed. [But see 8.20/11 above.]
1391
1392 18. If a pattern containing \R was studied, it was assumed that \R always
1393 matched two bytes, thus causing the minimum subject length to be
1394 incorrectly computed because \R can also match just one byte.
1395
1396 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
1397 was incorrectly computed.
1398
1399 20. If /S is present twice on a test pattern in pcretest input, it now
1400 *disables* studying, thereby overriding the use of -s on the command line
1401 (see 14 above). This is necessary for one or two tests to keep the output
1402 identical in both cases.
1403
1404 21. When (*ACCEPT) was used in an assertion that matched an empty string and
1405 PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
1406
1407 22. When an atomic group that contained a capturing parenthesis was
1408 successfully matched, but the branch in which it appeared failed, the
1409 capturing was not being forgotten if a higher numbered group was later
1410 captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
1411 group 1 to "a", when in fact it should be unset. This applied to multi-
1412 branched capturing and non-capturing groups, repeated or not, and also to
1413 positive assertions (capturing in negative assertions does not happen
1414 in PCRE) and also to nested atomic groups.
1415
1416 23. Add the ++ qualifier feature to pcretest, to show the remainder of the
1417 subject after a captured substring, to make it easier to tell which of a
1418 number of identical substrings has been captured.
1419
1420 24. The way atomic groups are processed by pcre_exec() has been changed so that
1421 if they are repeated, backtracking one repetition now resets captured
1422 values correctly. For example, if ((?>(a+)b)+aabab) is matched against
1423 "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
1424 "aaa". Previously, it would have been "a". As part of this code
1425 refactoring, the way recursive calls are handled has also been changed.
1426
1427 25. If an assertion condition captured any substrings, they were not passed
1428 back unless some other capturing happened later. For example, if
1429 (?(?=(a))a) was matched against "a", no capturing was returned.
1430
1431 26. When studying a pattern that contained subroutine calls or assertions,
1432 the code for finding the minimum length of a possible match was handling
1433 direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
1434 group 1 called group 2 while simultaneously a separate group 2 called group
1435 1). A stack overflow occurred in this case. I have fixed this by limiting
1436 the recursion depth to 10.
1437
1438 27. Updated RunTest.bat in the distribution to the version supplied by Tom
1439 Fortmann. This supports explicit test numbers on the command line, and has
1440 argument validation and error reporting.
1441
1442 28. An instance of \X with an unlimited repeat could fail if at any point the
1443 first character it looked at was a mark character.
1444
1445 29. Some minor code refactoring concerning Unicode properties and scripts
1446 should reduce the stack requirement of match() slightly.
1447
1448 30. Added the '=' option to pcretest to check the setting of unused capturing
1449 slots at the end of the pattern, which are documented as being -1, but are
1450 not included in the return count.
1451
1452 31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
1453 compiled something random. Now it gives a compile-time error (as does
1454 Perl).
1455
1456 32. A *MARK encountered during the processing of a positive assertion is now
1457 recorded and passed back (compatible with Perl).
1458
1459 33. If --only-matching or --colour was set on a pcregrep call whose pattern
1460 had alternative anchored branches, the search for a second match in a line
1461 was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
1462 matched the line "0102" twice. The same bug affected patterns that started
1463 with a backwards assertion. For example /\b01|\b02/ also matched "0102"
1464 twice.
1465
1466 34. Previously, PCRE did not allow quantification of assertions. However, Perl
1467 does, and because of capturing effects, quantifying parenthesized
1468 assertions may at times be useful. Quantifiers are now allowed for
1469 parenthesized assertions.
1470
1471 35. A minor code tidy in pcre_compile() when checking options for \R usage.
1472
1473 36. \g was being checked for fancy things in a character class, when it should
1474 just be a literal "g".
1475
1476 37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
1477 appearance of a nested POSIX class supersedes an apparent external class.
1478 For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
1479 unescaped square brackets may also appear as part of class names. For
1480 example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
1481 more like Perl. (But see 8.20/1 above.)
1482
1483 38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
1484 was because it thought it was \N{name}, which is not supported).
1485
1486 39. Add minix to OS list not supporting the -S option in pcretest.
1487
1488 40. PCRE tries to detect cases of infinite recursion at compile time, but it
1489 cannot analyze patterns in sufficient detail to catch mutual recursions
1490 such as ((?1))((?2)). There is now a runtime test that gives an error if a
1491 subgroup is called recursively as a subpattern for a second time at the
1492 same position in the subject string. In previous releases this might have
1493 been caught by the recursion limit, or it might have run out of stack.
1494
1495 41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
1496 happen only once. PCRE was, however incorrectly giving a compile time error
1497 "recursive call could loop indefinitely" because it cannot analyze the
1498 pattern in sufficient detail. The compile time test no longer happens when
1499 PCRE is compiling a conditional subpattern, but actual runaway loops are
1500 now caught at runtime (see 40 above).
1501
1502 42. It seems that Perl allows any characters other than a closing parenthesis
1503 to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
1504 has been changed to be the same.
1505
1506 43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
1507 as not to get warnings when autogen.sh is called. Also changed
1508 AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
1509
1510 44. To help people who use pcregrep to scan files containing exceedingly long
1511 lines, the following changes have been made:
1512
1513 (a) The default value of the buffer size parameter has been increased from
1514 8K to 20K. (The actual buffer used is three times this size.)
1515
1516 (b) The default can be changed by ./configure --with-pcregrep-bufsize when
1517 PCRE is built.
1518
1519 (c) A --buffer-size=n option has been added to pcregrep, to allow the size
1520 to be set at run time.
1521
1522 (d) Numerical values in pcregrep options can be followed by K or M, for
1523 example --buffer-size=50K.
1524
1525 (e) If a line being scanned overflows pcregrep's buffer, an error is now
1526 given and the return code is set to 2.
1527
1528 45. Add a pointer to the latest mark to the callout data block.
1529
1530 46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
1531 partial match of an empty string instead of no match. This was specific to
1532 the use of ".".
1533
1534 47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
1535 complete match instead of a partial match. This bug was dependent on both
1536 the PCRE_UTF8 and PCRE_DOTALL options being set.
1537
1538 48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
1539 starting byte set, because \b was not being ignored.
1540
1541
1542 Version 8.12 15-Jan-2011
1543 ------------------------
1544
1545 1. Fixed some typos in the markup of the man pages, and wrote a script that
1546 checks for such things as part of the documentation building process.
1547
1548 2. On a big-endian 64-bit system, pcregrep did not correctly process the
1549 --match-limit and --recursion-limit options (added for 8.11). In
1550 particular, this made one of the standard tests fail. (The integer value
1551 went into the wrong half of a long int.)
1552
1553 3. If the --colour option was given to pcregrep with -v (invert match), it
1554 did strange things, either producing crazy output, or crashing. It should,
1555 of course, ignore a request for colour when reporting lines that do not
1556 match.
1557
1558 4. Another pcregrep bug caused similar problems if --colour was specified with
1559 -M (multiline) and the pattern match finished with a line ending.
1560
1561 5. In pcregrep, when a pattern that ended with a literal newline sequence was
1562 matched in multiline mode, the following line was shown as part of the
1563 match. This seems wrong, so I have changed it.
1564
1565 6. Another pcregrep bug in multiline mode, when --colour was specified, caused
1566 the check for further matches in the same line (so they could be coloured)
1567 to overrun the end of the current line. If another match was found, it was
1568 incorrectly shown (and then shown again when found in the next line).
1569
1570 7. If pcregrep was compiled under Windows, there was a reference to the
1571 function pcregrep_exit() before it was defined. I am assuming this was
1572 the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
1573 reported by a user. I've moved the definition above the reference.
1574
1575
1576 Version 8.11 10-Dec-2010
1577 ------------------------
1578
1579 1. (*THEN) was not working properly if there were untried alternatives prior
1580 to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
1581 backtracked to try for "b" instead of moving to the next alternative branch
1582 at the same level (in this case, to look for "c"). The Perl documentation
1583 is clear that when (*THEN) is backtracked onto, it goes to the "next
1584 alternative in the innermost enclosing group".
1585
1586 2. (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
1587 such as (A(*COMMIT)B(*THEN)C|D) any failure after matching A should
1588 result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
1589 (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
1590 (*THEN).
1591
1592 3. If \s appeared in a character class, it removed the VT character from
1593 the class, even if it had been included by some previous item, for example
1594 in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
1595 of \s, but is part of the POSIX "space" class.)
1596
1597 4. A partial match never returns an empty string (because you can always
1598 match an empty string at the end of the subject); however the checking for
1599 an empty string was starting at the "start of match" point. This has been
1600 changed to the "earliest inspected character" point, because the returned
1601 data for a partial match starts at this character. This means that, for
1602 example, /(?<=abc)def/ gives a partial match for the subject "abc"
1603 (previously it gave "no match").
1604
1605 5. Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
1606 of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
1607 previously a full match would be given. However, setting PCRE_PARTIAL_HARD
1608 has an implication that the given string is incomplete (because a partial
1609 match is preferred over a full match). For this reason, these items now
1610 give a partial match in this situation. [Aside: previously, the one case
1611 /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
1612 match rather than a full match, which was wrong by the old rules, but is
1613 now correct.]
1614
1615 6. There was a bug in the handling of #-introduced comments, recognized when
1616 PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
1617 If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
1618 UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
1619 scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
1620 but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
1621 places in pcre_compile().
1622
1623 7. Related to (6) above, when pcre_compile() was skipping #-introduced
1624 comments when looking ahead for named forward references to subpatterns,
1625 the only newline sequence it recognized was NL. It now handles newlines
1626 according to the set newline convention.
1627
1628 8. SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
1629 former, but used strtoul(), whereas pcretest avoided strtoul() but did not
1630 cater for a lack of strerror(). These oversights have been fixed.
1631
1632 9. Added --match-limit and --recursion-limit to pcregrep.
1633
1634 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
1635
1636 11. When the -o option was used, pcregrep was setting a return code of 1, even
1637 when matches were found, and --line-buffered was not being honoured.
1638
1639 12. Added an optional parentheses number to the -o and --only-matching options
1640 of pcregrep.
1641
1642 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
1643 can match an empty string. The code to do it in pcretest and pcredemo
1644 needed fixing:
1645
1646 (a) When the newline convention was "crlf", pcretest got it wrong, skipping
1647 only one byte after an empty string match just before CRLF (this case
1648 just got forgotten; "any" and "anycrlf" were OK).
1649
1650 (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
1651 mode when an empty string match preceded an ASCII character followed by
1652 a non-ASCII character. (The code for advancing by one character rather
1653 than one byte was nonsense.)
1654
1655 (c) The pcredemo.c sample program did not have any code at all to handle
1656 the cases when CRLF is a valid newline sequence.
1657
1658 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
1659 as a starting offset was within the subject string. There is now a new
1660 error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
1661 negative or greater than the length of the string. In order to test this,
1662 pcretest is extended to allow the setting of negative starting offsets.
1663
1664 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
1665 starting offset points to the beginning of a UTF-8 character was
1666 unnecessarily clumsy. I tidied it up.
1667
1668 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
1669 bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
1670
1671 17. Nobody had reported that the --include_dir option, which was added in
1672 release 7.7 should have been called --include-dir (hyphen, not underscore)
1673 for compatibility with GNU grep. I have changed it to --include-dir, but
1674 left --include_dir as an undocumented synonym, and the same for
1675 --exclude-dir, though that is not available in GNU grep, at least as of
1676 release 2.5.4.
1677
1678 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
1679 characters from a string of bytes) have been redefined so as not to use
1680 loops, in order to improve performance in some environments. At the same
1681 time, I abstracted some of the common code into auxiliary macros to save
1682 repetition (this should not affect the compiled code).
1683
1684 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
1685 compile-time error is now given if \c is not followed by an ASCII
1686 character, that is, a byte less than 128. (In EBCDIC mode, the code is
1687 different, and any byte value is allowed.)
1688
1689 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
1690 START_OPTIMIZE option, which is now allowed at compile time - but just
1691 passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
1692 to pcregrep and other applications that have no direct access to PCRE
1693 options. The new /Y option in pcretest sets this option when calling
1694 pcre_compile().
1695
1696 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
1697 back references. Groups containing recursive back references were forced to
1698 be atomic by that change, but in the case of named groups, the amount of
1699 memory required was incorrectly computed, leading to "Failed: internal
1700 error: code overflow". This has been fixed.
1701
1702 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
1703 pcretest.c, to avoid build problems in some Borland environments.
1704
1705
1706 Version 8.10 25-Jun-2010
1707 ------------------------
1708
1709 1. Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
1710 THEN.
1711
1712 2. (*ACCEPT) was not working when inside an atomic group.
1713
1714 3. Inside a character class, \B is treated as a literal by default, but
1715 faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
1716 causes the error). The code is unchanged, but I tidied the documentation.
1717
1718 4. Inside a character class, PCRE always treated \R and \X as literals,
1719 whereas Perl faults them if its -w option is set. I have changed PCRE so
1720 that it faults them when PCRE_EXTRA is set.
1721
1722 5. Added support for \N, which always matches any character other than
1723 newline. (It is the same as "." when PCRE_DOTALL is not set.)
1724
1725 6. When compiling pcregrep with newer versions of gcc which may have
1726 FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
1727 declared with attribute warn_unused_result" were given. Just casting the
1728 result to (void) does not stop the warnings; a more elaborate fudge is
1729 needed. I've used a macro to implement this.
1730
1731 7. Minor change to pcretest.c to avoid a compiler warning.
1732
1733 8. Added four artifical Unicode properties to help with an option to make
1734 \s etc use properties (see next item). The new properties are: Xan
1735 (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
1736
1737 9. Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
1738 use Unicode properties. (*UCP) at the start of a pattern can be used to set
1739 this option. Modified pcretest to add /W to test this facility. Added
1740 REG_UCP to make it available via the POSIX interface.
1741
1742 10. Added --line-buffered to pcregrep.
1743
1744 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
1745 studied, and the match started with a letter with a code point greater than
1746 127 whose first byte was different to the first byte of the other case of
1747 the letter, the other case of this starting letter was not recognized
1748 (#976).
1749
1750 12. If a pattern that was studied started with a repeated Unicode property
1751 test, for example, \p{Nd}+, there was the theoretical possibility of
1752 setting up an incorrect bitmap of starting bytes, but fortunately it could
1753 not have actually happened in practice until change 8 above was made (it
1754 added property types that matched character-matching opcodes).
1755
1756 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
1757 possible starting bytes for non-anchored patterns.
1758
1759 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
1760 \R, and also a number of cases that involve Unicode properties, both
1761 explicit and implicit when PCRE_UCP is set.
1762
1763 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
1764 input, it could crash or give wrong results if characters with values
1765 greater than 0xc0 were present in the subject string. (Detail: it assumed
1766 UTF-8 input when processing these items.)
1767
1768 16. Added a lot of (int) casts to avoid compiler warnings in systems where
1769 size_t is 64-bit (#991).
1770
1771 17. Added a check for running out of memory when PCRE is compiled with
1772 --disable-stack-for-recursion (#990).
1773
1774 18. If the last data line in a file for pcretest does not have a newline on
1775 the end, a newline was missing in the output.
1776
1777 19. The default pcre_chartables.c file recognizes only ASCII characters (values
1778 less than 128) in its various bitmaps. However, there is a facility for
1779 generating tables according to the current locale when PCRE is compiled. It
1780 turns out that in some environments, 0x85 and 0xa0, which are Unicode space
1781 characters, are recognized by isspace() and therefore were getting set in
1782 these tables, and indeed these tables seem to approximate to ISO 8859. This
1783 caused a problem in UTF-8 mode when pcre_study() was used to create a list
1784 of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
1785 which of course cannot start UTF-8 characters. I have changed the code so
1786 that only real ASCII characters (less than 128) and the correct starting
1787 bytes for UTF-8 encodings are set for characters greater than 127 when in
1788 UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
1789 altogether.)
1790
1791 20. Added the /T option to pcretest so as to be able to run tests with non-
1792 standard character tables, thus making it possible to include the tests
1793 used for 19 above in the standard set of tests.
1794
1795 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
1796 reference to a subpattern the other side of a comment that contains an
1797 opening parenthesis caused either an internal compiling error, or a
1798 reference to the wrong subpattern.
1799
1800
1801 Version 8.02 19-Mar-2010
1802 ------------------------
1803
1804 1. The Unicode data tables have been updated to Unicode 5.2.0.
1805
1806 2. Added the option --libs-cpp to pcre-config, but only when C++ support is
1807 configured.
1808
1809 3. Updated the licensing terms in the pcregexp.pas file, as agreed with the
1810 original author of that file, following a query about its status.
1811
1812 4. On systems that do not have stdint.h (e.g. Solaris), check for and include
1813 inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
1814
1815 5. A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
1816 quantifier applied to a forward-referencing subroutine call, could compile
1817 incorrect code or give the error "internal error: previously-checked
1818 referenced subpattern not found".
1819
1820 6. Both MS Visual Studio and Symbian OS have problems with initializing
1821 variables to point to external functions. For these systems, therefore,
1822 pcre_malloc etc. are now initialized to local functions that call the
1823 relevant global functions.
1824
1825 7. There were two entries missing in the vectors called coptable and poptable
1826 in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
1827 I've fixed the data, and added a kludgy way of testing at compile time that
1828 the lengths are correct (equal to the number of opcodes).
1829
1830 8. Following on from 7, I added a similar kludge to check the length of the
1831 eint vector in pcreposix.c.
1832
1833 9. Error texts for pcre_compile() are held as one long string to avoid too
1834 much relocation at load time. To find a text, the string is searched,
1835 counting zeros. There was no check for running off the end of the string,
1836 which could happen if a new error number was added without updating the
1837 string.
1838
1839 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
1840
1841 11. \K was not working if it appeared in an atomic group or in a group that
1842 was called as a "subroutine", or in an assertion. Perl 5.11 documents that
1843 \K is "not well defined" if used in an assertion. PCRE now accepts it if
1844 the assertion is positive, but not if it is negative.
1845
1846 12. Change 11 fortuitously reduced the size of the stack frame used in the
1847 "match()" function of pcre_exec.c by one pointer. Forthcoming
1848 implementation of support for (*MARK) will need an extra pointer on the
1849 stack; I have reserved it now, so that the stack frame size does not
1850 decrease.
1851
1852 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
1853 item in branch that calls a recursion is a subroutine call - as in the
1854 second branch in the above example - was incorrectly given the compile-
1855 time error "recursive call could loop indefinitely" because pcre_compile()
1856 was not correctly checking the subroutine for matching a non-empty string.
1857
1858 14. The checks for overrunning compiling workspace could trigger after an
1859 overrun had occurred. This is a "should never occur" error, but it can be
1860 triggered by pathological patterns such as hundreds of nested parentheses.
1861 The checks now trigger 100 bytes before the end of the workspace.
1862
1863 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
1864
1865
1866 Version 8.01 19-Jan-2010
1867 ------------------------
1868
1869 1. If a pattern contained a conditional subpattern with only one branch (in
1870 particular, this includes all (*DEFINE) patterns), a call to pcre_study()
1871 computed the wrong minimum data length (which is of course zero for such
1872 subpatterns). This could cause incorrect "no match" results.
1873
1874 2. For patterns such as (?i)a(?-i)b|c where an option setting at the start of
1875 the pattern is reset in the first branch, pcre_compile() failed with
1876 "internal error: code overflow at offset...". This happened only when
1877 the reset was to the original external option setting. (An optimization
1878 abstracts leading options settings into an external setting, which was the
1879 cause of this.)
1880
1881 3. A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
1882 of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
1883 assertion pattern did not match (meaning that the assertion was true), it
1884 was incorrectly treated as false if the SKIP had been reached during the
1885 matching. This also applied to assertions used as conditions.
1886
1887 4. If an item that is not supported by pcre_dfa_exec() was encountered in an
1888 assertion subpattern, including such a pattern used as a condition,
1889 unpredictable results occurred, instead of the error return
1890 PCRE_ERROR_DFA_UITEM.
1891
1892 5. The C++ GlobalReplace function was not working like Perl for the special
1893 situation when an empty string is matched. It now does the fancy magic
1894 stuff that is necessary.
1895
1896 6. In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
1897 removed. (These were left over from very, very early versions of PCRE.)
1898
1899 7. Some cosmetic changes to the code to make life easier when compiling it
1900 as part of something else:
1901
1902 (a) Change DEBUG to PCRE_DEBUG.
1903
1904 (b) In pcre_compile(), rename the member of the "branch_chain" structure
1905 called "current" as "current_branch", to prevent a collision with the
1906 Linux macro when compiled as a kernel module.
1907
1908 (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
1909 prevent a collision with the Linux macro when compiled as a kernel
1910 module.
1911
1912 8. In pcre_compile() there are some checks for integer overflows that used to
1913 cast potentially large values to (double). This has been changed to that
1914 when building, a check for int64_t is made, and if it is found, it is used
1915 instead, thus avoiding the use of floating point arithmetic. (There is no
1916 other use of FP in PCRE.) If int64_t is not found, the fallback is to
1917 double.
1918
1919 9. Added two casts to avoid signed/unsigned warnings from VS Studio Express
1920 2005 (difference between two addresses compared to an unsigned value).
1921
1922 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
1923 custom one, because of the following reported problem in Windows:
1924
1925 - libbz2 uses the Pascal calling convention (WINAPI) for the functions
1926 under Win32.
1927 - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
1928 therefore missing the function definition.
1929 - The compiler thus generates a "C" signature for the test function.
1930 - The linker fails to find the "C" function.
1931 - PCRE fails to configure if asked to do so against libbz2.
1932
1933 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
1934 messages were output:
1935
1936 Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
1937 rerunning libtoolize, to keep the correct libtool macros in-tree.
1938 Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
1939
1940 I have done both of these things.
1941
1942 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
1943 most of the time, it *can* run out if it is given a pattern that contains a
1944 runaway infinite recursion. I updated the discussion in the pcrestack man
1945 page.
1946
1947 13. Now that we have gone to the x.xx style of version numbers, the minor
1948 version may start with zero. Using 08 or 09 is a bad idea because users
1949 might check the value of PCRE_MINOR in their code, and 08 or 09 may be
1950 interpreted as invalid octal numbers. I've updated the previous comment in
1951 configure.ac, and also added a check that gives an error if 08 or 09 are
1952 used.
1953
1954 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
1955 causing partial matching to fail when the end of the subject matched \W
1956 in a UTF-8 pattern where \W was quantified with a minimum of 3.
1957
1958 15. There were some discrepancies between the declarations in pcre_internal.h
1959 of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
1960 their definitions. The declarations used "const uschar *" and the
1961 definitions used USPTR. Even though USPTR is normally defined as "const
1962 unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
1963 reported that: "This difference in casting confuses some C++ compilers, for
1964 example, SunCC recognizes above declarations as different functions and
1965 generates broken code for hbpcre." I have changed the declarations to use
1966 USPTR.
1967
1968 16. GNU libtool is named differently on some systems. The autogen.sh script now
1969 tries several variants such as glibtoolize (MacOSX) and libtoolize1x
1970 (FreeBSD).
1971
1972 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
1973 (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
1974 comment: "Figure out how to create a longlong from a string: strtoll and
1975 equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
1976 instance, but it only takes 2 args instead of 3!"
1977
1978 18. A subtle bug concerned with back references has been fixed by a change of
1979 specification, with a corresponding code fix. A pattern such as
1980 ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
1981 refers, was giving matches when it shouldn't. For example, xa=xaaa would
1982 match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
1983 same bug. Such groups have to be quantified to be useful, or contained
1984 inside another quantified group. (If there's no repetition, the reference
1985 can never match.) The problem arises because, having left the group and
1986 moved on to the rest of the pattern, a later failure that backtracks into
1987 the group uses the captured value from the final iteration of the group
1988 rather than the correct earlier one. I have fixed this in PCRE by forcing
1989 any group that contains a reference to itself to be an atomic group; that
1990 is, there cannot be any backtracking into it once it has completed. This is
1991 similar to recursive and subroutine calls.
1992
1993
1994 Version 8.00 19-Oct-09
1995 ----------------------
1996
1997 1. The table for translating pcre_compile() error codes into POSIX error codes
1998 was out-of-date, and there was no check on the pcre_compile() error code
1999 being within the table. This could lead to an OK return being given in
2000 error.
2001
2002 2. Changed the call to open a subject file in pcregrep from fopen(pathname,
2003 "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
2004 in a Windows environment.
2005
2006 3. The pcregrep --count option prints the count for each file even when it is
2007 zero, as does GNU grep. However, pcregrep was also printing all files when
2008 --files-with-matches was added. Now, when both options are given, it prints
2009 counts only for those files that have at least one match. (GNU grep just
2010 prints the file name in this circumstance, but including the count seems
2011 more useful - otherwise, why use --count?) Also ensured that the
2012 combination -clh just lists non-zero counts, with no names.
2013
2014 4. The long form of the pcregrep -F option was incorrectly implemented as
2015 --fixed_strings instead of --fixed-strings. This is an incompatible change,
2016 but it seems right to fix it, and I didn't think it was worth preserving
2017 the old behaviour.
2018
2019 5. The command line items --regex=pattern and --regexp=pattern were not
2020 recognized by pcregrep, which required --regex pattern or --regexp pattern
2021 (with a space rather than an '='). The man page documented the '=' forms,
2022 which are compatible with GNU grep; these now work.
2023
2024 6. No libpcreposix.pc file was created for pkg-config; there was just
2025 libpcre.pc and libpcrecpp.pc. The omission has been rectified.
2026
2027 7. Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
2028 when UCP support is not needed, by modifying the Python script that
2029 generates it from Unicode data files. This should not matter if the module
2030 is correctly used as a library, but I received one complaint about 50K of
2031 unwanted data. My guess is that the person linked everything into his
2032 program rather than using a library. Anyway, it does no harm.
2033
2034 8. A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
2035 was a minimum greater than 1 for a wide character in a possessive
2036 repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
2037 which had an unlimited repeat of a nested, fixed maximum repeat of a wide
2038 character. Chaos in the form of incorrect output or a compiling loop could
2039 result.
2040
2041 9. The restrictions on what a pattern can contain when partial matching is
2042 requested for pcre_exec() have been removed. All patterns can now be
2043 partially matched by this function. In addition, if there are at least two
2044 slots in the offset vector, the offset of the earliest inspected character
2045 for the match and the offset of the end of the subject are set in them when
2046 PCRE_ERROR_PARTIAL is returned.
2047
2048 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
2049 synonymous with PCRE_PARTIAL, for backwards compatibility, and
2050 PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
2051 and may be more useful for multi-segment matching.
2052
2053 11. Partial matching with pcre_exec() is now more intuitive. A partial match
2054 used to be given if ever the end of the subject was reached; now it is
2055 given only if matching could not proceed because another character was
2056 needed. This makes a difference in some odd cases such as Z(*FAIL) with the
2057 string "Z", which now yields "no match" instead of "partial match". In the
2058 case of pcre_dfa_exec(), "no match" is given if every matching path for the
2059 final character ended with (*FAIL).
2060
2061 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
2062 if the pattern had a "must contain" character that was already found in the
2063 earlier partial match, unless partial matching was again requested. For
2064 example, with the pattern /dog.(body)?/, the "must contain" character is
2065 "g". If the first part-match was for the string "dog", restarting with
2066 "sbody" failed. This bug has been fixed.
2067
2068 13. The string returned by pcre_dfa_exec() after a partial match has been
2069 changed so that it starts at the first inspected character rather than the
2070 first character of the match. This makes a difference only if the pattern
2071 starts with a lookbehind assertion or \b or \B (\K is not supported by
2072 pcre_dfa_exec()). It's an incompatible change, but it makes the two
2073 matching functions compatible, and I think it's the right thing to do.
2074
2075 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
2076 so that the demonstration program is easily available in environments where
2077 PCRE has not been installed from source.
2078
2079 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
2080 libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
2081 library.
2082
2083 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
2084 It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
2085 is not the first non-POSIX option to be added. Clearly some people find
2086 these options useful.
2087
2088 17. If a caller to the POSIX matching function regexec() passes a non-zero
2089 value for nmatch with a NULL value for pmatch, the value of
2090 nmatch is forced to zero.
2091
2092 18. RunGrepTest did not have a test for the availability of the -u option of
2093 the diff command, as RunTest does. It now checks in the same way as
2094 RunTest, and also checks for the -b option.
2095
2096 19. If an odd number of negated classes containing just a single character
2097 interposed, within parentheses, between a forward reference to a named
2098 subpattern and the definition of the subpattern, compilation crashed with
2099 an internal error, complaining that it could not find the referenced
2100 subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
2101 [The bug was that it was starting one character too far in when skipping
2102 over the character class, thus treating the ] as data rather than
2103 terminating the class. This meant it could skip too much.]
2104
2105 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
2106 /g option in pcretest when the pattern contains \K, which makes it possible
2107 to have an empty string match not at the start, even when the pattern is
2108 anchored. Updated pcretest and pcredemo to use this option.
2109
2110 21. If the maximum number of capturing subpatterns in a recursion was greater
2111 than the maximum at the outer level, the higher number was returned, but
2112 with unset values at the outer level. The correct (outer level) value is
2113 now given.
2114
2115 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
2116 PCRE did not set those parentheses (unlike Perl). I have now found a way to
2117 make it do so. The string so far is captured, making this feature
2118 compatible with Perl.
2119
2120 23. The tests have been re-organized, adding tests 11 and 12, to make it
2121 possible to check the Perl 5.10 features against Perl 5.10.
2122
2123 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
2124 pattern matches a fixed length string. PCRE did not allow this; now it
2125 does. Neither allows recursion.
2126
2127 25. I finally figured out how to implement a request to provide the minimum
2128 length of subject string that was needed in order to match a given pattern.
2129 (It was back references and recursion that I had previously got hung up
2130 on.) This code has now been added to pcre_study(); it finds a lower bound
2131 to the length of subject needed. It is not necessarily the greatest lower
2132 bound, but using it to avoid searching strings that are too short does give
2133 some useful speed-ups. The value is available to calling programs via
2134 pcre_fullinfo().
2135
2136 26. While implementing 25, I discovered to my embarrassment that pcretest had
2137 not been passing the result of pcre_study() to pcre_dfa_exec(), so the
2138 study optimizations had never been tested with that matching function.
2139 Oops. What is worse, even when it was passed study data, there was a bug in
2140 pcre_dfa_exec() that meant it never actually used it. Double oops. There
2141 were also very few tests of studied patterns with pcre_dfa_exec().
2142
2143 27. If (?| is used to create subpatterns with duplicate numbers, they are now
2144 allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
2145 on the other side of the coin, they are no longer allowed to have different
2146 names, because these cannot be distinguished in PCRE, and this has caused
2147 confusion. (This is a difference from Perl.)
2148
2149 28. When duplicate subpattern names are present (necessarily with different
2150 numbers, as required by 27 above), and a test is made by name in a
2151 conditional pattern, either for a subpattern having been matched, or for
2152 recursion in such a pattern, all the associated numbered subpatterns are
2153 tested, and the overall condition is true if the condition is true for any
2154 one of them. This is the way Perl works, and is also more like the way
2155 testing by number works.
2156
2157
2158 Version 7.9 11-Apr-09
2159 ---------------------
2160
2161 1. When building with support for bzlib/zlib (pcregrep) and/or readline
2162 (pcretest), all targets were linked against these libraries. This included
2163 libpcre, libpcreposix, and libpcrecpp, even though they do not use these
2164 libraries. This caused unwanted dependencies to be created. This problem
2165 has been fixed, and now only pcregrep is linked with bzlib/zlib and only
2166 pcretest is linked with readline.
2167
2168 2. The "typedef int BOOL" in pcre_internal.h that was included inside the
2169 "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
2170 moved outside it again, because FALSE and TRUE are already defined in AIX,
2171 but BOOL is not.
2172
2173 3. The pcre_config() function was treating the PCRE_MATCH_LIMIT and
2174 PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
2175
2176 4. The pcregrep documentation said spaces were inserted as well as colons (or
2177 hyphens) following file names and line numbers when outputting matching
2178 lines. This is not true; no spaces are inserted. I have also clarified the
2179 wording for the --colour (or --color) option.
2180
2181 5. In pcregrep, when --colour was used with -o, the list of matching strings
2182 was not coloured; this is different to GNU grep, so I have changed it to be
2183 the same.
2184
2185 6. When --colo(u)r was used in pcregrep, only the first matching substring in
2186 each matching line was coloured. Now it goes on to look for further matches
2187 of any of the test patterns, which is the same behaviour as GNU grep.
2188
2189 7. A pattern that could match an empty string could cause pcregrep to loop; it
2190 doesn't make sense to accept an empty string match in pcregrep, so I have
2191 locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
2192 seems to be how GNU grep behaves. [But see later change 40 for release
2193 8.33.]
2194
2195 8. The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
2196 start or after a newline", because the conditional assertion was not being
2197 correctly handled. The rule now is that both the assertion and what follows
2198 in the first alternative must satisfy the test.
2199
2200 9. If auto-callout was enabled in a pattern with a conditional group whose
2201 condition was an assertion, PCRE could crash during matching, both with
2202 pcre_exec() and pcre_dfa_exec().
2203
2204 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
2205 used for matching.
2206
2207 11. Unicode property support in character classes was not working for
2208 characters (bytes) greater than 127 when not in UTF-8 mode.
2209
2210 12. Added the -M command line option to pcretest.
2211
2212 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
2213
2214 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
2215
2216 16. Added comments and documentation about mis-use of no_arg in the C++
2217 wrapper.
2218
2219 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
2220 from Martin Jerabek that uses macro names for all relevant character and
2221 string constants.
2222
2223 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
2224 SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
2225 SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
2226 these, but not everybody uses configure.
2227
2228 19. A conditional group that had only one branch was not being correctly
2229 recognized as an item that could match an empty string. This meant that an
2230 enclosing group might also not be so recognized, causing infinite looping
2231 (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
2232 with the subject "ab", where knowledge that the repeated group can match
2233 nothing is needed in order to break the loop.
2234
2235 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
2236 exec(), but without supplying a callout function, matching went wrong.
2237
2238 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
2239 leak if the size of the offset vector was greater than 30. When the vector
2240 is smaller, the saved offsets during recursion go onto a local stack
2241 vector, but for larger vectors malloc() is used. It was failing to free
2242 when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
2243 error, in fact).
2244
2245 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
2246 heapframe that is used only when UTF-8 support is enabled. This caused no
2247 problem, but was untidy.
2248
2249 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
2250 CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
2251 included within another project.
2252
2253 24. Steven Van Ingelgem's patches to add more options to the CMake support,
2254 slightly modified by me:
2255
2256 (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
2257 not building pcregrep.
2258
2259 (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
2260 if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
2261
2262 25. Forward references, both numeric and by name, in patterns that made use of
2263 duplicate group numbers, could behave incorrectly or give incorrect errors,
2264 because when scanning forward to find the reference group, PCRE was not
2265 taking into account the duplicate group numbers. A pattern such as
2266 ^X(?3)(a)(?|(b)|(q))(Y) is an example.
2267
2268 26. Changed a few more instances of "const unsigned char *" to USPTR, making
2269 the feature of a custom pointer more persuasive (as requested by a user).
2270
2271 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
2272 pcretest.c, inside #ifndefs, because it seems they are sometimes already
2273 pre-defined.
2274
2275 28. Added support for (*UTF8) at the start of a pattern.
2276
2277 29. Arrange for flags added by the "release type" setting in CMake to be shown
2278 in the configuration summary.
2279
2280
2281 Version 7.8 05-Sep-08
2282 ---------------------
2283
2284 1. Replaced UCP searching code with optimized version as implemented for Ad
2285 Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
2286 stage table and inline lookup instead of a function, giving speed ups of 2
2287 to 5 times on some simple patterns that I tested. Permission was given to
2288 distribute the MultiStage2.py script that generates the tables (it's not in
2289 the tarball, but is in the Subversion repository).
2290
2291 2. Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
2292 scripts.
2293
2294 3. Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
2295 a group with a zero qualifier. The result of the study could be incorrect,
2296 or the function might crash, depending on the pattern.
2297
2298 4. Caseless matching was not working for non-ASCII characters in back
2299 references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
2300 It now works when Unicode Property Support is available.
2301
2302 5. In pcretest, an escape such as \x{de} in the data was always generating
2303 a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
2304 non-UTF-8 mode. If the value is greater than 255, it gives a warning about
2305 truncation.
2306
2307 6. Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
2308
2309 7. Added two (int) casts to pcregrep when printing the difference of two
2310 pointers, in case they are 64-bit values.
2311
2312 8. Added comments about Mac OS X stack usage to the pcrestack man page and to
2313 test 2 if it fails.
2314
2315 9. Added PCRE_CALL_CONVENTION just before the names of all exported functions,
2316 and a #define of that name to empty if it is not externally set. This is to
2317 allow users of MSVC to set it if necessary.
2318
2319 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
2320 the convenience functions in the pcre_get.c source file.
2321
2322 11. An option change at the start of a pattern that had top-level alternatives
2323 could cause overwriting and/or a crash. This command provoked a crash in
2324 some environments:
2325
2326 printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
2327
2328 This potential security problem was recorded as CVE-2008-2371.
2329
2330 12. For a pattern where the match had to start at the beginning or immediately
2331 after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
2332 pcre_dfa_exec() could read past the end of the passed subject if there was
2333 no match. To help with detecting such bugs (e.g. with valgrind), I modified
2334 pcretest so that it places the subject at the end of its malloc-ed buffer.
2335
2336 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
2337 exec() might read past the end of the data buffer in UTF-8 mode.
2338
2339 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
2340 the data contained the byte 0x85 as part of a UTF-8 character within its
2341 first line. This applied both to normal and DFA matching.
2342
2343 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
2344 /^[^d]*?$/8 failed to match "abc".
2345
2346 16. Added a missing copyright notice to pcrecpp_internal.h.
2347
2348 17. Make it more clear in the documentation that values returned from
2349 pcre_exec() in ovector are byte offsets, not character counts.
2350
2351 18. Tidied a few places to stop certain compilers from issuing warnings.
2352
2353 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
2354 supplied by Stefan Weber. I made a further small update for 7.8 because
2355 there is a change of source arrangements: the pcre_searchfuncs.c module is
2356 replaced by pcre_ucd.c.
2357
2358
2359 Version 7.7 07-May-08
2360 ---------------------
2361
2362 1. Applied Craig's patch to sort out a long long problem: "If we can't convert
2363 a string to a long long, pretend we don't even have a long long." This is
2364 done by checking for the strtoq, strtoll, and _strtoi64 functions.
2365
2366 2. Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
2367 pre-7.6 versions, which defined a global no_arg variable instead of putting
2368 it in the RE class. (See also #8 below.)
2369
2370 3. Remove a line of dead code, identified by coverity and reported by Nuno
2371 Lopes.
2372
2373 4. Fixed two related pcregrep bugs involving -r with --include or --exclude:
2374
2375 (1) The include/exclude patterns were being applied to the whole pathnames
2376 of files, instead of just to the final components.
2377
2378 (2) If there was more than one level of directory, the subdirectories were
2379 skipped unless they satisfied the include/exclude conditions. This is
2380 inconsistent with GNU grep (and could even be seen as contrary to the
2381 pcregrep specification - which I improved to make it absolutely clear).
2382 The action now is always to scan all levels of directory, and just
2383 apply the include/exclude patterns to regular files.
2384
2385 5. Added the --include_dir and --exclude_dir patterns to pcregrep, and used
2386 --exclude_dir in the tests to avoid scanning .svn directories.
2387
2388 6. Applied Craig's patch to the QuoteMeta function so that it escapes the
2389 NUL character as backslash + 0 rather than backslash + NUL, because PCRE
2390 doesn't support NULs in patterns.
2391
2392 7. Added some missing "const"s to declarations of static tables in
2393 pcre_compile.c and pcre_dfa_exec.c.
2394
2395 8. Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
2396 caused by fix #2 above. (Subsequently also a second patch to fix the
2397 first patch. And a third patch - this was a messy problem.)
2398
2399 9. Applied Craig's patch to remove the use of push_back().
2400
2401 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
2402 matching function regexec().
2403
2404 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
2405 which, however, unlike Perl's \g{...}, are subroutine calls, not back
2406 references. PCRE supports relative numbers with this syntax (I don't think
2407 Oniguruma does).
2408
2409 12. Previously, a group with a zero repeat such as (...){0} was completely
2410 omitted from the compiled regex. However, this means that if the group
2411 was called as a subroutine from elsewhere in the pattern, things went wrong
2412 (an internal error was given). Such groups are now left in the compiled
2413 pattern, with a new opcode that causes them to be skipped at execution
2414 time.
2415
2416 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
2417 to the way PCRE behaves:
2418
2419 (a) A lone ] character is dis-allowed (Perl treats it as data).
2420
2421 (b) A back reference to an unmatched subpattern matches an empty string
2422 (Perl fails the current match path).
2423
2424 (c) A data ] in a character class must be notated as \] because if the
2425 first data character in a class is ], it defines an empty class. (In
2426 Perl it is not possible to have an empty class.) The empty class []
2427 never matches; it forces failure and is equivalent to (*FAIL) or (?!).
2428 The negative empty class [^] matches any one character, independently
2429 of the DOTALL setting.
2430
2431 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
2432 non-existent subpattern following a character class starting with ']' and
2433 containing () gave an internal compiling error instead of "reference to
2434 non-existent subpattern". Fortunately, when the pattern did exist, the
2435 compiled code was correct. (When scanning forwards to check for the
2436 existence of the subpattern, it was treating the data ']' as terminating
2437 the class, so got the count wrong. When actually compiling, the reference
2438 was subsequently set up correctly.)
2439
2440 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
2441 it was being rejected as not supported by pcre_dfa_exec(), even though
2442 other assertions are supported. I have made pcre_dfa_exec() support
2443 (*FAIL).
2444
2445 16. The implementation of 13c above involved the invention of a new opcode,
2446 OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
2447 cannot be changed at match time, I realized I could make a small
2448 improvement to matching performance by compiling OP_ALLANY instead of
2449 OP_ANY for "." when DOTALL was set, and then removing the runtime tests
2450 on the OP_ANY path.
2451
2452 17. Compiling pcretest on Windows with readline support failed without the
2453 following two fixes: (1) Make the unistd.h include conditional on
2454 HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
2455
2456 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
2457 ncurses library to be included for pcretest when ReadLine support is
2458 requested, but also to allow for it to be overridden. This patch came from
2459 Daniel Bergström.
2460
2461 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
2462 as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
2463 any errors with the current Unicode tables. Thanks to Peter Kankowski for
2464 spotting this.
2465
2466
2467 Version 7.6 28-Jan-08
2468 ---------------------
2469
2470 1. A character class containing a very large number of characters with
2471 codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
2472 overflow.
2473
2474 2. Patch to cut out the "long long" test in pcrecpp_unittest when
2475 HAVE_LONG_LONG is not defined.
2476
2477 3. Applied Christian Ehrlicher's patch to update the CMake build files to
2478 bring them up to date and include new features. This patch includes:
2479
2480 - Fixed PH's badly added libz and libbz2 support.
2481 - Fixed a problem with static linking.
2482 - Added pcredemo. [But later removed - see 7 below.]
2483 - Fixed dftables problem and added an option.
2484 - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
2485 HAVE_LONG_LONG.
2486 - Added readline support for pcretest.
2487 - Added an listing of the option settings after cmake has run.
2488
2489 4. A user submitted a patch to Makefile that makes it easy to create
2490 "pcre.dll" under mingw when using Configure/Make. I added stuff to
2491 Makefile.am that cause it to include this special target, without
2492 affecting anything else. Note that the same mingw target plus all
2493 the other distribution libraries and programs are now supported
2494 when configuring with CMake (see 6 below) instead of with
2495 Configure/Make.
2496
2497 5. Applied Craig's patch that moves no_arg into the RE class in the C++ code.
2498 This is an attempt to solve the reported problem "pcrecpp::no_arg is not
2499 exported in the Windows port". It has not yet been confirmed that the patch
2500 solves the problem, but it does no harm.
2501
2502 6. Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
2503 NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
2504 with CMake, and also correct the comment about stack recursion.
2505
2506 7. Remove the automatic building of pcredemo from the ./configure system and
2507 from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
2508 of a program that users should build themselves after PCRE is installed, so
2509 building it automatically is not really right. What is more, it gave
2510 trouble in some build environments.
2511
2512 8. Further tidies to CMakeLists.txt from Sheri and Christian.
2513
2514
2515 Version 7.5 10-Jan-08
2516 ---------------------
2517
2518 1. Applied a patch from Craig: "This patch makes it possible to 'ignore'
2519 values in parens when parsing an RE using the C++ wrapper."
2520
2521 2. Negative specials like \S did not work in character classes in UTF-8 mode.
2522 Characters greater than 255 were excluded from the class instead of being
2523 included.
2524
2525 3. The same bug as (2) above applied to negated POSIX classes such as
2526 [:^space:].
2527
2528 4. PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
2529 defined or documented. It seems to have been a typo for PCRE_STATIC, so
2530 I have changed it.
2531
2532 5. The construct (?&) was not diagnosed as a syntax error (it referenced the
2533 first named subpattern) and a construct such as (?&a) would reference the
2534 first named subpattern whose name started with "a" (in other words, the
2535 length check was missing). Both these problems are fixed. "Subpattern name
2536 expected" is now given for (?&) (a zero-length name), and this patch also
2537 makes it give the same error for \k'' (previously it complained that that
2538 was a reference to a non-existent subpattern).
2539
2540 6. The erroneous patterns (?+-a) and (?-+a) give different error messages;
2541 this is right because (?- can be followed by option settings as well as by
2542 digits. I have, however, made the messages clearer.
2543
2544 7. Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
2545 than the number used in the conditional) now cause a compile-time error.
2546 This is actually not compatible with Perl, which accepts such patterns, but
2547 treats the conditional as always being FALSE (as PCRE used to), but it
2548 seems to me that giving a diagnostic is better.
2549
2550 8. Change "alphameric" to the more common word "alphanumeric" in comments
2551 and messages.
2552
2553 9. Fix two occurrences of "backslash" in comments that should have been
2554 "backspace".
2555
2556 10. Remove two redundant lines of code that can never be obeyed (their function
2557 was moved elsewhere).
2558
2559 11. The program that makes PCRE's Unicode character property table had a bug
2560 which caused it to generate incorrect table entries for sequences of
2561 characters that have the same character type, but are in different scripts.
2562 It amalgamated them into a single range, with the script of the first of
2563 them. In other words, some characters were in the wrong script. There were
2564 thirteen such cases, affecting characters in the following ranges:
2565
2566 U+002b0 - U+002c1
2567 U+0060c - U+0060d
2568 U+0061e - U+00612
2569 U+0064b - U+0065e
2570 U+0074d - U+0076d
2571 U+01800 - U+01805
2572 U+01d00 - U+01d77
2573 U+01d9b - U+01dbf
2574 U+0200b - U+0200f
2575 U+030fc - U+030fe
2576 U+03260 - U+0327f
2577 U+0fb46 - U+0fbb1
2578 U+10450 - U+1049d
2579
2580 12. The -o option (show only the matching part of a line) for pcregrep was not
2581 compatible with GNU grep in that, if there was more than one match in a
2582 line, it showed only the first of them. It now behaves in the same way as
2583 GNU grep.
2584
2585 13. If the -o and -v options were combined for pcregrep, it printed a blank
2586 line for every non-matching line. GNU grep prints nothing, and pcregrep now
2587 does the same. The return code can be used to tell if there were any
2588 non-matching lines.
2589
2590 14. Added --file-offsets and --line-offsets to pcregrep.
2591
2592 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
2593 infinitely looping recursion. The bug was that positive lookaheads were not
2594 being skipped when checking for a possible empty match (negative lookaheads
2595 and both kinds of lookbehind were skipped).
2596
2597 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
2598 inclusion of <windows.h> to before rather than after the definition of
2599 INVALID_FILE_ATTRIBUTES (patch from David Byron).
2600
2601 17. Specifying a possessive quantifier with a specific limit for a Unicode
2602 character property caused pcre_compile() to compile bad code, which led at
2603 runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
2604 are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
2605 caused the error; without that there was no problem.
2606
2607 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
2608
2609 19. Added --enable-pcretest-libreadline.
2610
2611 20. In pcrecpp.cc, the variable 'count' was incremented twice in
2612 RE::GlobalReplace(). As a result, the number of replacements returned was
2613 double what it should be. I removed one of the increments, but Craig sent a
2614 later patch that removed the other one (the right fix) and added unit tests
2615 that check the return values (which was not done before).
2616
2617 21. Several CMake things:
2618
2619 (1) Arranged that, when cmake is used on Unix, the libraries end up with
2620 the names libpcre and libpcreposix, not just pcre and pcreposix.
2621
2622 (2) The above change means that pcretest and pcregrep are now correctly
2623 linked with the newly-built libraries, not previously installed ones.
2624
2625 (3) Added PCRE_SUPPORT_LIBREADLINE, PCRE_SUPPORT_LIBZ, PCRE_SUPPORT_LIBBZ2.
2626
2627 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
2628 crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
2629 UTF-8 newline character). The key issue is that the pattern starts .*;
2630 this means that the match must be either at the beginning, or after a
2631 newline. The bug was in the code for advancing after a failed match and
2632 checking that the new position followed a newline. It was not taking
2633 account of UTF-8 characters correctly.
2634
2635 23. PCRE was behaving differently from Perl in the way it recognized POSIX
2636 character classes. PCRE was not treating the sequence [:...:] as a
2637 character class unless the ... were all letters. Perl, however, seems to
2638 allow any characters between [: and :], though of course it rejects as
2639 unknown any "names" that contain non-letters, because all the known class
2640 names consist only of letters. Thus, Perl gives an error for [[:1234:]],
2641 for example, whereas PCRE did not - it did not recognize a POSIX character
2642 class. This seemed a bit dangerous, so the code has been changed to be
2643 closer to Perl. The behaviour is not identical to Perl, because PCRE will
2644 diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
2645 treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
2646 Perl does, and where it didn't before.
2647
2648 24. Rewrite so as to remove the single use of %n from pcregrep because in some
2649 Windows environments %n is disabled by default.
2650
2651
2652 Version 7.4 21-Sep-07
2653 ---------------------
2654
2655 1. Change 7.3/28 was implemented for classes by looking at the bitmap. This
2656 means that a class such as [\s] counted as "explicit reference to CR or
2657 LF". That isn't really right - the whole point of the change was to try to
2658 help when there was an actual mention of one of the two characters. So now
2659 the change happens only if \r or \n (or a literal CR or LF) character is
2660 encountered.
2661
2662 2. The 32-bit options word was also used for 6 internal flags, but the numbers
2663 of both had grown to the point where there were only 3 bits left.
2664 Fortunately, there was spare space in the data structure, and so I have
2665 moved the internal flags into a new 16-bit field to free up more option
2666 bits.
2667
2668 3. The appearance of (?J) at the start of a pattern set the DUPNAMES option,
2669 but did not set the internal JCHANGED flag - either of these is enough to
2670 control the way the "get" function works - but the PCRE_INFO_JCHANGED
2671 facility is supposed to tell if (?J) was ever used, so now (?J) at the
2672 start sets both bits.
2673
2674 4. Added options (at build time, compile time, exec time) to change \R from
2675 matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
2676
2677 5. doc/pcresyntax.html was missing from the distribution.
2678
2679 6. Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
2680 compatibility, even though it is no longer used.
2681
2682 7. Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
2683 strtoull to pcrecpp.cc to select the available functions in WIN32 when the
2684 windows.h file is present (where different names are used). [This was
2685 reversed later after testing - see 16 below.]
2686
2687 8. Changed all #include <config.h> to #include "config.h". There were also
2688 some further <pcre.h> cases that I changed to "pcre.h".
2689
2690 9. When pcregrep was used with the --colour option, it missed the line ending
2691 sequence off the lines that it output.
2692
2693 10. It was pointed out to me that arrays of string pointers cause lots of
2694 relocations when a shared library is dynamically loaded. A technique of
2695 using a single long string with a table of offsets can drastically reduce
2696 these. I have refactored PCRE in four places to do this. The result is
2697 dramatic:
2698
2699 Originally: 290
2700 After changing UCP table: 187
2701 After changing error message table: 43
2702 After changing table of "verbs" 36
2703 After changing table of Posix names 22
2704
2705 Thanks to the folks working on Gregex for glib for this insight.
2706
2707 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
2708 unicode-properties was also set.
2709
2710 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
2711
2712 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
2713 checked only for CRLF.
2714
2715 14. Added casts to pcretest.c to avoid compiler warnings.
2716
2717 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
2718
2719 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
2720 and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
2721 entirely. This removes changes made in 7 above.
2722
2723 17. The CMake files have been updated, and there is now more information about
2724 building with CMake in the NON-UNIX-USE document.
2725
2726
2727 Version 7.3 28-Aug-07
2728 ---------------------
2729
2730 1. In the rejigging of the build system that eventually resulted in 7.1, the
2731 line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
2732 brackets there is not right, since it causes compilers to look for an
2733 installed pcre.h, not the version that is in the source that is being
2734 compiled (which of course may be different). I have changed it back to:
2735
2736 #include "pcre.h"
2737
2738 I have a vague recollection that the change was concerned with compiling in
2739 different directories, but in the new build system, that is taken care of
2740 by the VPATH setting the Makefile.
2741
2742 2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
2743 when the subject happened to end in the byte 0x85 (e.g. if the last
2744 character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
2745 characters but of course it shouldn't be taken as a newline when it is part
2746 of another character. The bug was that, for an unlimited repeat of . in
2747 not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
2748 characters when looking for a newline.
2749
2750 3. A small performance improvement in the DOTALL UTF-8 mode .* case.
2751
2752 4. Debugging: adjusted the names of opcodes for different kinds of parentheses
2753 in debug output.
2754
2755 5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
2756 long printing in the pcrecpp unittest when running under MinGW.
2757
2758 6. ESC_K was left out of the EBCDIC table.
2759
2760 7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
2761 parentheses; I made it 1000, which seemed large enough. Unfortunately, the
2762 limit also applies to "virtual nesting" when a pattern is recursive, and in
2763 this case 1000 isn't so big. I have been able to remove this limit at the
2764 expense of backing off one optimization in certain circumstances. Normally,
2765 when pcre_exec() would call its internal match() function recursively and
2766 immediately return the result unconditionally, it uses a "tail recursion"
2767 feature to save stack. However, when a subpattern that can match an empty
2768 string has an unlimited repetition quantifier, it no longer makes this
2769 optimization. That gives it a stack frame in which to save the data for
2770 checking that an empty string has been matched. Previously this was taken
2771 from the 1000-entry workspace that had been reserved. So now there is no
2772 explicit limit, but more stack is used.
2773
2774 8. Applied Daniel's patches to solve problems with the import/export magic
2775 syntax that is required for Windows, and which was going wrong for the
2776 pcreposix and pcrecpp parts of the library. These were overlooked when this
2777 problem was solved for the main library.
2778
2779 9. There were some crude static tests to avoid integer overflow when computing
2780 the size of patterns that contain repeated groups with explicit upper
2781 limits. As the maximum quantifier is 65535, the maximum group length was
2782 set at 30,000 so that the product of these two numbers did not overflow a
2783 32-bit integer. However, it turns out that people want to use groups that
2784 are longer than 30,000 bytes (though not repeat them that many times).
2785 Change 7.0/17 (the refactoring of the way the pattern size is computed) has
2786 made it possible to implement the integer overflow checks in a much more
2787 dynamic way, which I have now done. The artificial limitation on group
2788 length has been removed - we now have only the limit on the total length of
2789 the compiled pattern, which depends on the LINK_SIZE setting.
2790
2791 10. Fixed a bug in the documentation for get/copy named substring when
2792 duplicate names are permitted. If none of the named substrings are set, the
2793 functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
2794 empty string.
2795
2796 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
2797 instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
2798 because the ] is interpreted as the first data character and the
2799 terminating ] is not found. PCRE has been made compatible with Perl in this
2800 regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
2801 cause memory overwriting.
2802
2803 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
2804 string has been matched (to stop an infinite loop). It was not recognizing
2805 a conditional subpattern that could match an empty string if that
2806 subpattern was within another subpattern. For example, it looped when
2807 trying to match (((?(1)X|))*) but it was OK with ((?(1)X|)*) where the
2808 condition was not nested. This bug has been fixed.
2809
2810 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
2811 past the start of the subject in the presence of bytes with the top bit
2812 set, for example "\x8aBCD".
2813
2814 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
2815 (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
2816
2817 14. Optimized (?!) to (*FAIL).
2818
2819 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
2820 This restricts code points to be within the range 0 to 0x10FFFF, excluding
2821 the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
2822 full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
2823 does: it's just the validity check that is more restrictive.
2824
2825 16. Inserted checks for integer overflows during escape sequence (backslash)
2826 processing, and also fixed erroneous offset values for syntax errors during
2827 backslash processing.
2828
2829 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
2830 for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
2831
2832 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
2833 caused an overrun.
2834
2835 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
2836 something other than just ASCII characters) inside a group that had an
2837 unlimited repeat caused a loop at compile time (while checking to see
2838 whether the group could match an empty string).
2839
2840 20. Debugging a pattern containing \p or \P could cause a crash. For example,
2841 [\P{Any}] did so. (Error in the code for printing property names.)
2842
2843 21. An orphan \E inside a character class could cause a crash.
2844
2845 22. A repeated capturing bracket such as (A)? could cause a wild memory
2846 reference during compilation.
2847
2848 23. There are several functions in pcre_compile() that scan along a compiled
2849 expression for various reasons (e.g. to see if it's fixed length for look
2850 behind). There were bugs in these functions when a repeated \p or \P was
2851 present in the pattern. These operators have additional parameters compared
2852 with \d, etc, and these were not being taken into account when moving along
2853 the compiled data. Specifically:
2854
2855 (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
2856 length.
2857
2858 (b) An item such as \pL+ within a repeated group could cause crashes or
2859 loops.
2860
2861 (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
2862 "reference to non-existent subpattern" error.
2863
2864 (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
2865
2866 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
2867 characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
2868
2869 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
2870
2871 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
2872 character were causing crashes (broken optimization).
2873
2874 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
2875 \p or \P) caused a compile-time loop.
2876
2877 28. More problems have arisen in unanchored patterns when CRLF is a valid line
2878 break. For example, the unstudied pattern [\r\n]A does not match the string
2879 "\r\nA" because change 7.0/46 below moves the current point on by two
2880 characters after failing to match at the start. However, the pattern \nA
2881 *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
2882 the same is true. There doesn't seem any very clean way out of this, but
2883 what I have chosen to do makes the common cases work: PCRE now takes note
2884 of whether there can be an explicit match for \r or \n anywhere in the
2885 pattern, and if so, 7.0/46 no longer applies. As part of this change,
2886 there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
2887 pattern has explicit CR or LF references.
2888
2889 29. Added (*CR) etc for changing newline setting at start of pattern.
2890
2891
2892 Version 7.2 19-Jun-07
2893 ---------------------
2894
2895 1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
2896 which is apparently normally available under Windows.
2897
2898 2. Re-jig the pcregrep tests with different newline settings in an attempt
2899 to make them independent of the local environment's newline setting.
2900
2901 3. Add code to configure.ac to remove -g from the CFLAGS default settings.
2902
2903 4. Some of the "internals" tests were previously cut out when the link size
2904 was not 2, because the output contained actual offsets. The recent new
2905 "Z" feature of pcretest means that these can be cut out, making the tests
2906 usable with all link sizes.
2907
2908 5. Implemented Stan Switzer's goto replacement for longjmp() when not using
2909 stack recursion. This gives a massive performance boost under BSD, but just
2910 a small improvement under Linux. However, it saves one field in the frame
2911 in all cases.
2912
2913 6. Added more features from the forthcoming Perl 5.10:
2914
2915 (a) (?-n) (where n is a string of digits) is a relative subroutine or
2916 recursion call. It refers to the nth most recently opened parentheses.
2917
2918 (b) (?+n) is also a relative subroutine call; it refers to the nth next
2919 to be opened parentheses.
2920
2921 (c) Conditions that refer to capturing parentheses can be specified
2922 relatively, for example, (?(-2)... or (?(+3)...
2923
2924 (d) \K resets the start of the current match so that everything before
2925 is not part of it.
2926
2927 (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
2928
2929 (f) \g{name} is another synonym - part of Perl 5.10's unification of
2930 reference syntax.
2931
2932 (g) (?| introduces a group in which the numbering of parentheses in each
2933 alternative starts with the same number.
2934
2935 (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
2936
2937 7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
2938 PCRE_INFO_JCHANGED.
2939
2940 8. A pattern such as (.*(.)?)* caused pcre_exec() to fail by either not
2941 terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
2942 for detecting groups that can match an empty string.
2943
2944 9. A pattern with a very large number of alternatives (more than several
2945 hundred) was running out of internal workspace during the pre-compile
2946 phase, where pcre_compile() figures out how much memory will be needed. A
2947 bit of new cunning has reduced the workspace needed for groups with
2948 alternatives. The 1000-alternative test pattern now uses 12 bytes of
2949 workspace instead of running out of the 4096 that are available.
2950
2951 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
2952
2953 11. Applied patch from Google to remove an optimization that didn't quite work.
2954 The report of the bug said:
2955
2956 pcrecpp::RE("a*").FullMatch("aaa") matches, while
2957 pcrecpp::RE("a*?").FullMatch("aaa") does not, and
2958 pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
2959
2960 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
2961 it matched the wrong number of bytes.
2962
2963
2964 Version 7.1 24-Apr-07
2965 ---------------------
2966
2967 1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
2968 that is more "standard", making use of automake and other Autotools. There
2969 is some re-arrangement of the files and adjustment of comments consequent
2970 on this.
2971
2972 2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
2973 for recursive directory scanning broke on some systems because the files
2974 are not scanned in any specific order and on different systems the order
2975 was different. A call to "sort" has been inserted into RunGrepTest for the
2976 approprate test as a short-term fix. In the longer term there may be an
2977 alternative.
2978
2979 3. I had an email from Eric Raymond about problems translating some of PCRE's
2980 man pages to HTML (despite the fact that I distribute HTML pages, some
2981 people do their own conversions for various reasons). The problems
2982 concerned the use of low-level troff macros .br and .in. I have therefore
2983 removed all such uses from the man pages (some were redundant, some could
2984 be replaced by .nf/.fi pairs). The 132html script that I use to generate
2985 HTML has been updated to handle .nf/.fi and to complain if it encounters
2986 .br or .in.
2987
2988 4. Updated comments in configure.ac that get placed in config.h.in and also
2989 arranged for config.h to be included in the distribution, with the name
2990 config.h.generic, for the benefit of those who have to compile without
2991 Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
2992
2993 5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
2994 Weber: (1) pcre_internal.h was missing some function renames; (2) updated
2995 makevp.bat for the current PCRE, using the additional files
2996 makevp_c.txt, makevp_l.txt, and pcregexp.pas.
2997
2998 6. A Windows user reported a minor discrepancy with test 2, which turned out
2999 to be caused by a trailing space on an input line that had got lost in his
3000 copy. The trailing space was an accident, so I've just removed it.
3001
3002 7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
3003 that is needed.
3004
3005 8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
3006 as "const" (a) because they are and (b) because it helps the PHP
3007 maintainers who have recently made a script to detect big data structures
3008 in the php code that should be moved to the .rodata section. I remembered
3009 to update Builducptable as well, so it won't revert if ucptable.h is ever
3010 re-created.
3011
3012 9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
3013 pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
3014 order to be able to cut out the UTF-8 tables in the latter when UTF-8
3015 support is not required. This saves 1.5-2K of code, which is important in
3016 some applications.
3017
3018 Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
3019 so as not to refer to the tables, even though these functions will never be
3020 called when UTF-8 support is disabled. Otherwise there are problems with a
3021 shared library.
3022
3023 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
3024
3025 (a) It was defining its arguments as char * instead of void *.
3026
3027 (b) It was assuming that all moves were upwards in memory; this was true
3028 a long time ago when I wrote it, but is no longer the case.
3029
3030 The emulated memove() is provided for those environments that have neither
3031 memmove() nor bcopy(). I didn't think anyone used it these days, but that
3032 is clearly not the case, as these two bugs were recently reported.
3033
3034 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
3035 and Detrail to create the HTML documentation, the .txt form of the man
3036 pages, and it removes trailing spaces from listed files. It also creates
3037 pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
3038 case, it wraps all the #defines with #ifndefs. This script should be run
3039 before "make dist".
3040
3041 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
3042 with Unicode property support.
3043
3044 (a) For a maximizing quantifier, if the two different cases of the
3045 character were of different lengths in their UTF-8 codings (there are
3046 some cases like this - I found 11), and the matching function had to
3047 back up over a mixture of the two cases, it incorrectly assumed they
3048 were both the same length.
3049
3050 (b) When PCRE was configured to use the heap rather than the stack for
3051 recursion during matching, it was not correctly preserving the data for
3052 the other case of a UTF-8 character when checking ahead for a match
3053 while processing a minimizing repeat. If the check also involved
3054 matching a wide character, but failed, corruption could cause an
3055 erroneous result when trying to check for a repeat of the original
3056 character.
3057
3058 13. Some tidying changes to the testing mechanism:
3059
3060 (a) The RunTest script now detects the internal link size and whether there
3061 is UTF-8 and UCP support by running ./pcretest -C instead of relying on
3062 values substituted by "configure". (The RunGrepTest script already did
3063 this for UTF-8.) The configure.ac script no longer substitutes the
3064 relevant variables.
3065
3066 (b) The debugging options /B and /D in pcretest show the compiled bytecode
3067 with length and offset values. This means that the output is different
3068 for different internal link sizes. Test 2 is skipped for link sizes
3069 other than 2 because of this, bypassing the problem. Unfortunately,
3070 there was also a test in test 3 (the locale tests) that used /B and
3071 failed for link sizes other than 2. Rather than cut the whole test out,
3072 I have added a new /Z option to pcretest that replaces the length and
3073 offset values with spaces. This is now used to make test 3 independent
3074 of link size. (Test 2 will be tidied up later.)
3075
3076 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
3077 segmentation fault instead of returning the appropriate error message.
3078
3079 15. In multiline mode when the newline sequence was set to "any", the pattern
3080 ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
3081 This doesn't seem right; it now treats the CRLF combination as the line
3082 ending, and so does not match in that case. It's only a pattern such as ^$
3083 that would hit this one: something like ^ABC$ would have failed after \r
3084 and then tried again after \r\n.
3085
3086 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
3087 in an attempt to make files that differ only in their line terminators
3088 compare equal. This works on Linux.
3089
3090 17. Under certain error circumstances pcregrep might try to free random memory
3091 as it exited. This is now fixed, thanks to valgrind.
3092
3093 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
3094 "abc\r\n\r\n", it found an unwanted second match after the second \r. This
3095 was because its rules for how to advance for /g after matching an empty
3096 string at the end of a line did not allow for this case. They now check for
3097 it specially.
3098
3099 20. pcretest is supposed to handle patterns and data of any length, by
3100 extending its buffers when necessary. It was getting this wrong when the
3101 buffer for a data line had to be extended.
3102
3103 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
3104 CRLF as a newline sequence.
3105
3106 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
3107 out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
3108 I have nevertheless tidied it up.
3109
3110 23. Added some casts to kill warnings from HP-UX ia64 compiler.
3111
3112 24. Added a man page for pcre-config.
3113
3114
3115 Version 7.0 19-Dec-06
3116 ---------------------
3117
3118 1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
3119 moving to gcc 4.1.1.
3120
3121 2. The -S option for pcretest uses setrlimit(); I had omitted to #include
3122 sys/time.h, which is documented as needed for this function. It doesn't
3123 seem to matter on Linux, but it showed up on some releases of OS X.
3124
3125 3. It seems that there are systems where bytes whose values are greater than
3126 127 match isprint() in the "C" locale. The "C" locale should be the
3127 default when a C program starts up. In most systems, only ASCII printing
3128 characters match isprint(). This difference caused the output from pcretest
3129 to vary, making some of the tests fail. I have changed pcretest so that:
3130
3131 (a) When it is outputting text in the compiled version of a pattern, bytes
3132 other than 32-126 are always shown as hex escapes.
3133
3134 (b) When it is outputting text that is a matched part of a subject string,
3135 it does the same, unless a different locale has been set for the match
3136 (using the /L modifier). In this case, it uses isprint() to decide.
3137
3138 4. Fixed a major bug that caused incorrect computation of the amount of memory
3139 required for a compiled pattern when options that changed within the
3140 pattern affected the logic of the preliminary scan that determines the
3141 length. The relevant options are -x, and -i in UTF-8 mode. The result was
3142 that the computed length was too small. The symptoms of this bug were
3143 either the PCRE error "internal error: code overflow" from pcre_compile(),
3144 or a glibc crash with a message such as "pcretest: free(): invalid next
3145 size (fast)". Examples of patterns that provoked this bug (shown in
3146 pcretest format) are:
3147
3148 /(?-x: )/x
3149 /(?x)(?-x: \s*#\s*)/
3150 /((?i)[\x{c0}])/8
3151 /(?i:[\x{c0}])/8
3152
3153 HOWEVER: Change 17 below makes this fix obsolete as the memory computation
3154 is now done differently.
3155
3156 5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
3157 wrapper classes; (b) implement a new function in the C++ scanner that is
3158 more efficient than the old way of doing things because it avoids levels of
3159 recursion in the regex matching; (c) add a paragraph to the documentation
3160 for the FullMatch() function.
3161
3162 6. The escape sequence \n was being treated as whatever was defined as
3163 "newline". Not only was this contrary to the documentation, which states
3164 that \n is character 10 (hex 0A), but it also went horribly wrong when
3165 "newline" was defined as CRLF. This has been fixed.
3166
3167 7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
3168 was being set to -1 for the "end of line" case (supposedly a value that no
3169 character can have). Though this value is never used (the check for end of
3170 line is "zero bytes in current character"), it caused compiler complaints.
3171 I've changed it to 0xffffffff.
3172
3173 8. In pcre_version.c, the version string was being built by a sequence of
3174 C macros that, in the event of PCRE_PRERELEASE being defined as an empty
3175 string (as it is for production releases) called a macro with an empty
3176 argument. The C standard says the result of this is undefined. The gcc
3177 compiler treats it as an empty string (which was what was wanted) but it is
3178 reported that Visual C gives an error. The source has been hacked around to
3179 avoid this problem.
3180
3181 9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
3182 builds of pcretest, and changed the call to _setmode() to use _O_BINARY
3183 instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
3184 of them did).
3185
3186 10. Originally, pcretest opened its input and output without "b"; then I was
3187 told that "b" was needed in some environments, so it was added for release
3188 5.0 to both the input and output. (It makes no difference on Unix-like
3189 systems.) Later I was told that it is wrong for the input on Windows. I've
3190 now abstracted the modes into two macros, to make it easier to fiddle with
3191 them, and removed "b" from the input mode under Windows.
3192
3193 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
3194
3195 12. Added -help and --help to pcretest as an official way of being reminded
3196 of the options.
3197
3198 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
3199 and pcrecpp.cc because they annoy compilers at high warning levels.
3200
3201 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
3202
3203 15. Fixed an occurrence of == in configure.ac that should have been = (shell
3204 scripts are not C programs :-) and which was not noticed because it works
3205 on Linux.
3206
3207 16. pcretest is supposed to handle any length of pattern and data line (as one
3208 line or as a continued sequence of lines) by extending its input buffer if
3209 necessary. This feature was broken for very long pattern lines, leading to
3210 a string of junk being passed to pcre_compile() if the pattern was longer
3211 than about 50K.
3212
3213 17. I have done a major re-factoring of the way pcre_compile() computes the
3214 amount of memory needed for a compiled pattern. Previously, there was code
3215 that made a preliminary scan of the pattern in order to do this. That was
3216 OK when PCRE was new, but as the facilities have expanded, it has become
3217 harder and harder to keep it in step with the real compile phase, and there
3218 have been a number of bugs (see for example, 4 above). I have now found a
3219 cunning way of running the real compile function in a "fake" mode that
3220 enables it to compute how much memory it would need, while actually only
3221 ever using a few hundred bytes of working memory and without too many
3222 tests of the mode. This should make future maintenance and development
3223 easier. A side effect of this work is that the limit of 200 on the nesting
3224 depth of parentheses has been removed (though this was never a serious
3225 limitation, I suspect). However, there is a downside: pcre_compile() now
3226 runs more slowly than before (30% or more, depending on the pattern). I
3227 hope this isn't a big issue. There is no effect on runtime performance.
3228
3229 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
3230 newline (only possible for the last line of a file) and it was a
3231 pattern that set a locale (followed by /Lsomething), pcretest crashed.
3232
3233 19. Added additional timing features to pcretest. (1) The -tm option now times
3234 matching only, not compiling. (2) Both -t and -tm can be followed, as a
3235 separate command line item, by a number that specifies the number of
3236 repeats to use when timing. The default is 50000; this gives better
3237 precision, but takes uncomfortably long for very large patterns.
3238
3239 20. Extended pcre_study() to be more clever in cases where a branch of a
3240 subpattern has no definite first character. For example, (a*|b*)[cd] would
3241 previously give no result from pcre_study(). Now it recognizes that the
3242 first character must be a, b, c, or d.
3243
3244 21. There was an incorrect error "recursive call could loop indefinitely" if
3245 a subpattern (or the entire pattern) that was being tested for matching an
3246 empty string contained only one non-empty item after a nested subpattern.
3247 For example, the pattern (?>\x{100}*)\d(?R) provoked this error
3248 incorrectly, because the \d was being skipped in the check.
3249
3250 22. The pcretest program now has a new pattern option /B and a command line
3251 option -b, which is equivalent to adding /B to every pattern. This causes
3252 it to show the compiled bytecode, without the additional information that
3253 -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
3254 is the same as /B/I).
3255
3256 23. A new optimization is now able automatically to treat some sequences such
3257 as a*b as a*+b. More specifically, if something simple (such as a character
3258 or a simple class like \d) has an unlimited quantifier, and is followed by
3259 something that cannot possibly match the quantified thing, the quantifier
3260 is automatically "possessified".
3261
3262 24. A recursive reference to a subpattern whose number was greater than 39
3263 went wrong under certain circumstances in UTF-8 mode. This bug could also
3264 have affected the operation of pcre_study().
3265
3266 25. Realized that a little bit of performance could be had by replacing
3267 (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
3268
3269 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
3270
3271 27. Possessive quantifiers such as a++ were previously implemented by turning
3272 them into atomic groups such as ($>a+). Now they have their own opcodes,
3273 which improves performance. This includes the automatically created ones
3274 from 23 above.
3275
3276 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
3277 lookahead was broken if it was not anchored. PCRE was mistakenly expecting
3278 the first matched character to be a colon. This applied both to named and
3279 numbered groups.
3280
3281 29. The ucpinternal.h header file was missing its idempotency #ifdef.
3282
3283 30. I was sent a "project" file called libpcre.a.dev which I understand makes
3284 building PCRE on Windows easier, so I have included it in the distribution.
3285
3286 31. There is now a check in pcretest against a ridiculously large number being
3287 returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
3288 loop, the loop is abandoned.
3289
3290 32. Forward references to subpatterns in conditions such as (?(2)...) where
3291 subpattern 2 is defined later cause pcre_compile() to search forwards in
3292 the pattern for the relevant set of parentheses. This search went wrong
3293 when there were unescaped parentheses in a character class, parentheses
3294 escaped with \Q...\E, or parentheses in a #-comment in /x mode.
3295
3296 33. "Subroutine" calls and backreferences were previously restricted to
3297 referencing subpatterns earlier in the regex. This restriction has now
3298 been removed.
3299
3300 34. Added a number of extra features that are going to be in Perl 5.10. On the
3301 whole, these are just syntactic alternatives for features that PCRE had
3302 previously implemented using the Python syntax or my own invention. The
3303 other formats are all retained for compatibility.
3304
3305 (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
3306 as (?P<name>...). The new forms, as well as being in Perl 5.10, are
3307 also .NET compatible.
3308
3309 (b) A recursion or subroutine call to a named group can now be defined as
3310 (?&name) as well as (?P>name).
3311
3312 (c) A backreference to a named group can now be defined as \k<name> or
3313 \k'name' as well as (?P=name). The new forms, as well as being in Perl
3314 5.10, are also .NET compatible.
3315
3316 (d) A conditional reference to a named group can now use the syntax
3317 (?(<name>) or (?('name') as well as (?(name).
3318
3319 (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
3320 groups (named and numbered) that are never evaluated inline, but can be
3321 called as "subroutines" from elsewhere. In effect, the DEFINE condition
3322 is always false. There may be only one alternative in such a group.
3323
3324 (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
3325 as the simple (?(R). The condition is true only if the most recent
3326 recursion is that of the given number or name. It does not search out
3327 through the entire recursion stack.
3328
3329 (g) The escape \gN or \g{N} has been added, where N is a positive or
3330 negative number, specifying an absolute or relative reference.
3331
3332 35. Tidied to get rid of some further signed/unsigned compiler warnings and
3333 some "unreachable code" warnings.
3334
3335 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
3336 things, this adds five new scripts.
3337
3338 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
3339 There were also incompatibilities regarding the handling of \Q..\E inside
3340 character classes, for example with patterns like [\Qa\E-\Qz\E] where the
3341 hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
3342
3343 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
3344 matches an empty string, and forcibly breaks the loop. There were bugs in
3345 this code in non-simple cases. For a pattern such as ^(a()*)* matched
3346 against aaaa the result was just "a" rather than "aaaa", for example. Two
3347 separate and independent bugs (that affected different cases) have been
3348 fixed.
3349
3350 39. Refactored the code to abolish the use of different opcodes for small
3351 capturing bracket numbers. This is a tidy that I avoided doing when I
3352 removed the limit on the number of capturing brackets for 3.5 back in 2001.
3353 The new approach is not only tidier, it makes it possible to reduce the
3354 memory needed to fix the previous bug (38).
3355
3356 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
3357 sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
3358 processing dot, circumflex, or dollar metacharacters, or #-comments in /x
3359 mode.
3360
3361 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
3362 report.
3363
3364 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
3365 copy construction and assignment in the C++ wrapper.
3366
3367 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
3368 couple of bugs that could have given wrong results in the "--newline=crlf"
3369 case.
3370
3371 44. Added a number of casts and did some reorganization of signed/unsigned int
3372 variables following suggestions from Dair Grant. Also renamed the variable
3373 "this" as "item" because it is a C++ keyword.
3374
3375 45. Arranged for dftables to add
3376
3377 #include "pcre_internal.h"
3378
3379 to pcre_chartables.c because without it, gcc 4.x may remove the array
3380 definition from the final binary if PCRE is built into a static library and
3381 dead code stripping is activated.
3382
3383 46. For an unanchored pattern, if a match attempt fails at the start of a
3384 newline sequence, and the newline setting is CRLF or ANY, and the next two
3385 characters are CRLF, advance by two characters instead of one.
3386
3387
3388 Version 6.7 04-Jul-06
3389 ---------------------
3390
3391 1. In order to handle tests when input lines are enormously long, pcretest has
3392 been re-factored so that it automatically extends its buffers when
3393 necessary. The code is crude, but this _is_ just a test program. The
3394 default size has been increased from 32K to 50K.
3395
3396 2. The code in pcre_study() was using the value of the re argument before
3397 testing it for NULL. (Of course, in any sensible call of the function, it
3398 won't be NULL.)
3399
3400 3. The memmove() emulation function in pcre_internal.h, which is used on
3401 systems that lack both memmove() and bcopy() - that is, hardly ever -
3402 was missing a "static" storage class specifier.
3403
3404 4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
3405 containing an extended class (one that cannot be represented by a bitmap
3406 because it contains high-valued characters or Unicode property items, e.g.
3407 [\pZ]). Almost always one would set UTF-8 mode when processing such a
3408 pattern, but PCRE should not loop if you do not (it no longer does).
3409 [Detail: two cases were found: (a) a repeated subpattern containing an
3410 extended class; (b) a recursive reference to a subpattern that followed a
3411 previous extended class. It wasn't skipping over the extended class
3412 correctly when UTF-8 mode was not set.]
3413
3414 5. A negated single-character class was not being recognized as fixed-length
3415 in lookbehind assertions such as (?<=[^f]), leading to an incorrect
3416 compile error "lookbehind assertion is not fixed length".
3417
3418 6. The RunPerlTest auxiliary script was showing an unexpected difference
3419 between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
3420 write a Perl script that can interpret lines of an input file either as
3421 byte characters or as UTF-8, which is what "perltest" was being required to
3422 do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
3423 can't do is switch easily at run time between having the "use utf8;" pragma
3424 or not. In the end, I fudged it by using the RunPerlTest script to insert
3425 "use utf8;" explicitly for the UTF-8 tests.
3426
3427 7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
3428 the end of the subject string, contrary to the documentation and to what
3429 Perl does. This was true of both matching functions. Now it matches only at
3430 the start of the subject and immediately after *internal* newlines.
3431
3432 8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
3433 a pointer to an int instead of a pointer to an unsigned long int. This
3434 caused problems on 64-bit systems.
3435
3436 9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
3437 instance of the 'standard' template library not being so standard".
3438
3439 10. There was no check on the number of named subpatterns nor the maximum
3440 length of a subpattern name. The product of these values is used to compute
3441 the size of the memory block for a compiled pattern. By supplying a very
3442 long subpattern name and a large number of named subpatterns, the size
3443 computation could be caused to overflow. This is now prevented by limiting
3444 the length of names to 32 characters, and the number of named subpatterns
3445 to 10,000.
3446
3447 11. Subpatterns that are repeated with specific counts have to be replicated in
3448 the compiled pattern. The size of memory for this was computed from the
3449 length of the subpattern and the repeat count. The latter is limited to
3450 65535, but there was no limit on the former, meaning that integer overflow
3451 could in principle occur. The compiled length of a repeated subpattern is
3452 now limited to 30,000 bytes in order to prevent this.
3453
3454 12. Added the optional facility to have named substrings with the same name.
3455
3456 13. Added the ability to use a named substring as a condition, using the
3457 Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
3458 are numbers (not recommended). Forward references are permitted.
3459
3460 14. Added forward references in named backreferences (if you see what I mean).
3461
3462 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
3463 pattern could run off the end of the subject. For example, the pattern
3464 "(?s)(.{1,5})"8 did this with the subject "ab".
3465
3466 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
3467 PCRE_CASELESS was set when matching characters that were quantified with ?
3468 or *.
3469
3470 17. A character class other than a single negated character that had a minimum
3471 but no maximum quantifier - for example [ab]{6,} - was not handled
3472 correctly by pce_dfa_exec(). It would match only one character.
3473
3474 18. A valid (though odd) pattern that looked like a POSIX character
3475 class but used an invalid character after [ (for example [[,abc,]]) caused
3476 pcre_compile() to give the error "Failed: internal error: code overflow" or
3477 in some cases to crash with a glibc free() error. This could even happen if
3478 the pattern terminated after [[ but there just happened to be a sequence of
3479 letters, a binary zero, and a closing ] in the memory that followed.
3480
3481 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
3482 over the years. Originally (before any Unicode support), just the bottom 8
3483 bits were taken. Thus, for example, \500 really meant \100. Nowadays the
3484 output from "man perlunicode" includes this:
3485
3486 The regular expression compiler produces polymorphic opcodes. That
3487 is, the pattern adapts to the data and automatically switches to
3488 the Unicode character scheme when presented with Unicode data--or
3489 instead uses a traditional byte scheme when presented with byte
3490 data.
3491
3492 Sadly, a wide octal escape does not cause a switch, and in a string with
3493 no other multibyte characters, these octal escapes are treated as before.
3494 Thus, in Perl, the pattern /\500/ actually matches \100 but the pattern
3495 /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
3496 Unicode string.
3497
3498 I have not perpetrated such confusion in PCRE. Up till now, it took just
3499 the bottom 8 bits, as in old Perl. I have now made octal escapes with
3500 values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
3501 translate to the appropriate multibyte character.
3502
3503 29. Applied some refactoring to reduce the number of warnings from Microsoft
3504 and Borland compilers. This has included removing the fudge introduced
3505 seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
3506 a warning about an unused variable.
3507
3508 21. PCRE has not included VT (character 0x0b) in the set of whitespace
3509 characters since release 4.0, because Perl (from release 5.004) does not.
3510 [Or at least, is documented not to: some releases seem to be in conflict
3511 with the documentation.] However, when a pattern was studied with
3512 pcre_study() and all its branches started with \s, PCRE still included VT
3513 as a possible starting character. Of course, this did no harm; it just
3514 caused an unnecessary match attempt.
3515
3516 22. Removed a now-redundant internal flag bit that recorded the fact that case
3517 dependency changed within the pattern. This was once needed for "required
3518 byte" processing, but is no longer used. This recovers a now-scarce options
3519 bit. Also moved the least significant internal flag bit to the most-
3520 significant bit of the word, which was not previously used (hangover from
3521 the days when it was an int rather than a uint) to free up another bit for
3522 the future.
3523
3524 23. Added support for CRLF line endings as well as CR and LF. As well as the
3525 default being selectable at build time, it can now be changed at runtime
3526 via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
3527 specify that it is scanning data with non-default line endings.
3528
3529 24. Changed the definition of CXXLINK to make it agree with the definition of
3530 LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
3531
3532 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
3533 recursions. This makes a big different to stack usage for some patterns.
3534
3535 26. If a subpattern containing a named recursion or subroutine reference such
3536 as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
3537 the space required for the compiled pattern went wrong and gave too small a
3538 value. Depending on the environment, this could lead to "Failed: internal
3539 error: code overflow at offset 49" or "glibc detected double free or
3540 corruption" errors.
3541
3542 27. Applied patches from Google (a) to support the new newline modes and (b) to
3543 advance over multibyte UTF-8 characters in GlobalReplace.
3544
3545 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
3546 difference for some implementation of PCRE in some Windows version.
3547
3548 29. Added some extra testing facilities to pcretest:
3549
3550 \q<number> in a data line sets the "match limit" value
3551 \Q<number> in a data line sets the "match recursion limt" value
3552 -S <number> sets the stack size, where <number> is in megabytes
3553
3554 The -S option isn't available for Windows.
3555
3556
3557 Version 6.6 06-Feb-06
3558 ---------------------
3559
3560 1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
3561 in pcreposix.h. I have copied the definition from pcre.h.
3562
3563 2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
3564 because pcre.h is no longer a built file.
3565
3566 3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
3567 not normally included in the compiled code.
3568
3569
3570 Version 6.5 01-Feb-06
3571 ---------------------
3572
3573 1. When using the partial match feature with pcre_dfa_exec(), it was not
3574 anchoring the second and subsequent partial matches at the new starting
3575 point. This could lead to incorrect results. For example, with the pattern
3576 /1234/, partially matching against "123" and then "a4" gave a match.
3577
3578 2. Changes to pcregrep:
3579
3580 (a) All non-match returns from pcre_exec() were being treated as failures
3581 to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
3582 error message is output. Some extra information is given for the
3583 PCRE_ERROR_MATCHLIMIT and PCRE_ERROR_RECURSIONLIMIT errors, which are
3584 probably the only errors that are likely to be caused by users (by
3585 specifying a regex that has nested indefinite repeats, for instance).
3586 If there are more than 20 of these errors, pcregrep is abandoned.
3587
3588 (b) A binary zero was treated as data while matching, but terminated the
3589 output line if it was written out. This has been fixed: binary zeroes
3590 are now no different to any other data bytes.
3591
3592 (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
3593 used to set a locale for matching. The --locale=xxxx long option has
3594 been added (no short equivalent) to specify a locale explicitly on the
3595 pcregrep command, overriding the environment variables.
3596
3597 (d) When -B was used with -n, some line numbers in the output were one less
3598 than they should have been.
3599
3600 (e) Added the -o (--only-matching) option.
3601
3602 (f) If -A or -C was used with -c (count only), some lines of context were
3603 accidentally printed for the final match.
3604
3605 (g) Added the -H (--with-filename) option.
3606
3607 (h) The combination of options -rh failed to suppress file names for files
3608 that were found from directory arguments.
3609
3610 (i) Added the -D (--devices) and -d (--directories) options.
3611
3612 (j) Added the -F (--fixed-strings) option.
3613
3614 (k) Allow "-" to be used as a file name for -f as well as for a data file.
3615
3616 (l) Added the --colo(u)r option.
3617
3618 (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
3619 is not present by default.
3620
3621 3. A nasty bug was discovered in the handling of recursive patterns, that is,
3622 items such as (?R) or (?1), when the recursion could match a number of
3623 alternatives. If it matched one of the alternatives, but subsequently,
3624 outside the recursion, there was a failure, the code tried to back up into
3625 the recursion. However, because of the way PCRE is implemented, this is not
3626 possible, and the result was an incorrect result from the match.
3627
3628 In order to prevent this happening, the specification of recursion has
3629 been changed so that all such subpatterns are automatically treated as
3630 atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
3631
3632 4. I had overlooked the fact that, in some locales, there are characters for
3633 which isalpha() is true but neither isupper() nor islower() are true. In
3634 the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
3635 and ordfeminine) are like this. This affected the treatment of \w and \W
3636 when they appeared in character classes, but not when they appeared outside
3637 a character class. The bit map for "word" characters is now created
3638 separately from the results of isalnum() instead of just taking it from the
3639 upper, lower, and digit maps. (Plus the underscore character, of course.)
3640
3641 5. The above bug also affected the handling of POSIX character classes such as
3642 [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
3643 permanent tables. Instead, the bit maps for such a class were previously
3644 created as the appropriate unions of the upper, lower, and digit bitmaps.
3645 Now they are created by subtraction from the [[:word:]] class, which has
3646 its own bitmap.
3647
3648 6. The [[:blank:]] character class matches horizontal, but not vertical space.
3649 It is created by subtracting the vertical space characters (\x09, \x0a,
3650 \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
3651 subtraction was done in the overall bitmap for a character class, meaning
3652 that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
3653 be recognized. This bug has been fixed.
3654
3655 7. Patches from the folks at Google:
3656
3657 (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
3658 real life, but is still worth protecting against".
3659
3660 (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
3661 regular expressions".
3662
3663 (c) pcre_scanner.cc: avoid use of std::count() because not all systems
3664 have it.
3665
3666 (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
3667 "configure" and the latter not, in order to fix a problem somebody had
3668 with compiling the Arg class on HP-UX.
3669
3670 (e) Improve the error-handling of the C++ wrapper a little bit.
3671
3672 (f) New tests for checking recursion limiting.
3673
3674 8. The pcre_memmove() function, which is used only if the environment does not
3675 have a standard memmove() function (and is therefore rarely compiled),
3676 contained two bugs: (a) use of int instead of size_t, and (b) it was not
3677 returning a result (though PCRE never actually uses the result).
3678
3679 9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
3680 large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
3681 returned instead of calling malloc() with an overflowing number that would
3682 most likely cause subsequent chaos.
3683
3684 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
3685
3686 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
3687 with this option is matched, the nmatch and pmatch options of regexec() are
3688 ignored.
3689
3690 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
3691 provided in case anyone wants to the the POSIX interface with UTF-8
3692 strings.
3693
3694 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
3695 C++ linking (needed for some HP-UX environments).
3696
3697 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
3698 (unused parameter) and in the pcre_printint() function (omitted "default"
3699 switch label when the default is to do nothing).
3700
3701 15. Added some code to make it possible, when PCRE is compiled as a C++
3702 library, to replace subject pointers for pcre_exec() with a smart pointer
3703 class, thus making it possible to process discontinuous strings.
3704
3705 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
3706 much the same function. They were added by different people who were trying
3707 to make PCRE easy to compile on non-Unix systems. It has been suggested
3708 that PCRE_EXPORT be abolished now that there is more automatic apparatus
3709 for compiling on Windows systems. I have therefore replaced it with
3710 PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
3711 defaults to "extern" for C or "extern C" for C++, which works fine on
3712 Unix-like systems. It is now possible to override the value of PCRE_DATA_
3713 SCOPE with something explicit in config.h. In addition:
3714
3715 (a) pcreposix.h still had just "extern" instead of either of these macros;
3716 I have replaced it with PCRE_DATA_SCOPE.
3717
3718 (b) Functions such as _pcre_xclass(), which are internal to the library,
3719 but external in the C sense, all had PCRE_EXPORT in their definitions.
3720 This is apparently wrong for the Windows case, so I have removed it.
3721 (It makes no difference on Unix-like systems.)
3722
3723 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
3724 of recursive calls to match(). This is different to MATCH_LIMIT because
3725 that limits the total number of calls to match(), not all of which increase
3726 the depth of recursion. Limiting the recursion depth limits the amount of
3727 stack (or heap if NO_RECURSE is set) that is used. The default can be set
3728 when PCRE is compiled, and changed at run time. A patch from Google adds
3729 this functionality to the C++ interface.
3730
3731 18. Changes to the handling of Unicode character properties:
3732
3733 (a) Updated the table to Unicode 4.1.0.
3734
3735 (b) Recognize characters that are not in the table as "Cn" (undefined).
3736
3737 (c) I revised the way the table is implemented to a much improved format
3738 which includes recognition of ranges. It now supports the ranges that
3739 are defined in UnicodeData.txt, and it also amalgamates other
3740 characters into ranges. This has reduced the number of entries in the
3741 table from around 16,000 to around 3,000, thus reducing its size
3742 considerably. I realized I did not need to use a tree structure after
3743 all - a binary chop search is just as efficient. Having reduced the
3744 number of entries, I extended their size from 6 bytes to 8 bytes to
3745 allow for more data.
3746
3747 (d) Added support for Unicode script names via properties such as \p{Han}.
3748
3749 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
3750 matching that character.
3751
3752 20. When matching a repeated Unicode property with a minimum greater than zero,
3753 (for example \pL{2,}), PCRE could look past the end of the subject if it
3754 reached it while seeking the minimum number of characters. This could
3755 happen only if some of the characters were more than one byte long, because
3756 there is a check for at least the minimum number of bytes.
3757
3758 21. Refactored the implementation of \p and \P so as to be more general, to
3759 allow for more different types of property in future. This has changed the
3760 compiled form incompatibly. Anybody with saved compiled patterns that use
3761 \p or \P will have to recompile them.
3762
3763 22. Added "Any" and "L&" to the supported property types.
3764
3765 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
3766 but give a compile time error if the value is greater than 0xff.
3767
3768 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
3769 accidentally not being installed or uninstalled.
3770
3771 25. The pcre.h file was built from pcre.h.in, but the only changes that were
3772 made were to insert the current release number. This seemed silly, because
3773 it made things harder for people building PCRE on systems that don't run
3774 "configure". I have turned pcre.h into a distributed file, no longer built
3775 by "configure", with the version identification directly included. There is
3776 no longer a pcre.h.in file.
3777
3778 However, this change necessitated a change to the pcre-config script as
3779 well. It is built from pcre-config.in, and one of the substitutions was the
3780 release number. I have updated configure.ac so that ./configure now finds
3781 the release number by grepping pcre.h.
3782
3783 26. Added the ability to run the tests under valgrind.
3784
3785
3786 Version 6.4 05-Sep-05
3787 ---------------------
3788
3789 1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
3790 "--" to be printed when multiple files were scanned, even when none of the
3791 -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
3792 consider it to be a bug, and have restored the previous behaviour.
3793
3794 2. A couple of code tidies to get rid of compiler warnings.
3795
3796 3. The pcretest program used to cheat by referring to symbols in the library
3797 whose names begin with _pcre_. These are internal symbols that are not
3798 really supposed to be visible externally, and in some environments it is
3799 possible to suppress them. The cheating is now confined to including
3800 certain files from the library's source, which is a bit cleaner.
3801
3802 4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
3803 file's purpose clearer.
3804
3805 5. Reorganized pcre_ucp_findchar().
3806
3807
3808 Version 6.3 15-Aug-05
3809 ---------------------
3810
3811 1. The file libpcre.pc.in did not have general read permission in the tarball.
3812
3813 2. There were some problems when building without C++ support:
3814
3815 (a) If C++ support was not built, "make install" and "make test" still
3816 tried to test it.
3817
3818 (b) There were problems when the value of CXX was explicitly set. Some
3819 changes have been made to try to fix these, and ...
3820
3821 (c) --disable-cpp can now be used to explicitly disable C++ support.
3822
3823 (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
3824 backslash in a target when C++ was disabled. This confuses some
3825 versions of "make", apparently. Using an intermediate variable solves
3826 this. (Same for CPP_LOBJ.)
3827
3828 3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
3829 (non-Windows) now includes $(CFLAGS) because these flags are sometimes
3830 necessary on certain architectures.
3831
3832 4. Added a setting of -export-symbols-regex to the link command to remove
3833 those symbols that are exported in the C sense, but actually are local
3834 within the library, and not documented. Their names all begin with
3835 "_pcre_". This is not a perfect job, because (a) we have to except some
3836 symbols that pcretest ("illegally") uses, and (b) the facility isn't always
3837 available (and never for static libraries). I have made a note to try to
3838 find a way round (a) in the future.
3839
3840
3841 Version 6.2 01-Aug-05
3842 ---------------------
3843
3844 1. There was no test for integer overflow of quantifier values. A construction
3845 such as {1111111111111111} would give undefined results. What is worse, if
3846 a minimum quantifier for a parenthesized subpattern overflowed and became
3847 negative, the calculation of the memory size went wrong. This could have
3848 led to memory overwriting.
3849
3850 2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
3851
3852 3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
3853 operating environments where this matters.
3854
3855 4. Applied Giuseppe Maxia's patch to add additional features for controlling
3856 PCRE options from within the C++ wrapper.
3857
3858 5. Named capturing subpatterns were not being correctly counted when a pattern
3859 was compiled. This caused two problems: (a) If there were more than 100
3860 such subpatterns, the calculation of the memory needed for the whole
3861 compiled pattern went wrong, leading to an overflow error. (b) Numerical
3862 back references of the form \12, where the number was greater than 9, were
3863 not recognized as back references, even though there were sufficient
3864 previous subpatterns.
3865
3866 6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
3867 versions of gcc, e.g. 2.95.4.
3868
3869
3870 Version 6.1 21-Jun-05
3871 ---------------------
3872
3873 1. There was one reference to the variable "posix" in pcretest.c that was not
3874 surrounded by "#if !defined NOPOSIX".
3875
3876 2. Make it possible to compile pcretest without DFA support, UTF8 support, or
3877 the cross-check on the old pcre_info() function, for the benefit of the
3878 cut-down version of PCRE that is currently imported into Exim.
3879
3880 3. A (silly) pattern starting with (?i)(?-i) caused an internal space
3881 allocation error. I've done the easy fix, which wastes 2 bytes for sensible
3882 patterns that start (?i) but I don't think that matters. The use of (?i) is
3883 just an example; this all applies to the other options as well.
3884
3885 4. Since libtool seems to echo the compile commands it is issuing, the output
3886 from "make" can be reduced a bit by putting "@" in front of each libtool
3887 compile command.
3888
3889 5. Patch from the folks at Google for configure.in to be a bit more thorough
3890 in checking for a suitable C++ installation before trying to compile the
3891 C++ stuff. This should fix a reported problem when a compiler was present,
3892 but no suitable headers.
3893
3894 6. The man pages all had just "PCRE" as their title. I have changed them to
3895 be the relevant file name. I have also arranged that these names are
3896 retained in the file doc/pcre.txt, which is a concatenation in text format
3897 of all the man pages except the little individual ones for each function.
3898
3899 7. The NON-UNIX-USE file had not been updated for the different set of source
3900 files that come with release 6. I also added a few comments about the C++
3901 wrapper.
3902
3903
3904 Version 6.0 07-Jun-05
3905 ---------------------
3906
3907 1. Some minor internal re-organization to help with my DFA experiments.
3908
3909 2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
3910 didn't matter for the library itself when fully configured, but did matter
3911 when compiling without UCP support, or within Exim, where the ucp files are
3912 not imported.
3913
3914 3. Refactoring of the library code to split up the various functions into
3915 different source modules. The addition of the new DFA matching code (see
3916 below) to a single monolithic source would have made it really too
3917 unwieldy, quite apart from causing all the code to be include in a
3918 statically linked application, when only some functions are used. This is
3919 relevant even without the DFA addition now that patterns can be compiled in
3920 one application and matched in another.
3921
3922 The downside of splitting up is that there have to be some external
3923 functions and data tables that are used internally in different modules of
3924 the library but which are not part of the API. These have all had their
3925 names changed to start with "_pcre_" so that they are unlikely to clash
3926 with other external names.
3927
3928 4. Added an alternate matching function, pcre_dfa_exec(), which matches using
3929 a different (DFA) algorithm. Although it is slower than the original
3930 function, it does have some advantages for certain types of matching
3931 problem.
3932
3933 5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
3934 including restarting after a partial match.
3935
3936 6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
3937 defined when compiling for Windows was sent to me. I have put it into the
3938 code, though I have no means of testing or verifying it.
3939
3940 7. Added the pcre_refcount() auxiliary function.
3941
3942 8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
3943 match before or at the first newline in the subject string. In pcretest,
3944 the /f option on a pattern can be used to set this.
3945
3946 9. A repeated \w when used in UTF-8 mode with characters greater than 256
3947 would behave wrongly. This has been present in PCRE since release 4.0.
3948
3949 10. A number of changes to the pcregrep command:
3950
3951 (a) Refactored how -x works; insert ^(...)$ instead of setting
3952 PCRE_ANCHORED and checking the length, in preparation for adding
3953 something similar for -w.
3954
3955 (b) Added the -w (match as a word) option.
3956
3957 (c) Refactored the way lines are read and buffered so as to have more
3958 than one at a time available.
3959
3960 (d) Implemented a pcregrep test script.
3961
3962 (e) Added the -M (multiline match) option. This allows patterns to match
3963 over several lines of the subject. The buffering ensures that at least
3964 8K, or the rest of the document (whichever is the shorter) is available
3965 for matching (and similarly the previous 8K for lookbehind assertions).
3966
3967 (f) Changed the --help output so that it now says
3968
3969 -w, --word-regex(p)
3970
3971 instead of two lines, one with "regex" and the other with "regexp"
3972 because that confused at least one person since the short forms are the
3973 same. (This required a bit of code, as the output is generated
3974 automatically from a table. It wasn't just a text change.)
3975
3976 (g) -- can be used to terminate pcregrep options if the next thing isn't an
3977 option but starts with a hyphen. Could be a pattern or a path name
3978 starting with a hyphen, for instance.
3979
3980 (h) "-" can be given as a file name to represent stdin.
3981
3982 (i) When file names are being printed, "(standard input)" is used for
3983 the standard input, for compatibility with GNU grep. Previously
3984 "<stdin>" was used.
3985
3986 (j) The option --label=xxx can be used to supply a name to be used for
3987 stdin when file names are being printed. There is no short form.
3988
3989 (k) Re-factored the options decoding logic because we are going to add
3990 two more options that take data. Such options can now be given in four
3991 different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
3992
3993 (l) Added the -A, -B, and -C options for requesting that lines of context
3994 around matches be printed.
3995
3996 (m) Added the -L option to print the names of files that do not contain
3997 any matching lines, that is, the complement of -l.
3998
3999 (n) The return code is 2 if any file cannot be opened, but pcregrep does
4000 continue to scan other files.
4001
4002 (o) The -s option was incorrectly implemented. For compatibility with other
4003 greps, it now suppresses the error message for a non-existent or non-
4004 accessible file (but not the return code). There is a new option called
4005 -q that suppresses the output of matching lines, which was what -s was
4006 previously doing.
4007
4008 (p) Added --include and --exclude options to specify files for inclusion
4009 and exclusion when recursing.
4010
4011 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
4012 Hopefully, it now does.
4013
4014 12. Missing cast in pcre_study().
4015
4016 13. Added an "uninstall" target to the makefile.
4017
4018 14. Replaced "extern" in the function prototypes in Makefile.in with
4019 "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
4020 world, but is set differently for Windows.
4021
4022 15. Added a second compiling function called pcre_compile2(). The only
4023 difference is that it has an extra argument, which is a pointer to an
4024 integer error code. When there is a compile-time failure, this is set
4025 non-zero, in addition to the error test pointer being set to point to an
4026 error message. The new argument may be NULL if no error number is required
4027 (but then you may as well call pcre_compile(), which is now just a
4028 wrapper). This facility is provided because some applications need a
4029 numeric error indication, but it has also enabled me to tidy up the way
4030 compile-time errors are handled in the POSIX wrapper.
4031
4032 16. Added VPATH=.libs to the makefile; this should help when building with one
4033 prefix path and installing with another. (Or so I'm told by someone who
4034 knows more about this stuff than I do.)
4035
4036 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
4037 passes PCRE_DOTALL to the pcre_compile() function, making the "." character
4038 match everything, including newlines. This is not POSIX-compatible, but
4039 somebody wanted the feature. From pcretest it can be activated by using
4040 both the P and the s flags.
4041
4042 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
4043
4044 19. libpcre.pc was being incorrectly installed as executable.
4045
4046 20. A couple of places in pcretest check for end-of-line by looking for '\n';
4047 it now also looks for '\r' so that it will work unmodified on Windows.
4048
4049 21. Added Google's contributed C++ wrapper to the distribution.
4050
4051 22. Added some untidy missing memory free() calls in pcretest, to keep
4052 Electric Fence happy when testing.
4053
4054
4055
4056 Version 5.0 13-Sep-04
4057 ---------------------
4058
4059 1. Internal change: literal characters are no longer packed up into items
4060 containing multiple characters in a single byte-string. Each character
4061 is now matched using a separate opcode. However, there may be more than one
4062 byte in the character in UTF-8 mode.
4063
4064 2. The pcre_callout_block structure has two new fields: pattern_position and
4065 next_item_length. These contain the offset in the pattern to the next match
4066 item, and its length, respectively.
4067
4068 3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
4069 insertion of callouts before each pattern item. Added the /C option to
4070 pcretest to make use of this.
4071
4072 4. On the advice of a Windows user, the lines
4073
4074 #if defined(_WIN32) || defined(WIN32)
4075 _setmode( _fileno( stdout ), 0x8000 );
4076 #endif /* defined(_WIN32) || defined(WIN32) */
4077
4078 have been added to the source of pcretest. This apparently does useful
4079 magic in relation to line terminators.
4080
4081 5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
4082 for the benefit of those environments where the "b" makes a difference.
4083
4084 6. The icc compiler has the same options as gcc, but "configure" doesn't seem
4085 to know about it. I have put a hack into configure.in that adds in code
4086 to set GCC=yes if CC=icc. This seems to end up at a point in the
4087 generated configure script that is early enough to affect the setting of
4088 compiler options, which is what is needed, but I have no means of testing
4089 whether it really works. (The user who reported this had patched the
4090 generated configure script, which of course I cannot do.)
4091
4092 LATER: After change 22 below (new libtool files), the configure script
4093 seems to know about icc (and also ecc). Therefore, I have commented out
4094 this hack in configure.in.
4095
4096 7. Added support for pkg-config (2 patches were sent in).
4097
4098 8. Negated POSIX character classes that used a combination of internal tables
4099 were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
4100 [[:^ascii]]. Typically, they would match almost any characters. The other
4101 POSIX classes were not broken in this way.
4102
4103 9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
4104 to find the match, as PCRE was deluded into thinking that the match had to
4105 start at the start point or following a newline. The same bug applied to
4106 patterns with negative forward assertions or any backward assertions
4107 preceding ".*" at the start, unless the pattern required a fixed first
4108 character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
4109
4110 10. In UTF-8 mode, when moving forwards in the subject after a failed match
4111 starting at the last subject character, bytes beyond the end of the subject
4112 string were read.
4113
4114 11. Renamed the variable "class" as "classbits" to make life easier for C++
4115 users. (Previously there was a macro definition, but it apparently wasn't
4116 enough.)
4117
4118 12. Added the new field "tables" to the extra data so that tables can be passed
4119 in at exec time, or the internal tables can be re-selected. This allows
4120 a compiled regex to be saved and re-used at a later time by a different
4121 program that might have everything at different addresses.
4122
4123 13. Modified the pcre-config script so that, when run on Solaris, it shows a
4124 -R library as well as a -L library.
4125
4126 14. The debugging options of pcretest (-d on the command line or D on a
4127 pattern) showed incorrect output for anything following an extended class
4128 that contained multibyte characters and which was followed by a quantifier.
4129
4130 15. Added optional support for general category Unicode character properties
4131 via the \p, \P, and \X escapes. Unicode property support implies UTF-8
4132 support. It adds about 90K to the size of the library. The meanings of the
4133 inbuilt class escapes such as \d and \s have NOT been changed.
4134
4135 16. Updated pcredemo.c to include calls to free() to release the memory for the
4136 compiled pattern.
4137
4138 17. The generated file chartables.c was being created in the source directory
4139 instead of in the building directory. This caused the build to fail if the
4140 source directory was different from the building directory, and was
4141 read-only.
4142
4143 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
4144 file. No doubt somebody will tell me if they don't make sense... Also added
4145 Dan Mooney's comments about building on OpenVMS.
4146
4147 19. Added support for partial matching via the PCRE_PARTIAL option for
4148 pcre_exec() and the \P data escape in pcretest.
4149
4150 20. Extended pcretest with 3 new pattern features:
4151
4152 (i) A pattern option of the form ">rest-of-line" causes pcretest to
4153 write the compiled pattern to the file whose name is "rest-of-line".
4154 This is a straight binary dump of the data, with the saved pointer to
4155 the character tables forced to be NULL. The study data, if any, is
4156 written too. After writing, pcretest reads a new pattern.
4157
4158 (ii) If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
4159 compiled pattern from the given file. There must not be any
4160 occurrences of "<" in the file name (pretty unlikely); if there are,
4161 pcretest will instead treat the initial "<" as a pattern delimiter.
4162 After reading in the pattern, pcretest goes on to read data lines as
4163 usual.
4164
4165 (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
4166 and 16-bit fields in a compiled pattern, to simulate a pattern that
4167 was compiled on a host of opposite endianness.
4168
4169 21. The pcre-exec() function can now cope with patterns that were compiled on
4170 hosts of opposite endianness, with this restriction:
4171
4172 As for any compiled expression that is saved and used later, the tables
4173 pointer field cannot be preserved; the extra_data field in the arguments
4174 to pcre_exec() should be used to pass in a tables address if a value
4175 other than the default internal tables were used at compile time.
4176
4177 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
4178 now diagnosed as an error. Previously, most of the time, a negative number
4179 would have been treated as zero, but if in addition "ovector" was passed as
4180 NULL, a crash could occur.
4181
4182 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
4183 new versions from the libtool 1.5 distribution (the last one is a copy of
4184 a file called libtool.m4). This seems to have fixed the need to patch
4185 "configure" to support Darwin 1.3 (which I used to do). However, I still
4186 had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
4187 workstation).
4188
4189 24. Changed the PCRE licence to be the more standard "BSD" licence.
4190
4191
4192 Version 4.5 01-Dec-03
4193 ---------------------
4194
4195 1. There has been some re-arrangement of the code for the match() function so
4196 that it can be compiled in a version that does not call itself recursively.
4197 Instead, it keeps those local variables that need separate instances for
4198 each "recursion" in a frame on the heap, and gets/frees frames whenever it
4199 needs to "recurse". Keeping track of where control must go is done by means
4200 of setjmp/longjmp. The whole thing is implemented by a set of macros that
4201 hide most of the details from the main code, and operates only if
4202 NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
4203 "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
4204 operating.
4205
4206 To make it easier for callers to provide specially tailored get/free
4207 functions for this usage, two new functions, pcre_stack_malloc, and
4208 pcre_stack_free, are used. They are always called in strict stacking order,
4209 and the size of block requested is always the same.
4210
4211 The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
4212 PCRE has been compiled to use the stack or the heap for recursion. The
4213 -C option of pcretest uses this to show which version is compiled.
4214
4215 A new data escape \S, is added to pcretest; it causes the amounts of store
4216 obtained and freed by both kinds of malloc/free at match time to be added
4217 to the output.
4218
4219 2. Changed the locale test to use "fr_FR" instead of "fr" because that's
4220 what's available on my current Linux desktop machine.
4221
4222 3. When matching a UTF-8 string, the test for a valid string at the start has
4223 been extended. If start_offset is not zero, PCRE now checks that it points
4224 to a byte that is the start of a UTF-8 character. If not, it returns
4225 PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
4226 this is necessary because there may be backward assertions in the pattern.
4227 When matching the same subject several times, it may save resources to use
4228 PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
4229
4230 4. The code for checking the validity of UTF-8 strings has been tightened so
4231 that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
4232 containing "overlong sequences".
4233
4234 5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
4235 I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
4236 should have been "&", but it just so happened that all the cases this let
4237 through by mistake were picked up later in the function.
4238
4239 6. I had used a variable called "isblank" - this is a C99 function, causing
4240 some compilers to warn. To avoid this, I renamed it (as "blankclass").
4241
4242 7. Cosmetic: (a) only output another newline at the end of pcretest if it is
4243 prompting; (b) run "./pcretest /dev/null" at the start of the test script
4244 so the version is shown; (c) stop "make test" echoing "./RunTest".
4245
4246 8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
4247
4248 9. The prototype for memmove() for systems that don't have it was using
4249 size_t, but the inclusion of the header that defines size_t was later. I've
4250 moved the #includes for the C headers earlier to avoid this.
4251
4252 10. Added some adjustments to the code to make it easier to compiler on certain
4253 special systems:
4254
4255 (a) Some "const" qualifiers were missing.
4256 (b) Added the macro EXPORT before all exported functions; by default this
4257 is defined to be empty.
4258 (c) Changed the dftables auxiliary program (that builds chartables.c) so
4259 that it reads its output file name as an argument instead of writing
4260 to the standard output and assuming this can be redirected.
4261
4262 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
4263 class containing characters with values greater than 255, PCRE compilation
4264 went into a loop.
4265
4266 12. A recursive reference to a subpattern that was within another subpattern
4267 that had a minimum quantifier of zero caused PCRE to crash. For example,
4268 (x(y(?2))z)? provoked this bug with a subject that got as far as the
4269 recursion. If the recursively-called subpattern itself had a zero repeat,
4270 that was OK.
4271
4272 13. In pcretest, the buffer for reading a data line was set at 30K, but the
4273 buffer into which it was copied (for escape processing) was still set at
4274 1024, so long lines caused crashes.
4275
4276 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
4277 "internal error: code overflow...". This applied to any character class
4278 that was followed by a possessive quantifier.
4279
4280 15. Modified the Makefile to add libpcre.la as a prerequisite for
4281 libpcreposix.la because I was told this is needed for a parallel build to
4282 work.
4283
4284 16. If a pattern that contained .* following optional items at the start was
4285 studied, the wrong optimizing data was generated, leading to matching
4286 errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
4287 matching string must start with a or b or c. The correct conclusion for
4288 this pattern is that a match can start with any character.
4289
4290
4291 Version 4.4 13-Aug-03
4292 ---------------------
4293
4294 1. In UTF-8 mode, a character class containing characters with values between
4295 127 and 255 was not handled correctly if the compiled pattern was studied.
4296 In fixing this, I have also improved the studying algorithm for such
4297 classes (slightly).
4298
4299 2. Three internal functions had redundant arguments passed to them. Removal
4300 might give a very teeny performance improvement.
4301
4302 3. Documentation bug: the value of the capture_top field in a callout is *one
4303 more than* the number of the hightest numbered captured substring.
4304
4305 4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
4306 in incorrectly linking with a previously installed version. They now link
4307 explicitly with libpcre.la.
4308
4309 5. configure.in no longer needs to recognize Cygwin specially.
4310
4311 6. A problem in pcre.in for Windows platforms is fixed.
4312
4313 7. If a pattern was successfully studied, and the -d (or /D) flag was given to
4314 pcretest, it used to include the size of the study block as part of its
4315 output. Unfortunately, the structure contains a field that has a different
4316 size on different hardware architectures. This meant that the tests that
4317 showed this size failed. As the block is currently always of a fixed size,
4318 this information isn't actually particularly useful in pcretest output, so
4319 I have just removed it.
4320
4321 8. Three pre-processor statements accidentally did not start in column 1.
4322 Sadly, there are *still* compilers around that complain, even though
4323 standard C has not required this for well over a decade. Sigh.
4324
4325 9. In pcretest, the code for checking callouts passed small integers in the
4326 callout_data field, which is a void * field. However, some picky compilers
4327 complained about the casts involved for this on 64-bit systems. Now
4328 pcretest passes the address of the small integer instead, which should get
4329 rid of the warnings.
4330
4331 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
4332 both compile and run time, and gives an error if an invalid UTF-8 sequence
4333 is found. There is a option for disabling this check in cases where the
4334 string is known to be correct and/or the maximum performance is wanted.
4335
4336 11. In response to a bug report, I changed one line in Makefile.in from
4337
4338 -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
4339 to
4340 -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
4341
4342 to look similar to other lines, but I have no way of telling whether this
4343 is the right thing to do, as I do not use Windows. No doubt I'll get told
4344 if it's wrong...
4345
4346
4347 Version 4.3 21-May-03
4348 ---------------------
4349
4350 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
4351 Makefile.
4352
4353 2. Some refactoring to improve the quality of the code:
4354
4355 (i) The utf8_table... variables are now declared "const".
4356
4357 (ii) The code for \cx, which used the "case flipping" table to upper case
4358 lower case letters, now just substracts 32. This is ASCII-specific,
4359 but the whole concept of \cx is ASCII-specific, so it seems
4360 reasonable.
4361
4362 (iii) PCRE was using its character types table to recognize decimal and
4363 hexadecimal digits in the pattern. This is silly, because it handles
4364 only 0-9, a-f, and A-F, but the character types table is locale-
4365 specific, which means strange things might happen. A private
4366 table is now used for this - though it costs 256 bytes, a table is
4367 much faster than multiple explicit tests. Of course, the standard
4368 character types table is still used for matching digits in subject
4369 strings against \d.
4370
4371 (iv) Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
4372 ending in _t are). So I've renamed it as ESC_tee.
4373
4374 3. The first argument for regexec() in the POSIX wrapper should have been
4375 defined as "const".
4376
4377 4. Changed pcretest to use malloc() for its buffers so that they can be
4378 Electric Fenced for debugging.
4379
4380 5. There were several places in the code where, in UTF-8 mode, PCRE would try
4381 to read one or more bytes before the start of the subject string. Often this
4382 had no effect on PCRE's behaviour, but in some circumstances it could
4383 provoke a segmentation fault.
4384
4385 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
4386 to try to read one or more bytes before the start of the subject string.
4387
4388 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
4389 UTF-8 support could misbehave in various ways if the subject string
4390 contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
4391 area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
4392 back over UTF-8 characters.)
4393
4394
4395 Version 4.2 14-Apr-03
4396 ---------------------
4397
4398 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
4399
4400 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
4401 [ON_WINDOWS]: new variable, "#" on non-Windows platforms
4402 [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
4403 [WIN_PREFIX]: new variable, "cyg" for Cygwin
4404 * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
4405 and BUILD_EXEEXT
4406 Note: automatic setting of the BUILD variables is not yet working
4407 set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
4408 compile-time but not at link-time
4409 [LINK]: use for linking executables only
4410 make different versions for Windows and non-Windows
4411 [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
4412 libraries
4413 [LINK_FOR_BUILD]: new variable
4414 [OBJEXT]: use throughout
4415 [EXEEXT]: use throughout
4416 <winshared>: new target
4417 <wininstall>: new target
4418 <dftables.o>: use native compiler
4419 <dftables>: use native linker
4420 <install>: handle Windows platform correctly
4421 <clean>: ditto
4422 <check>: ditto
4423 copy DLL to top builddir before testing
4424
4425 As part of these changes, -no-undefined was removed again. This was reported
4426 to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
4427 in any case.
4428
4429 3. Some tidies to get rid of compiler warnings:
4430
4431 . In the match_data structure, match_limit was an unsigned long int, whereas
4432 match_call_count was an int. I've made them both unsigned long ints.
4433
4434 . In pcretest the fact that a const uschar * doesn't automatically cast to
4435 a void * provoked a warning.
4436
4437 . Turning on some more compiler warnings threw up some "shadow" variables
4438 and a few more missing casts.
4439
4440 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
4441 option, a class that contained a single character with a value between 128
4442 and 255 (e.g. /[\xFF]/) caused PCRE to crash.
4443
4444 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
4445 option, a class that contained several characters, but with at least one
4446 whose value was between 128 and 255 caused PCRE to crash.
4447
4448
4449 Version 4.1 12-Mar-03
4450 ---------------------
4451
4452 1. Compiling with gcc -pedantic found a couple of places where casts were
4453 needed, and a string in dftables.c that was longer than standard compilers are
4454 required to support.
4455
4456 2. Compiling with Sun's compiler found a few more places where the code could
4457 be tidied up in order to avoid warnings.
4458
4459 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
4460 first of these names is deprecated in the latest Autoconf in favour of the name
4461 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
4462 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
4463 analogy I have changed it to CFLAGS_FOR_BUILD.
4464
4465 4. Added -no-undefined to the linking command in the Makefile, because this is
4466 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
4467 linking step for the pcreposix library.
4468
4469 5. PCRE was failing to diagnose the case of two named groups with the same
4470 name.
4471
4472 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
4473 literal character that is needed in the subject for a match, and scans along to
4474 ensure that it is present before embarking on the full matching process. This
4475 saves time in cases of nested unlimited repeats that are never going to match.
4476 Problem: the scan can take a lot of time if the subject is very long (e.g.
4477 megabytes), thus penalizing straightforward matches. It is now done only if the
4478 amount of subject to be scanned is less than 1000 bytes.
4479
4480 7. A lesser problem with the same optimization is that it was recording the
4481 first character of an anchored pattern as "needed", thus provoking a search
4482 right along the subject, even when the first match of the pattern was going to
4483 fail. The "needed" character is now not set for anchored patterns, unless it
4484 follows something in the pattern that is of non-fixed length. Thus, it still
4485 fulfils its original purpose of finding quick non-matches in cases of nested
4486 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
4487
4488
4489 Version 4.0 17-Feb-03
4490 ---------------------
4491
4492 1. If a comment in an extended regex that started immediately after a meta-item
4493 extended to the end of string, PCRE compiled incorrect data. This could lead to
4494 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
4495
4496 2. Moved to autoconf 2.53 and libtool 1.4.2.
4497
4498 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
4499 the special perltest8 script is no longer needed - all the tests can be run
4500 from a single perltest script.
4501
4502 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
4503 by \s. It has now been removed in PCRE. This means it isn't recognized as
4504 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
4505 class [:space:] *does* include VT, thereby creating a mess.
4506
4507 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
4508 space and tab.
4509
4510 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
4511 its new features into the main test script, reducing the number of scripts.
4512
4513 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
4514 were backward compatible, and made the (?i) apply to the whole pattern, as if
4515 /i were given. Now it behaves more logically, and applies the option setting
4516 only to what follows. PCRE has been changed to follow suit. However, if it
4517 finds options settings right at the start of the pattern, it extracts them into
4518 the global options, as before. Thus, they show up in the info data.
4519
4520 8. Added support for the \Q...\E escape sequence. Characters in between are
4521 treated as literals. This is slightly different from Perl in that $ and @ are
4522 also handled as literals inside the quotes. In Perl, they will cause variable
4523 interpolation. Note the following examples:
4524
4525 Pattern PCRE matches Perl matches
4526
4527 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
4528 \Qabc\$xyz\E abc\$xyz abc\$xyz
4529 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
4530
4531 For compatibility with Perl, \Q...\E sequences are recognized inside character
4532 classes as well as outside them.
4533
4534 9. Re-organized 3 code statements in pcretest to avoid "overflow in
4535 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
4536 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
4537 signed/unsigned warnings.
4538
4539 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
4540 option for pcretest, so I've replaced it by a simple function that does just
4541 that job.
4542
4543 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
4544 "pcregrep -".
4545
4546 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
4547 Java package. This provides some syntactic sugar for simple cases of what my
4548 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
4549 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
4550 item, you can use this simplified notation. Note that only makes sense with
4551 greedy quantifiers. Consequently, the use of the possessive quantifier forces
4552 greediness, whatever the setting of the PCRE_UNGREEDY option.
4553
4554 13. A change of greediness default within a pattern was not taking effect at
4555 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
4556 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
4557 was abstracted outside.
4558
4559 14. PCRE now supports the \G assertion. It is true when the current matching
4560 position is at the start point of the match. This differs from \A when the
4561 starting offset is non-zero. Used with the /g option of pcretest (or similar
4562 code), it works in the same way as it does for Perl's /g option. If all
4563 alternatives of a regex begin with \G, the expression is anchored to the start
4564 match position, and the "anchored" flag is set in the compiled expression.
4565
4566 15. Some bugs concerning the handling of certain option changes within patterns
4567 have been fixed. These applied to options other than (?ims). For example,
4568 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
4569 been the other way round. Some of this was related to change 7 above.
4570
4571 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
4572 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
4573 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
4574 POSIX classes only within a class (e.g. /[[:alpha:]]/).
4575
4576 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
4577 mode. Unlike ".", it always matches newline, whatever the setting of
4578 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
4579 assertions. Perl allows it, but it doesn't (in general) work because it can't
4580 calculate the length of the lookbehind. At least, that's the case for Perl
4581 5.8.0 - I've been told they are going to document that it doesn't work in
4582 future.
4583
4584 18. Added an error diagnosis for escapes that PCRE does not support: these are
4585 \L, \l, \N, \P, \p, \U, \u, and \X.
4586
4587 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
4588 reading past the end of the pattern in cases such as /[abcd/.
4589
4590 20. PCRE was getting more memory than necessary for patterns with classes that
4591 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
4592
4593 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
4594 compiling PCRE for use with Virtual Pascal.
4595
4596 22. Small fix to the Makefile to make it work properly if the build is done
4597 outside the source tree.
4598
4599 23. Added a new extension: a condition to go with recursion. If a conditional
4600 subpattern starts with (?(R) the "true" branch is used if recursion has
4601 happened, whereas the "false" branch is used only at the top level.
4602
4603 24. When there was a very long string of literal characters (over 255 bytes
4604 without UTF support, over 250 bytes with UTF support), the computation of how
4605 much memory was required could be incorrect, leading to segfaults or other
4606 strange effects.
4607
4608 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
4609 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
4610 there was a subsequent back reference to those brackets. This meant that, for
4611 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
4612 possible to check for precisely this case. All we can do is abandon the
4613 optimization if .* occurs inside capturing brackets when there are any back
4614 references whatsoever. (See below for a better fix that came later.)
4615
4616 26. The handling of the optimization for finding the first character of a
4617 non-anchored pattern, and for finding a character that is required later in the
4618 match were failing in some cases. This didn't break the matching; it just
4619 failed to optimize when it could. The way this is done has been re-implemented.
4620
4621 27. Fixed typo in error message for invalid (?R item (it said "(?p").
4622
4623 28. Added a new feature that provides some of the functionality that Perl
4624 provides with (?{...}). The facility is termed a "callout". The way it is done
4625 in PCRE is for the caller to provide an optional function, by setting
4626 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
4627 global variable. By default it is unset, which disables all calling out. To get
4628 the function called, the regex must include (?C) at appropriate points. This
4629 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
4630 This provides a means of identifying different callout points. When PCRE
4631 reaches such a point in the regex, if pcre_callout has been set, the external
4632 function is called. It is provided with data in a structure called
4633 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
4634 matching continues; if it returns a non-zero value, the match at the current
4635 point fails. However, backtracking will occur if possible. [This was changed
4636 later and other features added - see item 49 below.]
4637
4638 29. pcretest is upgraded to test the callout functionality. It provides a
4639 callout function that displays information. By default, it shows the start of
4640 the match and the current position in the text. There are some new data escapes
4641 to vary what happens:
4642
4643 \C+ in addition, show current contents of captured substrings
4644 \C- do not supply a callout function
4645 \C!n return 1 when callout number n is reached
4646 \C!n!m return 1 when callout number n is reached for the mth time
4647
4648 30. If pcregrep was called with the -l option and just a single file name, it
4649 output "<stdin>" if a match was found, instead of the file name.
4650
4651 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
4652 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
4653 pcre_exec(). This saves a malloc/free per call. The default value of
4654 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
4655 when configuring.
4656
4657 32. The default maximum size of a compiled pattern is 64K. There have been a
4658 few cases of people hitting this limit. The code now uses macros to handle the
4659 storing of links as offsets within the compiled pattern. It defaults to 2-byte
4660 links, but this can be changed to 3 or 4 bytes by --with-link-size when
4661 configuring. Tests 2 and 5 work only with 2-byte links because they output
4662 debugging information about compiled patterns.
4663
4664 33. Internal code re-arrangements:
4665
4666 (a) Moved the debugging function for printing out a compiled regex into
4667 its own source file (printint.c) and used #include to pull it into
4668 pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
4669 separate copies.
4670
4671 (b) Defined the list of op-code names for debugging as a macro in
4672 internal.h so that it is next to the definition of the opcodes.
4673
4674 (c) Defined a table of op-code lengths for simpler skipping along compiled
4675 code. This is again a macro in internal.h so that it is next to the
4676 definition of the opcodes.
4677
4678 34. Added support for recursive calls to individual subpatterns, along the
4679 lines of Robin Houston's patch (but implemented somewhat differently).
4680
4681 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
4682 allow it to read and process whole directories in Win32. This code was
4683 contributed by Lionel Fourquaux; it has not been tested by me.
4684
4685 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
4686 used to name a group. Names consist of alphanumerics and underscores, and must
4687 be unique. Back references use the syntax (?P=name) and recursive calls use
4688 (?P>name) which is a PCRE extension to the Python extension. Groups still have
4689 numbers. The function pcre_fullinfo() can be used after compilation to extract
4690 a name/number map. There are three relevant calls:
4691
4692 PCRE_INFO_NAMEENTRYSIZE yields the size of each entry in the map
4693 PCRE_INFO_NAMECOUNT yields the number of entries
4694 PCRE_INFO_NAMETABLE yields a pointer to the map.
4695
4696 The map is a vector of fixed-size entries. The size of each entry depends on
4697 the length of the longest name used. The first two bytes of each entry are the
4698 group number, most significant byte first. There follows the corresponding
4699 name, zero terminated. The names are in alphabetical order.
4700
4701 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
4702 case instead of 255. Making it the same both with and without UTF-8 support
4703 means that the same test output works with both.
4704
4705 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
4706 calling malloc() with a zero argument.
4707
4708 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
4709 optimization. I've improved things by keeping a bitmap of backreferences with
4710 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
4711 fact referenced, the optimization can be applied. It is unlikely that a
4712 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
4713 the match to follow \n) will appear inside brackets with a number greater than
4714 31, but if it does, any back reference > 31 suppresses the optimization.
4715
4716 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
4717 of disabling numbered capturing parentheses. Any opening parenthesis that is
4718 not followed by ? behaves as if it were followed by ?: but named parentheses
4719 can still be used for capturing (and they will acquire numbers in the usual
4720 way).
4721
4722 41. Redesigned the return codes from the match() function into yes/no/error so
4723 that errors can be passed back from deep inside the nested calls. A malloc
4724 failure while inside a recursive subpattern call now causes the
4725 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
4726
4727 42. It is now possible to set a limit on the number of times the match()
4728 function is called in a call to pcre_exec(). This facility makes it possible to
4729 limit the amount of recursion and backtracking, though not in a directly
4730 obvious way, because the match() function is used in a number of different
4731 circumstances. The count starts from zero for each position in the subject
4732 string (for non-anchored patterns). The default limit is, for compatibility, a
4733 large number, namely 10 000 000. You can change this in two ways:
4734
4735 (a) When configuring PCRE before making, you can use --with-match-limit=n
4736 to set a default value for the compiled library.
4737
4738 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
4739 a different value is set. See 45 below.
4740
4741 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
4742
4743 43. Added a new function pcre_config(int, void *) to enable run-time extraction
4744 of things that can be changed at compile time. The first argument specifies
4745 what is wanted and the second points to where the information is to be placed.
4746 The current list of available information is:
4747
4748 PCRE_CONFIG_UTF8
4749
4750 The output is an integer that is set to one if UTF-8 support is available;
4751 otherwise it is set to zero.
4752
4753 PCRE_CONFIG_NEWLINE
4754
4755 The output is an integer that it set to the value of the code that is used for
4756 newline. It is either LF (10) or CR (13).
4757
4758 PCRE_CONFIG_LINK_SIZE
4759
4760 The output is an integer that contains the number of bytes used for internal
4761 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
4762
4763 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
4764
4765 The output is an integer that contains the threshold above which the POSIX
4766 interface uses malloc() for output vectors. See item 31 above.
4767
4768 PCRE_CONFIG_MATCH_LIMIT
4769
4770 The output is an unsigned integer that contains the default limit of the number
4771 of match() calls in a pcre_exec() execution. See 42 above.
4772
4773 44. pcretest has been upgraded by the addition of the -C option. This causes it
4774 to extract all the available output from the new pcre_config() function, and to
4775 output it. The program then exits immediately.
4776
4777 45. A need has arisen to pass over additional data with calls to pcre_exec() in
4778 order to support additional features. One way would have been to define
4779 pcre_exec2() (for example) with extra arguments, but this would not have been
4780 extensible, and would also have required all calls to the original function to
4781 be mapped to the new one. Instead, I have chosen to extend the mechanism that
4782 is used for passing in "extra" data from pcre_study().
4783
4784 The pcre_extra structure is now exposed and defined in pcre.h. It currently
4785 contains the following fields:
4786
4787 flags a bitmap indicating which of the following fields are set
4788 study_data opaque data from pcre_study()
4789 match_limit a way of specifying a limit on match() calls for a specific
4790 call to pcre_exec()
4791 callout_data data for callouts (see 49 below)
4792
4793 The flag bits are also defined in pcre.h, and are
4794
4795 PCRE_EXTRA_STUDY_DATA
4796 PCRE_EXTRA_MATCH_LIMIT
4797 PCRE_EXTRA_CALLOUT_DATA
4798
4799 The pcre_study() function now returns one of these new pcre_extra blocks, with
4800 the actual study data pointed to by the study_data field, and the
4801 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
4802 before. That is, this change is entirely upwards-compatible and requires no
4803 change to existing code.
4804
4805 If you want to pass in additional data to pcre_exec(), you can either place it
4806 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
4807 block.
4808
4809 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
4810 data string contains the escape sequence \M, pcretest calls pcre_exec() several
4811 times with different match limits, until it finds the minimum value needed for
4812 pcre_exec() to complete. The value is then output. This can be instructive; for
4813 most simple matches the number is quite small, but for pathological cases it
4814 gets very large very quickly.
4815
4816 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
4817 returns the size of the data block pointed to by the study_data field in a
4818 pcre_extra block, that is, the value that was passed as the argument to
4819 pcre_malloc() when PCRE was getting memory in which to place the information
4820 created by pcre_study(). The fourth argument should point to a size_t variable.
4821 pcretest has been extended so that this information is shown after a successful
4822 pcre_study() call when information about the compiled regex is being displayed.
4823
4824 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
4825 because what follows is always an absolute path. (Later: it turns out that this
4826 is more than cosmetic for MinGW, because it doesn't like empty path
4827 components.)
4828
4829 49. Some changes have been made to the callout feature (see 28 above):
4830
4831 (i) A callout function now has three choices for what it returns:
4832
4833 0 => success, carry on matching
4834 > 0 => failure at this point, but backtrack if possible
4835 < 0 => serious error, return this value from pcre_exec()
4836
4837 Negative values should normally be chosen from the set of PCRE_ERROR_xxx
4838 values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
4839 "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
4840 use by callout functions. It will never be used by PCRE itself.
4841
4842 (ii) The pcre_extra structure (see 45 above) has a void * field called
4843 callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
4844 pcre_callout_block structure has a field of the same name. The contents of
4845 the field passed in the pcre_extra structure are passed to the callout
4846 function in the corresponding field in the callout block. This makes it
4847 easier to use the same callout-containing regex from multiple threads. For
4848 testing, the pcretest program has a new data escape
4849
4850 \C*n pass the number n (may be negative) as callout_data
4851
4852 If the callout function in pcretest receives a non-zero value as
4853 callout_data, it returns that value.
4854
4855 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
4856 there were some redundant $(CFLAGS) in commands that are now specified as
4857 $(LINK), which already includes $(CFLAGS).
4858
4859 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
4860 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
4861 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
4862 one-byte characters throughout. Note that case-insensitive matching applies
4863 only to characters whose values are less than 256. PCRE doesn't support the
4864 notion of cases for higher-valued characters.
4865
4866 (i) A character class whose characters are all within 0-255 is handled as
4867 a bit map, and the map is inverted for negative classes. Previously, a
4868 character > 255 always failed to match such a class; however it should
4869 match if the class was a negative one (e.g. [^ab]). This has been fixed.
4870
4871 (ii) A negated character class with a single character < 255 is coded as
4872 "not this character" (OP_NOT). This wasn't working properly when the test
4873 character was multibyte, either singly or repeated.
4874
4875 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
4876 mode, for example: \x{100}{2,3}.
4877
4878 (iv) The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
4879 singly or repeated) now correctly test multibyte characters. However,
4880 PCRE doesn't recognize any characters with values greater than 255 as
4881 digits, spaces, or word characters. Such characters always match \D, \S,
4882 and \W, and never match \d, \s, or \w.
4883
4884 (v) Classes may now contain characters and character ranges with values
4885 greater than 255. For example: [ab\x{100}-\x{400}].
4886
4887 (vi) pcregrep now has a --utf-8 option (synonym -u) which makes it call
4888 PCRE in UTF-8 mode.
4889
4890 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
4891 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
4892 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
4893 value.)
4894
4895 53. The single man page has become too large. I have therefore split it up into
4896 a number of separate man pages. These also give rise to individual HTML pages;
4897 these are now put in a separate directory, and there is an index.html page that
4898 lists them all. Some hyperlinking between the pages has been installed.
4899
4900 54. Added convenience functions for handling named capturing parentheses.
4901
4902 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
4903 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
4904 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
4905 are faulted.
4906
4907 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
4908 calling configure. These values are used when compiling the dftables.c program
4909 which is run to generate the source of the default character tables. They
4910 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
4911 you will need to set these values.
4912
4913 57. Updated the building process for Windows DLL, as provided by Fred Cox.
4914
4915
4916 Version 3.9 02-Jan-02
4917 ---------------------
4918
4919 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
4920
4921 2. If --disable-static was given, the building process failed when trying to
4922 build pcretest and pcregrep. (For some reason it was using libtool to compile
4923 them, which is not right, as they aren't part of the library.)
4924
4925
4926 Version 3.8 18-Dec-01
4927 ---------------------
4928
4929 1. The experimental UTF-8 code was completely screwed up. It was packing the
4930 bytes in the wrong order. How dumb can you get?
4931
4932
4933 Version 3.7 29-Oct-01
4934 ---------------------
4935
4936 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
4937 This caused pcretest, when used on the test data, to segfault. Unfortunately,
4938 this didn't happen under Solaris 8, where I normally test things.
4939
4940 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
4941 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
4942 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
4943 this fix an hour or so after the initial 3.7 release.)
4944
4945
4946 Version 3.6 23-Oct-01
4947 ---------------------
4948
4949 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
4950 offsets passed as NULL with zero offset count.
4951
4952 2. The config.guess and config.sub files had not been updated when I moved to
4953 the latest autoconf.
4954
4955
4956 Version 3.5 15-Aug-01
4957 ---------------------
4958
4959 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
4960 had been forgotten.
4961
4962 2. By using declared but undefined structures, we can avoid using "void"
4963 definitions in pcre.h while keeping the internal definitions of the structures
4964 private.
4965
4966 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
4967 user point of view, this means that both static and shared libraries are built
4968 by default, but this can be individually controlled. More of the work of
4969 handling this static/shared cases is now inside libtool instead of PCRE's make
4970 file.
4971
4972 4. The pcretest utility is now installed along with pcregrep because it is
4973 useful for users (to test regexs) and by doing this, it automatically gets
4974 relinked by libtool. The documentation has been turned into a man page, so
4975 there are now .1, .txt, and .html versions in /doc.
4976
4977 5. Upgrades to pcregrep:
4978 (i) Added long-form option names like gnu grep.
4979 (ii) Added --help to list all options with an explanatory phrase.
4980 (iii) Added -r, --recursive to recurse into sub-directories.
4981 (iv) Added -f, --file to read patterns from a file.
4982
4983 6. pcre_exec() was referring to its "code" argument before testing that
4984 argument for NULL (and giving an error if it was NULL).
4985
4986 7. Upgraded Makefile.in to allow for compiling in a different directory from
4987 the source directory.
4988
4989 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
4990 options bits, the pointer it was passed was to an int instead of to an unsigned
4991 long int. This mattered only on 64-bit systems.
4992
4993 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
4994 generated) instead of pcre.in, which it its source. Also made the same change
4995 in several of the .c files.
4996
4997 10. A new release of gcc defines printf() as a macro, which broke pcretest
4998 because it had an ifdef in the middle of a string argument for printf(). Fixed
4999 by using separate calls to printf().
5000
5001 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
5002 script, to force use of CR or LF instead of \n in the source. On non-Unix
5003 systems, the value can be set in config.h.
5004
5005 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
5006 absolute limit. Changed the text of the error message to make this clear, and
5007 likewise updated the man page.
5008
5009 13. The limit of 99 on the number of capturing subpatterns has been removed.
5010 The new limit is 65535, which I hope will not be a "real" limit.
5011
5012
5013 Version 3.4 22-Aug-00
5014 ---------------------
5015
5016 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
5017
5018 2. Diagnose condition (?(0) as an error instead of crashing on matching.
5019
5020
5021 Version 3.3 01-Aug-00
5022 ---------------------
5023
5024 1. If an octal character was given, but the value was greater than \377, it
5025 was not getting masked to the least significant bits, as documented. This could
5026 lead to crashes in some systems.
5027
5028 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
5029 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
5030
5031 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
5032 These just pass their arguments on to (pcre_free)(), but they are provided
5033 because some uses of PCRE bind it to non-C systems that can call its functions,
5034 but cannot call free() or pcre_free() directly.
5035
5036 4. Add "make test" as a synonym for "make check". Corrected some comments in
5037 the Makefile.
5038
5039 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
5040 Makefile.
5041
5042 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
5043 command called pgrep for grepping around the active processes.
5044
5045 7. Added the beginnings of support for UTF-8 character strings.
5046
5047 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
5048 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
5049 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
5050 out for the ar command.)
5051
5052
5053 Version 3.2 12-May-00
5054 ---------------------
5055
5056 This is purely a bug fixing release.
5057
5058 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
5059 of ZA. This was just one example of several cases that could provoke this bug,
5060 which was introduced by change 9 of version 2.00. The code for breaking
5061 infinite loops after an iteration that matches an empty string was't working
5062 correctly.
5063
5064 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
5065 when matched against abbab (for example). After matching an empty string, it
5066 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
5067 caused it to match further down the string than it should.
5068
5069 3. The code contained an inclusion of sys/types.h. It isn't clear why this
5070 was there because it doesn't seem to be needed, and it causes trouble on some
5071 systems, as it is not a Standard C header. It has been removed.
5072
5073 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
5074 were reported on the Macintosh. The changes were from
5075
5076 while ((c = *(++ptr)) != 0 && c != '\n');
5077 to
5078 while ((c = *(++ptr)) != 0 && c != '\n') ;
5079
5080 Totally extraordinary, but if that's what it takes...
5081
5082 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
5083 available. Added HAVE_BCOPY and an autoconf test for it; if neither
5084 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
5085 assumes the way PCRE uses memmove() (always moving upwards).
5086
5087 6. PCRE is being used in one environment where strchr() is not available. There
5088 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
5089 faster code anyway.
5090
5091
5092 Version 3.1 09-Feb-00
5093 ---------------------
5094
5095 The only change in this release is the fixing of some bugs in Makefile.in for
5096 the "install" target:
5097
5098 (1) It was failing to install pcreposix.h.
5099
5100 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
5101
5102
5103 Version 3.0 01-Feb-00
5104 ---------------------
5105
5106 1. Add support for the /+ modifier to perltest (to output $` like it does in
5107 pcretest).
5108
5109 2. Add support for the /g modifier to perltest.
5110
5111 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
5112 matches null strings.
5113
5114 4. Fix perltest so that it doesn't do unwanted things when fed an empty
5115 pattern. Perl treats empty patterns specially - it reuses the most recent
5116 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
5117 effect.
5118
5119 5. The POSIX interface was broken in that it was just handing over the POSIX
5120 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
5121 required a bigger vector, with some working space on the end. This means that
5122 the POSIX wrapper now has to get and free some memory, and copy the results.
5123
5124 6. Added some simple autoconf support, placing the test data and the
5125 documentation in separate directories, re-organizing some of the
5126 information files, and making it build pcre-config (a GNU standard). Also added
5127 libtool support for building PCRE as a shared library, which is now the
5128 default.
5129
5130 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
5131 09 are not valid octal constants. Single digits will be used for minor values
5132 less than 10.
5133
5134 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
5135 existing programs that set these in the POSIX interface can use PCRE without
5136 modification.
5137
5138 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
5139 return all that pcre_info() returns, plus additional data. The pcre_info()
5140 function is retained for compatibility, but is considered to be obsolete.
5141
5142 10. Added experimental recursion feature (?R) to handle one common case that
5143 Perl 5.6 will be able to do with (?p{...}).
5144
5145 11. Added support for POSIX character classes like [:alpha:], which Perl is
5146 adopting.
5147
5148
5149 Version 2.08 31-Aug-99
5150 ----------------------
5151
5152 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
5153 trying to match at the startoffset position, but instead was moving forward to
5154 the next newline as if a previous match had failed.
5155
5156 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
5157 and could get into a loop if a null string was matched other than at the start
5158 of the subject.
5159
5160 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
5161 be distinguished at compile time, and for completeness also added PCRE_DATE.
5162
5163 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
5164 in GnuWin32 environments.
5165
5166
5167 Version 2.07 29-Jul-99
5168 ----------------------
5169
5170 1. The documentation is now supplied in plain text form and HTML as well as in
5171 the form of man page sources.
5172
5173 2. C++ compilers don't like assigning (void *) values to other pointer types.
5174 In particular this affects malloc(). Although there is no problem in Standard
5175 C, I've put in casts to keep C++ compilers happy.
5176
5177 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
5178 should be (const char *).
5179
5180 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
5181 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
5182 However, I haven't made this a standard facility. The documentation doesn't
5183 mention it, and the Makefile doesn't support it.
5184
5185 5. The Makefile now contains an "install" target, with editable destinations at
5186 the top of the file. The pcretest program is not installed.
5187
5188 6. pgrep -V now gives the PCRE version number and date.
5189
5190 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
5191 causing the entire string to be ignored, instead of just the last character.
5192
5193 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
5194 non-matching string, it can take a very, very long time, even for strings of
5195 quite modest length, because of the nested recursion. PCRE now does better in
5196 some of these cases. It does this by remembering the last required literal
5197 character in the pattern, and pre-searching the subject to ensure it is present
5198 before running the real match. In other words, it applies a heuristic to detect
5199 some types of certain failure quickly, and in the above example, if presented
5200 with a string that has no trailing " it gives "no match" very quickly.
5201
5202 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
5203 other alternatives are tried instead.
5204
5205
5206 Version 2.06 09-Jun-99
5207 ----------------------
5208
5209 1. Change pcretest's output for amount of store used to show just the code
5210 space, because the remainder (the data block) varies in size between 32-bit and
5211 64-bit systems.
5212
5213 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
5214 start matching at. This allows lookbehinds to work when searching for multiple
5215 occurrences in a string.
5216
5217 3. Added additional options to pcretest for testing multiple occurrences:
5218
5219 /+ outputs the rest of the string that follows a match
5220 /g loops for multiple occurrences, using the new startoffset argument
5221 /G loops for multiple occurrences by passing an incremented pointer
5222
5223 4. PCRE wasn't doing the "first character" optimization for patterns starting
5224 with \b or \B, though it was doing it for other lookbehind assertions. That is,
5225 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
5226 the letter 'x'. On long subject strings, this gives a significant speed-up.
5227
5228
5229 Version 2.05 21-Apr-99
5230 ----------------------
5231
5232 1. Changed the type of magic_number from int to long int so that it works
5233 properly on 16-bit systems.
5234
5235 2. Fixed a bug which caused patterns starting with .* not to work correctly
5236 when the subject string contained newline characters. PCRE was assuming
5237 anchoring for such patterns in all cases, which is not correct because .* will
5238 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
5239 DOTALL is set at top level; otherwise it knows that patterns starting with .*
5240 must be retried after every newline in the subject.
5241
5242
5243 Version 2.04 18-Feb-99
5244 ----------------------
5245
5246 1. For parenthesized subpatterns with repeats whose minimum was zero, the
5247 computation of the store needed to hold the pattern was incorrect (too large).
5248 If such patterns were nested a few deep, this could multiply and become a real
5249 problem.
5250
5251 2. Added /M option to pcretest to show the memory requirement of a specific
5252 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
5253
5254 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
5255 compiled in such a way that the backtracking after subsequent failure was
5256 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
5257 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
5258
5259
5260 Version 2.03 02-Feb-99
5261 ----------------------
5262
5263 1. Fixed typo and small mistake in man page.
5264
5265 2. Added 4th condition (GPL supersedes if conflict) and created separate
5266 LICENCE file containing the conditions.
5267
5268 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
5269 Perl, that is the internal \ allows the delimiter to be included in the
5270 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
5271 the final delimiter, add \ to the end of the pattern (to test the error).
5272
5273 4. Added the convenience functions for extracting substrings after a successful
5274 match. Updated pcretest to make it able to test these functions.
5275
5276
5277 Version 2.02 14-Jan-99
5278 ----------------------
5279
5280 1. Initialized the working variables associated with each extraction so that
5281 their saving and restoring doesn't refer to uninitialized store.
5282
5283 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
5284 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
5285 fix the problem.
5286
5287 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
5288 calls, and wasn't printing the correct value for compiling calls. Increased the
5289 default value of LOOPREPEAT, and the number of significant figures in the
5290 times.
5291
5292 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
5293
5294 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
5295 a building problem on Windows NT with a FAT file system.
5296
5297
5298 Version 2.01 21-Oct-98
5299 ----------------------
5300
5301 1. Changed the API for pcre_compile() to allow for the provision of a pointer
5302 to character tables built by pcre_maketables() in the current locale. If NULL
5303 is passed, the default tables are used.
5304
5305
5306 Version 2.00 24-Sep-98
5307 ----------------------
5308
5309 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
5310 it any more.
5311
5312 2. Allow quantification of (?>) groups, and make it work correctly.
5313
5314 3. The first character computation wasn't working for (?>) groups.
5315
5316 4. Correct the implementation of \Z (it is permitted to match on the \n at the
5317 end of the subject) and add 5.005's \z, which really does match only at the
5318 very end of the subject.
5319
5320 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
5321
5322 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
5323 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
5324 localized options. All options to pcre_study() were also removed.
5325
5326 7. Add other new features from 5.005:
5327
5328 $(?<= positive lookbehind
5329 $(?<! negative lookbehind
5330 (?imsx-imsx) added the unsetting capability
5331 such a setting is global if at outer level; local otherwise
5332 (?imsx-imsx:) non-capturing groups with option setting
5333 (?(cond)re|re) conditional pattern matching
5334
5335 A backreference to itself in a repeated group matches the previous
5336 captured string.
5337
5338 8. General tidying up of studying (both automatic and via "study")
5339 consequential on the addition of new assertions.
5340
5341 9. As in 5.005, unlimited repeated groups that could match an empty substring
5342 are no longer faulted at compile time. Instead, the loop is forcibly broken at
5343 runtime if any iteration does actually match an empty substring.
5344
5345 10. Include the RunTest script in the distribution.
5346
5347 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
5348 discrepancies, some of which were old and were also with respect to 5.004. They
5349 have now been fixed.
5350
5351
5352 Version 1.09 28-Apr-98
5353 ----------------------
5354
5355 1. A negated single character class followed by a quantifier with a minimum
5356 value of one (e.g. [^x]{1,6} ) was not compiled correctly. This could lead to
5357 program crashes, or just wrong answers. This did not apply to negated classes
5358 containing more than one character, or to minima other than one.
5359
5360
5361 Version 1.08 27-Mar-98
5362 ----------------------
5363
5364 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
5365
5366 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
5367 latter must appear before anything that relies on it in the pattern.
5368
5369
5370 Version 1.07 16-Feb-98
5371 ----------------------
5372
5373 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
5374 repeat of a potentially empty string).
5375
5376
5377 Version 1.06 23-Jan-98
5378 ----------------------
5379
5380 1. Added Markus Oberhumer's little patches for C++.
5381
5382 2. Literal strings longer than 255 characters were broken.
5383
5384
5385 Version 1.05 23-Dec-97
5386 ----------------------
5387
5388 1. Negated character classes containing more than one character were failing if
5389 PCRE_CASELESS was set at run time.
5390
5391
5392 Version 1.04 19-Dec-97
5393 ----------------------
5394
5395 1. Corrected the man page, where some "const" qualifiers had been omitted.
5396
5397 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
5398 input syntax.
5399
5400 3. Fixed memory leak which occurred when a regex with back references was
5401 matched with an offsets vector that wasn't big enough. The temporary memory
5402 that is used in this case wasn't being freed if the match failed.
5403
5404 4. Tidied pcretest to ensure it frees memory that it gets.
5405
5406 5. Temporary memory was being obtained in the case where the passed offsets
5407 vector was exactly big enough.
5408
5409 6. Corrected definition of offsetof() from change 5 below.
5410
5411 7. I had screwed up change 6 below and broken the rules for the use of
5412 setjmp(). Now fixed.
5413
5414
5415 Version 1.03 18-Dec-97
5416 ----------------------
5417
5418 1. A erroneous regex with a missing opening parenthesis was correctly
5419 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
5420 on some systems.
5421
5422 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
5423 it was reported that one broken compiler failed on the former because "code" is
5424 also an independent variable.
5425
5426 3. The erroneous regex a[]b caused an array overrun reference.
5427
5428 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
5429 fail on data ending with that character. (It was going on too far, and checking
5430 the next character, typically a binary zero.) This was specific to the
5431 optimized code for single-character negative classes.
5432
5433 5. Added a contributed patch from the TIN world which does the following:
5434
5435 + Add an undef for memmove, in case the the system defines a macro for it.
5436
5437 + Add a definition of offsetof(), in case there isn't one. (I don't know
5438 the reason behind this - offsetof() is part of the ANSI standard - but
5439 it does no harm).
5440
5441 + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
5442 most of the places where whitespace preceded '#'. I have given up and
5443 allowed the remaining 2 cases to be at the margin.
5444
5445 + Rename some variables in pcre to eliminate shadowing. This seems very
5446 pedantic, but does no harm, of course.
5447
5448 6. Moved the call to setjmp() into its own function, to get rid of warnings
5449 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
5450
5451 7. Constructs such as \d{8,} were compiling into the equivalent of
5452 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
5453 outcome, but in this particular case used more store than had been allocated,
5454 which caused the bug to be discovered because it threw up an internal error.
5455
5456 8. The debugging code in both pcre and pcretest for outputting the compiled
5457 form of a regex was going wrong in the case of back references followed by
5458 curly-bracketed repeats.
5459
5460
5461 Version 1.02 12-Dec-97
5462 ----------------------
5463
5464 1. Typos in pcre.3 and comments in the source fixed.
5465
5466 2. Applied a contributed patch to get rid of places where it used to remove
5467 'const' from variables, and fixed some signed/unsigned and uninitialized
5468 variable warnings.
5469
5470 3. Added the "runtest" target to Makefile.
5471
5472 4. Set default compiler flag to -O2 rather than just -O.
5473
5474
5475 Version 1.01 19-Nov-97
5476 ----------------------
5477
5478 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
5479 like /([ab]*)*/, that is, for classes with more than one character in them.
5480
5481 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
5482 as /((?>a*))*/ (a PCRE_EXTRA facility).
5483
5484
5485 Version 1.00 18-Nov-97
5486 ----------------------
5487
5488 1. Added compile-time macros to support systems such as SunOS4 which don't have
5489 memmove() or strerror() but have other things that can be used instead.
5490
5491 2. Arranged that "make clean" removes the executables.
5492
5493
5494 Version 0.99 27-Oct-97
5495 ----------------------
5496
5497 1. Fixed bug in code for optimizing classes with only one character. It was
5498 initializing a 32-byte map regardless, which could cause it to run off the end
5499 of the memory it had got.
5500
5501 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
5502
5503
5504 Version 0.98 22-Oct-97
5505 ----------------------
5506
5507 1. Fixed bug in code for handling temporary memory usage when there are more
5508 back references than supplied space in the ovector. This could cause segfaults.
5509
5510
5511 Version 0.97 21-Oct-97
5512 ----------------------
5513
5514 1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
5515
5516 2. Optimized negated single characters not to use a bit map.
5517
5518 3. Brought error texts together as macro definitions; clarified some of them;
5519 fixed one that was wrong - it said "range out of order" when it meant "invalid
5520 escape sequence".
5521
5522 4. Changed some char * arguments to const char *.
5523
5524 5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
5525
5526 6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
5527 pcretest.
5528
5529
5530 Version 0.96 16-Oct-97
5531 ----------------------
5532
5533 1. Added a simple "pgrep" utility to the distribution.
5534
5535 2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
5536 unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
5537 where "ddd" means "one or more decimal digits".
5538
5539 3. Fixed serious bug. If a pattern had a back reference, but the call to
5540 pcre_exec() didn't supply a large enough ovector to record the related
5541 identifying subpattern, the match always failed. PCRE now remembers the number
5542 of the largest back reference, and gets some temporary memory in which to save
5543 the offsets during matching if necessary, in order to ensure that
5544 backreferences always work.
5545
5546 4. Increased the compatibility with Perl in a number of ways:
5547
5548 (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
5549 to request this handling. The option can be set at compile or exec time.
5550
5551 (b) $ matches before a terminating newline by default; an option
5552 PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
5553 mode). The option can be set at compile or exec time.
5554
5555 (c) The handling of \ followed by a digit other than 0 is now supposed to be
5556 the same as Perl's. If the decimal number it represents is less than 10
5557 or there aren't that many previous left capturing parentheses, an octal
5558 escape is read. Inside a character class, it's always an octal escape,
5559 even if it is a single digit.
5560
5561 (d) An escaped but undefined alphabetic character is taken as a literal,
5562 unless PCRE_EXTRA is set. Currently this just reserves the remaining
5563 escapes.
5564
5565 (e) {0} is now permitted. (The previous item is removed from the compiled
5566 pattern).
5567
5568 5. Changed all the names of code files so that the basic parts are no longer
5569 than 10 characters, and abolished the teeny "globals.c" file.
5570
5571 6. Changed the handling of character classes; they are now done with a 32-byte
5572 bit map always.
5573
5574 7. Added the -d and /D options to pcretest to make it possible to look at the
5575 internals of compilation without having to recompile pcre.
5576
5577
5578 Version 0.95 23-Sep-97
5579 ----------------------
5580
5581 1. Fixed bug in pre-pass concerning escaped "normal" characters such as \x5c or
5582 \x20 at the start of a run of normal characters. These were being treated as
5583 real characters, instead of the source characters being re-checked.
5584
5585
5586 Version 0.94 18-Sep-97
5587 ----------------------
5588
5589 1. The functions are now thread-safe, with the caveat that the global variables
5590 containing pointers to malloc() and free() or alternative functions are the
5591 same for all threads.
5592
5593 2. Get pcre_study() to generate a bitmap of initial characters for non-
5594 anchored patterns when this is possible, and use it if passed to pcre_exec().
5595
5596
5597 Version 0.93 15-Sep-97
5598 ----------------------
5599
5600 1. /(b)|(:+)/ was computing an incorrect first character.
5601
5602 2. Add pcre_study() to the API and the passing of pcre_extra to pcre_exec(),
5603 but not actually doing anything yet.
5604
5605 3. Treat "-" characters in classes that cannot be part of ranges as literals,
5606 as Perl does (e.g. [-az] or [az-]).
5607
5608 4. Set the anchored flag if a branch starts with .* or .*? because that tests
5609 all possible positions.
5610
5611 5. Split up into different modules to avoid including unneeded functions in a
5612 compiled binary. However, compile and exec are still in one module. The "study"
5613 function is split off.
5614
5615 6. The character tables are now in a separate module whose source is generated
5616 by an auxiliary program - but can then be edited by hand if required. There are
5617 now no calls to isalnum(), isspace(), isdigit(), isxdigit(), tolower() or
5618 toupper() in the code.
5619
5620 7. Turn the malloc/free funtions variables into pcre_malloc and pcre_free and
5621 make them global. Abolish the function for setting them, as the caller can now
5622 set them directly.
5623
5624
5625 Version 0.92 11-Sep-97
5626 ----------------------
5627
5628 1. A repeat with a fixed maximum and a minimum of 1 for an ordinary character
5629 (e.g. /a{1,3}/) was broken (I mis-optimized it).
5630
5631 2. Caseless matching was not working in character classes if the characters in
5632 the pattern were in upper case.
5633
5634 3. Make ranges like [W-c] work in the same way as Perl for caseless matching.
5635
5636 4. Make PCRE_ANCHORED public and accept as a compile option.
5637
5638 5. Add an options word to pcre_exec() and accept PCRE_ANCHORED and
5639 PCRE_CASELESS at run time. Add escapes \A and \I to pcretest to cause it to
5640 pass them.
5641
5642 6. Give an error if bad option bits passed at compile or run time.
5643
5644 7. Add PCRE_MULTILINE at compile and exec time, and (?m) as well. Add \M to
5645 pcretest to cause it to pass that flag.
5646
5647 8. Add pcre_info(), to get the number of identifying subpatterns, the stored
5648 options, and the first character, if set.
5649
5650 9. Recognize C+ or C{n,m} where n >= 1 as providing a fixed starting character.
5651
5652
5653 Version 0.91 10-Sep-97
5654 ----------------------
5655
5656 1. PCRE was failing to diagnose unlimited repeats of subpatterns that could
5657 match the empty string as in /(a*)*/. It was looping and ultimately crashing.
5658
5659 2. PCRE was looping on encountering an indefinitely repeated back reference to
5660 a subpattern that had matched an empty string, e.g. /(a|)\1*/. It now does what
5661 Perl does - treats the match as successful.
5662
5663 ****

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