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

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1525 - (show annotations)
Wed Feb 11 16:48:35 2015 UTC (4 years, 8 months ago) by ph10
File size: 272014 byte(s)
Error occurred while calculating annotation data.
Fix bug that did not allow zero case for (a)*+ when ovector was too small to 
capture.
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
69 Version 8.36 26-September-2014
70 ------------------------------
71
72 1. Got rid of some compiler warnings in the C++ modules that were shown up by
73 -Wmissing-field-initializers and -Wunused-parameter.
74
75 2. The tests for quantifiers being too big (greater than 65535) were being
76 applied after reading the number, and stupidly assuming that integer
77 overflow would give a negative number. The tests are now applied as the
78 numbers are read.
79
80 3. Tidy code in pcre_exec.c where two branches that used to be different are
81 now the same.
82
83 4. The JIT compiler did not generate match limit checks for certain
84 bracketed expressions with quantifiers. This may lead to exponential
85 backtracking, instead of returning with PCRE_ERROR_MATCHLIMIT. This
86 issue should be resolved now.
87
88 5. Fixed an issue, which occures when nested alternatives are optimized
89 with table jumps.
90
91 6. Inserted two casts and changed some ints to size_t in the light of some
92 reported 64-bit compiler warnings (Bugzilla 1477).
93
94 7. Fixed a bug concerned with zero-minimum possessive groups that could match
95 an empty string, which sometimes were behaving incorrectly in the
96 interpreter (though correctly in the JIT matcher). This pcretest input is
97 an example:
98
99 '\A(?:[^"]++|"(?:[^"]*+|"")*+")++'
100 NON QUOTED "QUOT""ED" AFTER "NOT MATCHED
101
102 the interpreter was reporting a match of 'NON QUOTED ' only, whereas the
103 JIT matcher and Perl both matched 'NON QUOTED "QUOT""ED" AFTER '. The test
104 for an empty string was breaking the inner loop and carrying on at a lower
105 level, when possessive repeated groups should always return to a higher
106 level as they have no backtrack points in them. The empty string test now
107 occurs at the outer level.
108
109 8. Fixed a bug that was incorrectly auto-possessifying \w+ in the pattern
110 ^\w+(?>\s*)(?<=\w) which caused it not to match "test test".
111
112 9. Give a compile-time error for \o{} (as Perl does) and for \x{} (which Perl
113 doesn't).
114
115 10. Change 8.34/15 introduced a bug that caused the amount of memory needed
116 to hold a pattern to be incorrectly computed (too small) when there were
117 named back references to duplicated names. This could cause "internal
118 error: code overflow" or "double free or corruption" or other memory
119 handling errors.
120
121 11. When named subpatterns had the same prefixes, back references could be
122 confused. For example, in this pattern:
123
124 /(?P<Name>a)?(?P<Name2>b)?(?(<Name>)c|d)*l/
125
126 the reference to 'Name' was incorrectly treated as a reference to a
127 duplicate name.
128
129 12. A pattern such as /^s?c/mi8 where the optional character has more than
130 one "other case" was incorrectly compiled such that it would only try to
131 match starting at "c".
132
133 13. When a pattern starting with \s was studied, VT was not included in the
134 list of possible starting characters; this should have been part of the
135 8.34/18 patch.
136
137 14. If a character class started [\Qx]... where x is any character, the class
138 was incorrectly terminated at the ].
139
140 15. If a pattern that started with a caseless match for a character with more
141 than one "other case" was studied, PCRE did not set up the starting code
142 unit bit map for the list of possible characters. Now it does. This is an
143 optimization improvement, not a bug fix.
144
145 16. The Unicode data tables have been updated to Unicode 7.0.0.
146
147 17. Fixed a number of memory leaks in pcregrep.
148
149 18. Avoid a compiler warning (from some compilers) for a function call with
150 a cast that removes "const" from an lvalue by using an intermediate
151 variable (to which the compiler does not object).
152
153 19. Incorrect code was compiled if a group that contained an internal recursive
154 back reference was optional (had quantifier with a minimum of zero). This
155 example compiled incorrect code: /(((a\2)|(a*)\g<-1>))*/ and other examples
156 caused segmentation faults because of stack overflows at compile time.
157
158 20. A pattern such as /((?(R)a|(?1)))+/, which contains a recursion within a
159 group that is quantified with an indefinite repeat, caused a compile-time
160 loop which used up all the system stack and provoked a segmentation fault.
161 This was not the same bug as 19 above.
162
163 21. Add PCRECPP_EXP_DECL declaration to operator<< in pcre_stringpiece.h.
164 Patch by Mike Frysinger.
165
166
167 Version 8.35 04-April-2014
168 --------------------------
169
170 1. A new flag is set, when property checks are present in an XCLASS.
171 When this flag is not set, PCRE can perform certain optimizations
172 such as studying these XCLASS-es.
173
174 2. The auto-possessification of character sets were improved: a normal
175 and an extended character set can be compared now. Furthermore
176 the JIT compiler optimizes more character set checks.
177
178 3. Got rid of some compiler warnings for potentially uninitialized variables
179 that show up only when compiled with -O2.
180
181 4. A pattern such as (?=ab\K) that uses \K in an assertion can set the start
182 of a match later then the end of the match. The pcretest program was not
183 handling the case sensibly - it was outputting from the start to the next
184 binary zero. It now reports this situation in a message, and outputs the
185 text from the end to the start.
186
187 5. Fast forward search is improved in JIT. Instead of the first three
188 characters, any three characters with fixed position can be searched.
189 Search order: first, last, middle.
190
191 6. Improve character range checks in JIT. Characters are read by an inprecise
192 function now, which returns with an unknown value if the character code is
193 above a certain threshold (e.g: 256). The only limitation is that the value
194 must be bigger than the threshold as well. This function is useful when
195 the characters above the threshold are handled in the same way.
196
197 7. The macros whose names start with RAWUCHAR are placeholders for a future
198 mode in which only the bottom 21 bits of 32-bit data items are used. To
199 make this more memorable for those maintaining the code, the names have
200 been changed to start with UCHAR21, and an extensive comment has been added
201 to their definition.
202
203 8. Add missing (new) files sljitNativeTILEGX.c and sljitNativeTILEGX-encoder.c
204 to the export list in Makefile.am (they were accidentally omitted from the
205 8.34 tarball).
206
207 9. The informational output from pcretest used the phrase "starting byte set"
208 which is inappropriate for the 16-bit and 32-bit libraries. As the output
209 for "first char" and "need char" really means "non-UTF-char", I've changed
210 "byte" to "char", and slightly reworded the output. The documentation about
211 these values has also been (I hope) clarified.
212
213 10. Another JIT related optimization: use table jumps for selecting the correct
214 backtracking path, when more than four alternatives are present inside a
215 bracket.
216
217 11. Empty match is not possible, when the minimum length is greater than zero,
218 and there is no \K in the pattern. JIT should avoid empty match checks in
219 such cases.
220
221 12. In a caseless character class with UCP support, when a character with more
222 than one alternative case was not the first character of a range, not all
223 the alternative cases were added to the class. For example, s and \x{17f}
224 are both alternative cases for S: the class [RST] was handled correctly,
225 but [R-T] was not.
226
227 13. The configure.ac file always checked for pthread support when JIT was
228 enabled. This is not used in Windows, so I have put this test inside a
229 check for the presence of windows.h (which was already tested for).
230
231 14. Improve pattern prefix search by a simplified Boyer-Moore algorithm in JIT.
232 The algorithm provides a way to skip certain starting offsets, and usually
233 faster than linear prefix searches.
234
235 15. Change 13 for 8.20 updated RunTest to check for the 'fr' locale as well
236 as for 'fr_FR' and 'french'. For some reason, however, it then used the
237 Windows-specific input and output files, which have 'french' screwed in.
238 So this could never have worked. One of the problems with locales is that
239 they aren't always the same. I have now updated RunTest so that it checks
240 the output of the locale test (test 3) against three different output
241 files, and it allows the test to pass if any one of them matches. With luck
242 this should make the test pass on some versions of Solaris where it was
243 failing. Because of the uncertainty, the script did not used to stop if
244 test 3 failed; it now does. If further versions of a French locale ever
245 come to light, they can now easily be added.
246
247 16. If --with-pcregrep-bufsize was given a non-integer value such as "50K",
248 there was a message during ./configure, but it did not stop. This now
249 provokes an error. The invalid example in README has been corrected.
250 If a value less than the minimum is given, the minimum value has always
251 been used, but now a warning is given.
252
253 17. If --enable-bsr-anycrlf was set, the special 16/32-bit test failed. This
254 was a bug in the test system, which is now fixed. Also, the list of various
255 configurations that are tested for each release did not have one with both
256 16/32 bits and --enable-bar-anycrlf. It now does.
257
258 18. pcretest was missing "-C bsr" for displaying the \R default setting.
259
260 19. Little endian PowerPC systems are supported now by the JIT compiler.
261
262 20. The fast forward newline mechanism could enter to an infinite loop on
263 certain invalid UTF-8 input. Although we don't support these cases
264 this issue can be fixed by a performance optimization.
265
266 21. Change 33 of 8.34 is not sufficient to ensure stack safety because it does
267 not take account if existing stack usage. There is now a new global
268 variable called pcre_stack_guard that can be set to point to an external
269 function to check stack availability. It is called at the start of
270 processing every parenthesized group.
271
272 22. A typo in the code meant that in ungreedy mode the max/min qualifier
273 behaved like a min-possessive qualifier, and, for example, /a{1,3}b/U did
274 not match "ab".
275
276 23. When UTF was disabled, the JIT program reported some incorrect compile
277 errors. These messages are silenced now.
278
279 24. Experimental support for ARM-64 and MIPS-64 has been added to the JIT
280 compiler.
281
282 25. Change all the temporary files used in RunGrepTest to be different to those
283 used by RunTest so that the tests can be run simultaneously, for example by
284 "make -j check".
285
286
287 Version 8.34 15-December-2013
288 -----------------------------
289
290 1. Add pcre[16|32]_jit_free_unused_memory to forcibly free unused JIT
291 executable memory. Patch inspired by Carsten Klein.
292
293 2. ./configure --enable-coverage defined SUPPORT_GCOV in config.h, although
294 this macro is never tested and has no effect, because the work to support
295 coverage involves only compiling and linking options and special targets in
296 the Makefile. The comment in config.h implied that defining the macro would
297 enable coverage support, which is totally false. There was also support for
298 setting this macro in the CMake files (my fault, I just copied it from
299 configure). SUPPORT_GCOV has now been removed.
300
301 3. Make a small performance improvement in strlen16() and strlen32() in
302 pcretest.
303
304 4. Change 36 for 8.33 left some unreachable statements in pcre_exec.c,
305 detected by the Solaris compiler (gcc doesn't seem to be able to diagnose
306 these cases). There was also one in pcretest.c.
307
308 5. Cleaned up a "may be uninitialized" compiler warning in pcre_exec.c.
309
310 6. In UTF mode, the code for checking whether a group could match an empty
311 string (which is used for indefinitely repeated groups to allow for
312 breaking an infinite loop) was broken when the group contained a repeated
313 negated single-character class with a character that occupied more than one
314 data item and had a minimum repetition of zero (for example, [^\x{100}]* in
315 UTF-8 mode). The effect was undefined: the group might or might not be
316 deemed as matching an empty string, or the program might have crashed.
317
318 7. The code for checking whether a group could match an empty string was not
319 recognizing that \h, \H, \v, \V, and \R must match a character.
320
321 8. Implemented PCRE_INFO_MATCH_EMPTY, which yields 1 if the pattern can match
322 an empty string. If it can, pcretest shows this in its information output.
323
324 9. Fixed two related bugs that applied to Unicode extended grapheme clusters
325 that were repeated with a maximizing qualifier (e.g. \X* or \X{2,5}) when
326 matched by pcre_exec() without using JIT:
327
328 (a) If the rest of the pattern did not match after a maximal run of
329 grapheme clusters, the code for backing up to try with fewer of them
330 did not always back up over a full grapheme when characters that do not
331 have the modifier quality were involved, e.g. Hangul syllables.
332
333 (b) If the match point in a subject started with modifier character, and
334 there was no match, the code could incorrectly back up beyond the match
335 point, and potentially beyond the first character in the subject,
336 leading to a segfault or an incorrect match result.
337
338 10. A conditional group with an assertion condition could lead to PCRE
339 recording an incorrect first data item for a match if no other first data
340 item was recorded. For example, the pattern (?(?=ab)ab) recorded "a" as a
341 first data item, and therefore matched "ca" after "c" instead of at the
342 start.
343
344 11. Change 40 for 8.33 (allowing pcregrep to find empty strings) showed up a
345 bug that caused the command "echo a | ./pcregrep -M '|a'" to loop.
346
347 12. The source of pcregrep now includes z/OS-specific code so that it can be
348 compiled for z/OS as part of the special z/OS distribution.
349
350 13. Added the -T and -TM options to pcretest.
351
352 14. The code in pcre_compile.c for creating the table of named capturing groups
353 has been refactored. Instead of creating the table dynamically during the
354 actual compiling pass, the information is remembered during the pre-compile
355 pass (on the stack unless there are more than 20 named groups, in which
356 case malloc() is used) and the whole table is created before the actual
357 compile happens. This has simplified the code (it is now nearly 150 lines
358 shorter) and prepared the way for better handling of references to groups
359 with duplicate names.
360
361 15. A back reference to a named subpattern when there is more than one of the
362 same name now checks them in the order in which they appear in the pattern.
363 The first one that is set is used for the reference. Previously only the
364 first one was inspected. This change makes PCRE more compatible with Perl.
365
366 16. Unicode character properties were updated from Unicode 6.3.0.
367
368 17. The compile-time code for auto-possessification has been refactored, based
369 on a patch by Zoltan Herczeg. It now happens after instead of during
370 compilation. The code is cleaner, and more cases are handled. The option
371 PCRE_NO_AUTO_POSSESS is added for testing purposes, and the -O and /O
372 options in pcretest are provided to set it. It can also be set by
373 (*NO_AUTO_POSSESS) at the start of a pattern.
374
375 18. The character VT has been added to the default ("C" locale) set of
376 characters that match \s and are generally treated as white space,
377 following this same change in Perl 5.18. There is now no difference between
378 "Perl space" and "POSIX space". Whether VT is treated as white space in
379 other locales depends on the locale.
380
381 19. The code for checking named groups as conditions, either for being set or
382 for being recursed, has been refactored (this is related to 14 and 15
383 above). Processing unduplicated named groups should now be as fast at
384 numerical groups, and processing duplicated groups should be faster than
385 before.
386
387 20. Two patches to the CMake build system, by Alexander Barkov:
388
389 (1) Replace the "source" command by "." in CMakeLists.txt because
390 "source" is a bash-ism.
391
392 (2) Add missing HAVE_STDINT_H and HAVE_INTTYPES_H to config-cmake.h.in;
393 without these the CMake build does not work on Solaris.
394
395 21. Perl has changed its handling of \8 and \9. If there is no previously
396 encountered capturing group of those numbers, they are treated as the
397 literal characters 8 and 9 instead of a binary zero followed by the
398 literals. PCRE now does the same.
399
400 22. Following Perl, added \o{} to specify codepoints in octal, making it
401 possible to specify values greater than 0777 and also making them
402 unambiguous.
403
404 23. Perl now gives an error for missing closing braces after \x{... instead of
405 treating the string as literal. PCRE now does the same.
406
407 24. RunTest used to grumble if an inappropriate test was selected explicitly,
408 but just skip it when running all tests. This make it awkward to run ranges
409 of tests when one of them was inappropriate. Now it just skips any
410 inappropriate tests, as it always did when running all tests.
411
412 25. If PCRE_AUTO_CALLOUT and PCRE_UCP were set for a pattern that contained
413 character types such as \d or \w, too many callouts were inserted, and the
414 data that they returned was rubbish.
415
416 26. In UCP mode, \s was not matching two of the characters that Perl matches,
417 namely NEL (U+0085) and MONGOLIAN VOWEL SEPARATOR (U+180E), though they
418 were matched by \h. The code has now been refactored so that the lists of
419 the horizontal and vertical whitespace characters used for \h and \v (which
420 are defined only in one place) are now also used for \s.
421
422 27. Add JIT support for the 64 bit TileGX architecture.
423 Patch by Jiong Wang (Tilera Corporation).
424
425 28. Possessive quantifiers for classes (both explicit and automatically
426 generated) now use special opcodes instead of wrapping in ONCE brackets.
427
428 29. Whereas an item such as A{4}+ ignored the possessivenes of the quantifier
429 (because it's meaningless), this was not happening when PCRE_CASELESS was
430 set. Not wrong, but inefficient.
431
432 30. Updated perltest.pl to add /u (force Unicode mode) when /W (use Unicode
433 properties for \w, \d, etc) is present in a test regex. Otherwise if the
434 test contains no characters greater than 255, Perl doesn't realise it
435 should be using Unicode semantics.
436
437 31. Upgraded the handling of the POSIX classes [:graph:], [:print:], and
438 [:punct:] when PCRE_UCP is set so as to include the same characters as Perl
439 does in Unicode mode.
440
441 32. Added the "forbid" facility to pcretest so that putting tests into the
442 wrong test files can sometimes be quickly detected.
443
444 33. There is now a limit (default 250) on the depth of nesting of parentheses.
445 This limit is imposed to control the amount of system stack used at compile
446 time. It can be changed at build time by --with-parens-nest-limit=xxx or
447 the equivalent in CMake.
448
449 34. Character classes such as [A-\d] or [a-[:digit:]] now cause compile-time
450 errors. Perl warns for these when in warning mode, but PCRE has no facility
451 for giving warnings.
452
453 35. Change 34 for 8.13 allowed quantifiers on assertions, because Perl does.
454 However, this was not working for (?!) because it is optimized to (*FAIL),
455 for which PCRE does not allow quantifiers. The optimization is now disabled
456 when a quantifier follows (?!). I can't see any use for this, but it makes
457 things uniform.
458
459 36. Perl no longer allows group names to start with digits, so I have made this
460 change also in PCRE. It simplifies the code a bit.
461
462 37. In extended mode, Perl ignores spaces before a + that indicates a
463 possessive quantifier. PCRE allowed a space before the quantifier, but not
464 before the possessive +. It now does.
465
466 38. The use of \K (reset reported match start) within a repeated possessive
467 group such as (a\Kb)*+ was not working.
468
469 40. Document that the same character tables must be used at compile time and
470 run time, and that the facility to pass tables to pcre_exec() and
471 pcre_dfa_exec() is for use only with saved/restored patterns.
472
473 41. Applied Jeff Trawick's patch CMakeLists.txt, which "provides two new
474 features for Builds with MSVC:
475
476 1. Support pcre.rc and/or pcreposix.rc (as is already done for MinGW
477 builds). The .rc files can be used to set FileDescription and many other
478 attributes.
479
480 2. Add an option (-DINSTALL_MSVC_PDB) to enable installation of .pdb files.
481 This allows higher-level build scripts which want .pdb files to avoid
482 hard-coding the exact files needed."
483
484 42. Added support for [[:<:]] and [[:>:]] as used in the BSD POSIX library to
485 mean "start of word" and "end of word", respectively, as a transition aid.
486
487 43. A minimizing repeat of a class containing codepoints greater than 255 in
488 non-UTF 16-bit or 32-bit modes caused an internal error when PCRE was
489 compiled to use the heap for recursion.
490
491 44. Got rid of some compiler warnings for unused variables when UTF but not UCP
492 is configured.
493
494
495 Version 8.33 28-May-2013
496 ------------------------
497
498 1. Added 'U' to some constants that are compared to unsigned integers, to
499 avoid compiler signed/unsigned warnings. Added (int) casts to unsigned
500 variables that are added to signed variables, to ensure the result is
501 signed and can be negated.
502
503 2. Applied patch by Daniel Richard G for quashing MSVC warnings to the
504 CMake config files.
505
506 3. Revise the creation of config.h.generic so that all boolean macros are
507 #undefined, whereas non-boolean macros are #ifndef/#endif-ed. This makes
508 overriding via -D on the command line possible.
509
510 4. Changing the definition of the variable "op" in pcre_exec.c from pcre_uchar
511 to unsigned int is reported to make a quite noticeable speed difference in
512 a specific Windows environment. Testing on Linux did also appear to show
513 some benefit (and it is clearly not harmful). Also fixed the definition of
514 Xop which should be unsigned.
515
516 5. Related to (4), changing the definition of the intermediate variable cc
517 in repeated character loops from pcre_uchar to pcre_uint32 also gave speed
518 improvements.
519
520 6. Fix forward search in JIT when link size is 3 or greater. Also removed some
521 unnecessary spaces.
522
523 7. Adjust autogen.sh and configure.ac to lose warnings given by automake 1.12
524 and later.
525
526 8. Fix two buffer over read issues in 16 and 32 bit modes. Affects JIT only.
527
528 9. Optimizing fast_forward_start_bits in JIT.
529
530 10. Adding support for callouts in JIT, and fixing some issues revealed
531 during this work. Namely:
532
533 (a) Unoptimized capturing brackets incorrectly reset on backtrack.
534
535 (b) Minimum length was not checked before the matching is started.
536
537 11. The value of capture_last that is passed to callouts was incorrect in some
538 cases when there was a capture on one path that was subsequently abandoned
539 after a backtrack. Also, the capture_last value is now reset after a
540 recursion, since all captures are also reset in this case.
541
542 12. The interpreter no longer returns the "too many substrings" error in the
543 case when an overflowing capture is in a branch that is subsequently
544 abandoned after a backtrack.
545
546 13. In the pathological case when an offset vector of size 2 is used, pcretest
547 now prints out the matched string after a yield of 0 or 1.
548
549 14. Inlining subpatterns in recursions, when certain conditions are fulfilled.
550 Only supported by the JIT compiler at the moment.
551
552 15. JIT compiler now supports 32 bit Macs thanks to Lawrence Velazquez.
553
554 16. Partial matches now set offsets[2] to the "bumpalong" value, that is, the
555 offset of the starting point of the matching process, provided the offsets
556 vector is large enough.
557
558 17. The \A escape now records a lookbehind value of 1, though its execution
559 does not actually inspect the previous character. This is to ensure that,
560 in partial multi-segment matching, at least one character from the old
561 segment is retained when a new segment is processed. Otherwise, if there
562 are no lookbehinds in the pattern, \A might match incorrectly at the start
563 of a new segment.
564
565 18. Added some #ifdef __VMS code into pcretest.c to help VMS implementations.
566
567 19. Redefined some pcre_uchar variables in pcre_exec.c as pcre_uint32; this
568 gives some modest performance improvement in 8-bit mode.
569
570 20. Added the PCRE-specific property \p{Xuc} for matching characters that can
571 be expressed in certain programming languages using Universal Character
572 Names.
573
574 21. Unicode validation has been updated in the light of Unicode Corrigendum #9,
575 which points out that "non characters" are not "characters that may not
576 appear in Unicode strings" but rather "characters that are reserved for
577 internal use and have only local meaning".
578
579 22. When a pattern was compiled with automatic callouts (PCRE_AUTO_CALLOUT) and
580 there was a conditional group that depended on an assertion, if the
581 assertion was false, the callout that immediately followed the alternation
582 in the condition was skipped when pcre_exec() was used for matching.
583
584 23. Allow an explicit callout to be inserted before an assertion that is the
585 condition for a conditional group, for compatibility with automatic
586 callouts, which always insert a callout at this point.
587
588 24. In 8.31, (*COMMIT) was confined to within a recursive subpattern. Perl also
589 confines (*SKIP) and (*PRUNE) in the same way, and this has now been done.
590
591 25. (*PRUNE) is now supported by the JIT compiler.
592
593 26. Fix infinite loop when /(?<=(*SKIP)ac)a/ is matched against aa.
594
595 27. Fix the case where there are two or more SKIPs with arguments that may be
596 ignored.
597
598 28. (*SKIP) is now supported by the JIT compiler.
599
600 29. (*THEN) is now supported by the JIT compiler.
601
602 30. Update RunTest with additional test selector options.
603
604 31. The way PCRE handles backtracking verbs has been changed in two ways.
605
606 (1) Previously, in something like (*COMMIT)(*SKIP), COMMIT would override
607 SKIP. Now, PCRE acts on whichever backtracking verb is reached first by
608 backtracking. In some cases this makes it more Perl-compatible, but Perl's
609 rather obscure rules do not always do the same thing.
610
611 (2) Previously, backtracking verbs were confined within assertions. This is
612 no longer the case for positive assertions, except for (*ACCEPT). Again,
613 this sometimes improves Perl compatibility, and sometimes does not.
614
615 32. A number of tests that were in test 2 because Perl did things differently
616 have been moved to test 1, because either Perl or PCRE has changed, and
617 these tests are now compatible.
618
619 32. Backtracking control verbs are now handled in the same way in JIT and
620 interpreter.
621
622 33. An opening parenthesis in a MARK/PRUNE/SKIP/THEN name in a pattern that
623 contained a forward subroutine reference caused a compile error.
624
625 34. Auto-detect and optimize limited repetitions in JIT.
626
627 35. Implement PCRE_NEVER_UTF to lock out the use of UTF, in particular,
628 blocking (*UTF) etc.
629
630 36. In the interpreter, maximizing pattern repetitions for characters and
631 character types now use tail recursion, which reduces stack usage.
632
633 37. The value of the max lookbehind was not correctly preserved if a compiled
634 and saved regex was reloaded on a host of different endianness.
635
636 38. Implemented (*LIMIT_MATCH) and (*LIMIT_RECURSION). As part of the extension
637 of the compiled pattern block, expand the flags field from 16 to 32 bits
638 because it was almost full.
639
640 39. Try madvise first before posix_madvise.
641
642 40. Change 7 for PCRE 7.9 made it impossible for pcregrep to find empty lines
643 with a pattern such as ^$. It has taken 4 years for anybody to notice! The
644 original change locked out all matches of empty strings. This has been
645 changed so that one match of an empty string per line is recognized.
646 Subsequent searches on the same line (for colouring or for --only-matching,
647 for example) do not recognize empty strings.
648
649 41. Applied a user patch to fix a number of spelling mistakes in comments.
650
651 42. Data lines longer than 65536 caused pcretest to crash.
652
653 43. Clarified the data type for length and startoffset arguments for pcre_exec
654 and pcre_dfa_exec in the function-specific man pages, where they were
655 explicitly stated to be in bytes, never having been updated. I also added
656 some clarification to the pcreapi man page.
657
658 44. A call to pcre_dfa_exec() with an output vector size less than 2 caused
659 a segmentation fault.
660
661
662 Version 8.32 30-November-2012
663 -----------------------------
664
665 1. Improved JIT compiler optimizations for first character search and single
666 character iterators.
667
668 2. Supporting IBM XL C compilers for PPC architectures in the JIT compiler.
669 Patch by Daniel Richard G.
670
671 3. Single character iterator optimizations in the JIT compiler.
672
673 4. Improved JIT compiler optimizations for character ranges.
674
675 5. Rename the "leave" variable names to "quit" to improve WinCE compatibility.
676 Reported by Giuseppe D'Angelo.
677
678 6. The PCRE_STARTLINE bit, indicating that a match can occur only at the start
679 of a line, was being set incorrectly in cases where .* appeared inside
680 atomic brackets at the start of a pattern, or where there was a subsequent
681 *PRUNE or *SKIP.
682
683 7. Improved instruction cache flush for POWER/PowerPC.
684 Patch by Daniel Richard G.
685
686 8. Fixed a number of issues in pcregrep, making it more compatible with GNU
687 grep:
688
689 (a) There is now no limit to the number of patterns to be matched.
690
691 (b) An error is given if a pattern is too long.
692
693 (c) Multiple uses of --exclude, --exclude-dir, --include, and --include-dir
694 are now supported.
695
696 (d) --exclude-from and --include-from (multiple use) have been added.
697
698 (e) Exclusions and inclusions now apply to all files and directories, not
699 just to those obtained from scanning a directory recursively.
700
701 (f) Multiple uses of -f and --file-list are now supported.
702
703 (g) In a Windows environment, the default for -d has been changed from
704 "read" (the GNU grep default) to "skip", because otherwise the presence
705 of a directory in the file list provokes an error.
706
707 (h) The documentation has been revised and clarified in places.
708
709 9. Improve the matching speed of capturing brackets.
710
711 10. Changed the meaning of \X so that it now matches a Unicode extended
712 grapheme cluster.
713
714 11. Patch by Daniel Richard G to the autoconf files to add a macro for sorting
715 out POSIX threads when JIT support is configured.
716
717 12. Added support for PCRE_STUDY_EXTRA_NEEDED.
718
719 13. In the POSIX wrapper regcomp() function, setting re_nsub field in the preg
720 structure could go wrong in environments where size_t is not the same size
721 as int.
722
723 14. Applied user-supplied patch to pcrecpp.cc to allow PCRE_NO_UTF8_CHECK to be
724 set.
725
726 15. The EBCDIC support had decayed; later updates to the code had included
727 explicit references to (e.g.) \x0a instead of CHAR_LF. There has been a
728 general tidy up of EBCDIC-related issues, and the documentation was also
729 not quite right. There is now a test that can be run on ASCII systems to
730 check some of the EBCDIC-related things (but is it not a full test).
731
732 16. The new PCRE_STUDY_EXTRA_NEEDED option is now used by pcregrep, resulting
733 in a small tidy to the code.
734
735 17. Fix JIT tests when UTF is disabled and both 8 and 16 bit mode are enabled.
736
737 18. If the --only-matching (-o) option in pcregrep is specified multiple
738 times, each one causes appropriate output. For example, -o1 -o2 outputs the
739 substrings matched by the 1st and 2nd capturing parentheses. A separating
740 string can be specified by --om-separator (default empty).
741
742 19. Improving the first n character searches.
743
744 20. Turn case lists for horizontal and vertical white space into macros so that
745 they are defined only once.
746
747 21. This set of changes together give more compatible Unicode case-folding
748 behaviour for characters that have more than one other case when UCP
749 support is available.
750
751 (a) The Unicode property table now has offsets into a new table of sets of
752 three or more characters that are case-equivalent. The MultiStage2.py
753 script that generates these tables (the pcre_ucd.c file) now scans
754 CaseFolding.txt instead of UnicodeData.txt for character case
755 information.
756
757 (b) The code for adding characters or ranges of characters to a character
758 class has been abstracted into a generalized function that also handles
759 case-independence. In UTF-mode with UCP support, this uses the new data
760 to handle characters with more than one other case.
761
762 (c) A bug that is fixed as a result of (b) is that codepoints less than 256
763 whose other case is greater than 256 are now correctly matched
764 caselessly. Previously, the high codepoint matched the low one, but not
765 vice versa.
766
767 (d) The processing of \h, \H, \v, and \ in character classes now makes use
768 of the new class addition function, using character lists defined as
769 macros alongside the case definitions of 20 above.
770
771 (e) Caseless back references now work with characters that have more than
772 one other case.
773
774 (f) General caseless matching of characters with more than one other case
775 is supported.
776
777 22. Unicode character properties were updated from Unicode 6.2.0
778
779 23. Improved CMake support under Windows. Patch by Daniel Richard G.
780
781 24. Add support for 32-bit character strings, and UTF-32
782
783 25. Major JIT compiler update (code refactoring and bugfixing).
784 Experimental Sparc 32 support is added.
785
786 26. Applied a modified version of Daniel Richard G's patch to create
787 pcre.h.generic and config.h.generic by "make" instead of in the
788 PrepareRelease script.
789
790 27. Added a definition for CHAR_NULL (helpful for the z/OS port), and use it in
791 pcre_compile.c when checking for a zero character.
792
793 28. Introducing a native interface for JIT. Through this interface, the compiled
794 machine code can be directly executed. The purpose of this interface is to
795 provide fast pattern matching, so several sanity checks are not performed.
796 However, feature tests are still performed. The new interface provides
797 1.4x speedup compared to the old one.
798
799 29. If pcre_exec() or pcre_dfa_exec() was called with a negative value for
800 the subject string length, the error given was PCRE_ERROR_BADOFFSET, which
801 was confusing. There is now a new error PCRE_ERROR_BADLENGTH for this case.
802
803 30. In 8-bit UTF-8 mode, pcretest failed to give an error for data codepoints
804 greater than 0x7fffffff (which cannot be represented in UTF-8, even under
805 the "old" RFC 2279). Instead, it ended up passing a negative length to
806 pcre_exec().
807
808 31. Add support for GCC's visibility feature to hide internal functions.
809
810 32. Running "pcretest -C pcre8" or "pcretest -C pcre16" gave a spurious error
811 "unknown -C option" after outputting 0 or 1.
812
813 33. There is now support for generating a code coverage report for the test
814 suite in environments where gcc is the compiler and lcov is installed. This
815 is mainly for the benefit of the developers.
816
817 34. If PCRE is built with --enable-valgrind, certain memory regions are marked
818 unaddressable using valgrind annotations, allowing valgrind to detect
819 invalid memory accesses. This is mainly for the benefit of the developers.
820
821 25. (*UTF) can now be used to start a pattern in any of the three libraries.
822
823 26. Give configure error if --enable-cpp but no C++ compiler found.
824
825
826 Version 8.31 06-July-2012
827 -------------------------
828
829 1. Fixing a wrong JIT test case and some compiler warnings.
830
831 2. Removed a bashism from the RunTest script.
832
833 3. Add a cast to pcre_exec.c to fix the warning "unary minus operator applied
834 to unsigned type, result still unsigned" that was given by an MS compiler
835 on encountering the code "-sizeof(xxx)".
836
837 4. Partial matching support is added to the JIT compiler.
838
839 5. Fixed several bugs concerned with partial matching of items that consist
840 of more than one character:
841
842 (a) /^(..)\1/ did not partially match "aba" because checking references was
843 done on an "all or nothing" basis. This also applied to repeated
844 references.
845
846 (b) \R did not give a hard partial match if \r was found at the end of the
847 subject.
848
849 (c) \X did not give a hard partial match after matching one or more
850 characters at the end of the subject.
851
852 (d) When newline was set to CRLF, a pattern such as /a$/ did not recognize
853 a partial match for the string "\r".
854
855 (e) When newline was set to CRLF, the metacharacter "." did not recognize
856 a partial match for a CR character at the end of the subject string.
857
858 6. If JIT is requested using /S++ or -s++ (instead of just /S+ or -s+) when
859 running pcretest, the text "(JIT)" added to the output whenever JIT is
860 actually used to run the match.
861
862 7. Individual JIT compile options can be set in pcretest by following -s+[+]
863 or /S+[+] with a digit between 1 and 7.
864
865 8. OP_NOT now supports any UTF character not just single-byte ones.
866
867 9. (*MARK) control verb is now supported by the JIT compiler.
868
869 10. The command "./RunTest list" lists the available tests without actually
870 running any of them. (Because I keep forgetting what they all are.)
871
872 11. Add PCRE_INFO_MAXLOOKBEHIND.
873
874 12. Applied a (slightly modified) user-supplied patch that improves performance
875 when the heap is used for recursion (compiled with --disable-stack-for-
876 recursion). Instead of malloc and free for each heap frame each time a
877 logical recursion happens, frames are retained on a chain and re-used where
878 possible. This sometimes gives as much as 30% improvement.
879
880 13. As documented, (*COMMIT) is now confined to within a recursive subpattern
881 call.
882
883 14. As documented, (*COMMIT) is now confined to within a positive assertion.
884
885 15. It is now possible to link pcretest with libedit as an alternative to
886 libreadline.
887
888 16. (*COMMIT) control verb is now supported by the JIT compiler.
889
890 17. The Unicode data tables have been updated to Unicode 6.1.0.
891
892 18. Added --file-list option to pcregrep.
893
894 19. Added binary file support to pcregrep, including the -a, --binary-files,
895 -I, and --text options.
896
897 20. The madvise function is renamed for posix_madvise for QNX compatibility
898 reasons. Fixed by Giuseppe D'Angelo.
899
900 21. Fixed a bug for backward assertions with REVERSE 0 in the JIT compiler.
901
902 22. Changed the option for creating symbolic links for 16-bit man pages from
903 -s to -sf so that re-installing does not cause issues.
904
905 23. Support PCRE_NO_START_OPTIMIZE in JIT as (*MARK) support requires it.
906
907 24. Fixed a very old bug in pcretest that caused errors with restarted DFA
908 matches in certain environments (the workspace was not being correctly
909 retained). Also added to pcre_dfa_exec() a simple plausibility check on
910 some of the workspace data at the beginning of a restart.
911
912 25. \s*\R was auto-possessifying the \s* when it should not, whereas \S*\R
913 was not doing so when it should - probably a typo introduced by SVN 528
914 (change 8.10/14).
915
916 26. When PCRE_UCP was not set, \w+\x{c4} was incorrectly auto-possessifying the
917 \w+ when the character tables indicated that \x{c4} was a word character.
918 There were several related cases, all because the tests for doing a table
919 lookup were testing for characters less than 127 instead of 255.
920
921 27. If a pattern contains capturing parentheses that are not used in a match,
922 their slots in the ovector are set to -1. For those that are higher than
923 any matched groups, this happens at the end of processing. In the case when
924 there were back references that the ovector was too small to contain
925 (causing temporary malloc'd memory to be used during matching), and the
926 highest capturing number was not used, memory off the end of the ovector
927 was incorrectly being set to -1. (It was using the size of the temporary
928 memory instead of the true size.)
929
930 28. To catch bugs like 27 using valgrind, when pcretest is asked to specify an
931 ovector size, it uses memory at the end of the block that it has got.
932
933 29. Check for an overlong MARK name and give an error at compile time. The
934 limit is 255 for the 8-bit library and 65535 for the 16-bit library.
935
936 30. JIT compiler update.
937
938 31. JIT is now supported on jailbroken iOS devices. Thanks for Ruiger
939 Rill for the patch.
940
941 32. Put spaces around SLJIT_PRINT_D in the JIT compiler. Required by CXX11.
942
943 33. Variable renamings in the PCRE-JIT compiler. No functionality change.
944
945 34. Fixed typos in pcregrep: in two places there was SUPPORT_LIBZ2 instead of
946 SUPPORT_LIBBZ2. This caused a build problem when bzip2 but not gzip (zlib)
947 was enabled.
948
949 35. Improve JIT code generation for greedy plus quantifier.
950
951 36. When /((?:a?)*)*c/ or /((?>a?)*)*c/ was matched against "aac", it set group
952 1 to "aa" instead of to an empty string. The bug affected repeated groups
953 that could potentially match an empty string.
954
955 37. Optimizing single character iterators in JIT.
956
957 38. Wide characters specified with \uxxxx in JavaScript mode are now subject to
958 the same checks as \x{...} characters in non-JavaScript mode. Specifically,
959 codepoints that are too big for the mode are faulted, and in a UTF mode,
960 disallowed codepoints are also faulted.
961
962 39. If PCRE was compiled with UTF support, in three places in the DFA
963 matcher there was code that should only have been obeyed in UTF mode, but
964 was being obeyed unconditionally. In 8-bit mode this could cause incorrect
965 processing when bytes with values greater than 127 were present. In 16-bit
966 mode the bug would be provoked by values in the range 0xfc00 to 0xdc00. In
967 both cases the values are those that cannot be the first data item in a UTF
968 character. The three items that might have provoked this were recursions,
969 possessively repeated groups, and atomic groups.
970
971 40. Ensure that libpcre is explicitly listed in the link commands for pcretest
972 and pcregrep, because some OS require shared objects to be explicitly
973 passed to ld, causing the link step to fail if they are not.
974
975 41. There were two incorrect #ifdefs in pcre_study.c, meaning that, in 16-bit
976 mode, patterns that started with \h* or \R* might be incorrectly matched.
977
978
979 Version 8.30 04-February-2012
980 -----------------------------
981
982 1. Renamed "isnumber" as "is_a_number" because in some Mac environments this
983 name is defined in ctype.h.
984
985 2. Fixed a bug in fixed-length calculation for lookbehinds that would show up
986 only in quite long subpatterns.
987
988 3. Removed the function pcre_info(), which has been obsolete and deprecated
989 since it was replaced by pcre_fullinfo() in February 2000.
990
991 4. For a non-anchored pattern, if (*SKIP) was given with a name that did not
992 match a (*MARK), and the match failed at the start of the subject, a
993 reference to memory before the start of the subject could occur. This bug
994 was introduced by fix 17 of release 8.21.
995
996 5. A reference to an unset group with zero minimum repetition was giving
997 totally wrong answers (in non-JavaScript-compatibility mode). For example,
998 /(another)?(\1?)test/ matched against "hello world test". This bug was
999 introduced in release 8.13.
1000
1001 6. Add support for 16-bit character strings (a large amount of work involving
1002 many changes and refactorings).
1003
1004 7. RunGrepTest failed on msys because \r\n was replaced by whitespace when the
1005 command "pattern=`printf 'xxx\r\njkl'`" was run. The pattern is now taken
1006 from a file.
1007
1008 8. Ovector size of 2 is also supported by JIT based pcre_exec (the ovector size
1009 rounding is not applied in this particular case).
1010
1011 9. The invalid Unicode surrogate codepoints U+D800 to U+DFFF are now rejected
1012 if they appear, or are escaped, in patterns.
1013
1014 10. Get rid of a number of -Wunused-but-set-variable warnings.
1015
1016 11. The pattern /(?=(*:x))(q|)/ matches an empty string, and returns the mark
1017 "x". The similar pattern /(?=(*:x))((*:y)q|)/ did not return a mark at all.
1018 Oddly, Perl behaves the same way. PCRE has been fixed so that this pattern
1019 also returns the mark "x". This bug applied to capturing parentheses,
1020 non-capturing parentheses, and atomic parentheses. It also applied to some
1021 assertions.
1022
1023 12. Stephen Kelly's patch to CMakeLists.txt allows it to parse the version
1024 information out of configure.ac instead of relying on pcre.h.generic, which
1025 is not stored in the repository.
1026
1027 13. Applied Dmitry V. Levin's patch for a more portable method for linking with
1028 -lreadline.
1029
1030 14. ZH added PCRE_CONFIG_JITTARGET; added its output to pcretest -C.
1031
1032 15. Applied Graycode's patch to put the top-level frame on the stack rather
1033 than the heap when not using the stack for recursion. This gives a
1034 performance improvement in many cases when recursion is not deep.
1035
1036 16. Experimental code added to "pcretest -C" to output the stack frame size.
1037
1038
1039 Version 8.21 12-Dec-2011
1040 ------------------------
1041
1042 1. Updating the JIT compiler.
1043
1044 2. JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
1045 are added as well.
1046
1047 3. Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
1048 PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
1049 calling _pcre_jit_exec. Some extra comments are added.
1050
1051 4. (*MARK) settings inside atomic groups that do not contain any capturing
1052 parentheses, for example, (?>a(*:m)), were not being passed out. This bug
1053 was introduced by change 18 for 8.20.
1054
1055 5. Supporting of \x, \U and \u in JavaScript compatibility mode based on the
1056 ECMA-262 standard.
1057
1058 6. Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
1059 erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
1060 This bug was probably introduced by change 9 of 8.13.
1061
1062 7. While fixing 6 above, I noticed that a number of other items were being
1063 incorrectly rejected as "not fixed length". This arose partly because newer
1064 opcodes had not been added to the fixed-length checking code. I have (a)
1065 corrected the bug and added tests for these items, and (b) arranged for an
1066 error to occur if an unknown opcode is encountered while checking for fixed
1067 length instead of just assuming "not fixed length". The items that were
1068 rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
1069 (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
1070 repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
1071
1072 8. A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
1073 being incorrectly compiled and would have given unpredicatble results.
1074
1075 9. A possessively repeated subpattern with minimum repeat count greater than
1076 one behaved incorrectly. For example, (A){2,}+ behaved as if it was
1077 (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
1078 the first (A) could occur when it should not.
1079
1080 10. Add a cast and remove a redundant test from the code.
1081
1082 11. JIT should use pcre_malloc/pcre_free for allocation.
1083
1084 12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
1085 best practice nowadays, and helps with cross-compiling. (If the exec_prefix
1086 is anything other than /usr, -L is still shown).
1087
1088 13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
1089
1090 14. Perl does not support \N without a following name in a [] class; PCRE now
1091 also gives an error.
1092
1093 15. If a forward reference was repeated with an upper limit of around 2000,
1094 it caused the error "internal error: overran compiling workspace". The
1095 maximum number of forward references (including repeats) was limited by the
1096 internal workspace, and dependent on the LINK_SIZE. The code has been
1097 rewritten so that the workspace expands (via pcre_malloc) if necessary, and
1098 the default depends on LINK_SIZE. There is a new upper limit (for safety)
1099 of around 200,000 forward references. While doing this, I also speeded up
1100 the filling in of repeated forward references.
1101
1102 16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
1103 incorrectly expecting the subject to contain another "a" after the start.
1104
1105 17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
1106 in the match, the SKIP should be ignored. This was not happening; instead
1107 the SKIP was being treated as NOMATCH. For patterns such as
1108 /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
1109 tested.
1110
1111 18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
1112 now much more compatible with Perl, in particular in cases where the result
1113 is a non-match for a non-anchored pattern. For example, if
1114 /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
1115 "m", where previously it did not return a name. A side effect of this
1116 change is that for partial matches, the last encountered mark name is
1117 returned, as for non matches. A number of tests that were previously not
1118 Perl-compatible have been moved into the Perl-compatible test files. The
1119 refactoring has had the pleasing side effect of removing one argument from
1120 the match() function, thus reducing its stack requirements.
1121
1122 19. If the /S+ option was used in pcretest to study a pattern using JIT,
1123 subsequent uses of /S (without +) incorrectly behaved like /S+.
1124
1125 21. Retrieve executable code size support for the JIT compiler and fixing
1126 some warnings.
1127
1128 22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
1129 not work when the shorter character appeared right at the end of the
1130 subject string.
1131
1132 23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
1133 systems.
1134
1135 24. Added PCRE_INFO_JITSIZE to pass on the value from (21) above, and also
1136 output it when the /M option is used in pcretest.
1137
1138 25. The CheckMan script was not being included in the distribution. Also, added
1139 an explicit "perl" to run Perl scripts from the PrepareRelease script
1140 because this is reportedly needed in Windows.
1141
1142 26. If study data was being save in a file and studying had not found a set of
1143 "starts with" bytes for the pattern, the data written to the file (though
1144 never used) was taken from uninitialized memory and so caused valgrind to
1145 complain.
1146
1147 27. Updated RunTest.bat as provided by Sheri Pierce.
1148
1149 28. Fixed a possible uninitialized memory bug in pcre_jit_compile.c.
1150
1151 29. Computation of memory usage for the table of capturing group names was
1152 giving an unnecessarily large value.
1153
1154
1155 Version 8.20 21-Oct-2011
1156 ------------------------
1157
1158 1. Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
1159 a POSIX class. After further experiments with Perl, which convinced me that
1160 Perl has bugs and confusions, a closing square bracket is no longer allowed
1161 in a POSIX name. This bug also affected patterns with classes that started
1162 with full stops.
1163
1164 2. If a pattern such as /(a)b|ac/ is matched against "ac", there is no
1165 captured substring, but while checking the failing first alternative,
1166 substring 1 is temporarily captured. If the output vector supplied to
1167 pcre_exec() was not big enough for this capture, the yield of the function
1168 was still zero ("insufficient space for captured substrings"). This cannot
1169 be totally fixed without adding another stack variable, which seems a lot
1170 of expense for a edge case. However, I have improved the situation in cases
1171 such as /(a)(b)x|abc/ matched against "abc", where the return code
1172 indicates that fewer than the maximum number of slots in the ovector have
1173 been set.
1174
1175 3. Related to (2) above: when there are more back references in a pattern than
1176 slots in the output vector, pcre_exec() uses temporary memory during
1177 matching, and copies in the captures as far as possible afterwards. It was
1178 using the entire output vector, but this conflicts with the specification
1179 that only 2/3 is used for passing back captured substrings. Now it uses
1180 only the first 2/3, for compatibility. This is, of course, another edge
1181 case.
1182
1183 4. Zoltan Herczeg's just-in-time compiler support has been integrated into the
1184 main code base, and can be used by building with --enable-jit. When this is
1185 done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
1186 runtime --no-jit option is given.
1187
1188 5. When the number of matches in a pcre_dfa_exec() run exactly filled the
1189 ovector, the return from the function was zero, implying that there were
1190 other matches that did not fit. The correct "exactly full" value is now
1191 returned.
1192
1193 6. If a subpattern that was called recursively or as a subroutine contained
1194 (*PRUNE) or any other control that caused it to give a non-standard return,
1195 invalid errors such as "Error -26 (nested recursion at the same subject
1196 position)" or even infinite loops could occur.
1197
1198 7. If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
1199 computing the minimum length on reaching *ACCEPT, and so ended up with the
1200 wrong value of 1 rather than 0. Further investigation indicates that
1201 computing a minimum subject length in the presence of *ACCEPT is difficult
1202 (think back references, subroutine calls), and so I have changed the code
1203 so that no minimum is registered for a pattern that contains *ACCEPT.
1204
1205 8. If (*THEN) was present in the first (true) branch of a conditional group,
1206 it was not handled as intended. [But see 16 below.]
1207
1208 9. Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
1209 Sheri Pierce.
1210
1211 10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
1212 the first byte in a match must be "a".
1213
1214 11. Change 17 for 8.13 increased the recursion depth for patterns like
1215 /a(?:.)*?a/ drastically. I've improved things by remembering whether a
1216 pattern contains any instances of (*THEN). If it does not, the old
1217 optimizations are restored. It would be nice to do this on a per-group
1218 basis, but at the moment that is not feasible.
1219
1220 12. In some environments, the output of pcretest -C is CRLF terminated. This
1221 broke RunTest's code that checks for the link size. A single white space
1222 character after the value is now allowed for.
1223
1224 13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
1225 For "fr", it uses the Windows-specific input and output files.
1226
1227 14. If (*THEN) appeared in a group that was called recursively or as a
1228 subroutine, it did not work as intended. [But see next item.]
1229
1230 15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
1231 pattern fragments (but not containing any | characters). If A and B are
1232 matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
1233 was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
1234 D. In other words, Perl considers parentheses that do not contain any |
1235 characters to be part of a surrounding alternative, whereas PCRE was
1236 treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
1237 differently. PCRE now behaves in the same way as Perl, except in the case
1238 of subroutine/recursion calls such as (?1) which have in any case always
1239 been different (but PCRE had them first :-).
1240
1241 16. Related to 15 above: Perl does not treat the | in a conditional group as
1242 creating alternatives. Such a group is treated in the same way as an
1243 ordinary group without any | characters when processing (*THEN). PCRE has
1244 been changed to match Perl's behaviour.
1245
1246 17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
1247 RunGrepTest script failed.
1248
1249 18. Change 22 for version 13 caused atomic groups to use more stack. This is
1250 inevitable for groups that contain captures, but it can lead to a lot of
1251 stack use in large patterns. The old behaviour has been restored for atomic
1252 groups that do not contain any capturing parentheses.
1253
1254 19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
1255 suppress the check for a minimum subject length at run time. (If it was
1256 given to pcre_exec() or pcre_dfa_exec() it did work.)
1257
1258 20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
1259 fail to work when decoding hex characters in data strings in EBCDIC
1260 environments.
1261
1262 21. It appears that in at least one Mac OS environment, the isxdigit() function
1263 is implemented as a macro that evaluates to its argument more than once,
1264 contravening the C 90 Standard (I haven't checked a later standard). There
1265 was an instance in pcretest which caused it to go wrong when processing
1266 \x{...} escapes in subject strings. The has been rewritten to avoid using
1267 things like p++ in the argument of isxdigit().
1268
1269
1270 Version 8.13 16-Aug-2011
1271 ------------------------
1272
1273 1. The Unicode data tables have been updated to Unicode 6.0.0.
1274
1275 2. Two minor typos in pcre_internal.h have been fixed.
1276
1277 3. Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
1278 pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
1279 in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
1280
1281 4. There were a number of related bugs in the code for matching backrefences
1282 caselessly in UTF-8 mode when codes for the characters concerned were
1283 different numbers of bytes. For example, U+023A and U+2C65 are an upper
1284 and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
1285 (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
1286 code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
1287 2-byte code at the end of the subject (it thought there wasn't enough data
1288 left).
1289
1290 5. Comprehensive information about what went wrong is now returned by
1291 pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
1292 as the output vector has at least 2 elements. The offset of the start of
1293 the failing character and a reason code are placed in the vector.
1294
1295 6. When the UTF-8 string check fails for pcre_compile(), the offset that is
1296 now returned is for the first byte of the failing character, instead of the
1297 last byte inspected. This is an incompatible change, but I hope it is small
1298 enough not to be a problem. It makes the returned offset consistent with
1299 pcre_exec() and pcre_dfa_exec().
1300
1301 7. pcretest now gives a text phrase as well as the error number when
1302 pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
1303 failure, the offset and reason code are output.
1304
1305 8. When \R was used with a maximizing quantifier it failed to skip backwards
1306 over a \r\n pair if the subsequent match failed. Instead, it just skipped
1307 back over a single character (\n). This seems wrong (because it treated the
1308 two characters as a single entity when going forwards), conflicts with the
1309 documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
1310 behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
1311 has been changed.
1312
1313 9. Some internal refactoring has changed the processing so that the handling
1314 of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
1315 time (the PCRE_DOTALL option was changed this way some time ago: version
1316 7.7 change 16). This has made it possible to abolish the OP_OPT op code,
1317 which was always a bit of a fudge. It also means that there is one less
1318 argument for the match() function, which reduces its stack requirements
1319 slightly. This change also fixes an incompatibility with Perl: the pattern
1320 (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
1321
1322 10. More internal refactoring has drastically reduced the number of recursive
1323 calls to match() for possessively repeated groups such as (abc)++ when
1324 using pcre_exec().
1325
1326 11. While implementing 10, a number of bugs in the handling of groups were
1327 discovered and fixed:
1328
1329 (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
1330 (a|)*(?1) gave a compile-time internal error.
1331 ((a|)+)+ did not notice that the outer group could match an empty string.
1332 (^a|^)+ was not marked as anchored.
1333 (.*a|.*)+ was not marked as matching at start or after a newline.
1334
1335 12. Yet more internal refactoring has removed another argument from the match()
1336 function. Special calls to this function are now indicated by setting a
1337 value in a variable in the "match data" data block.
1338
1339 13. Be more explicit in pcre_study() instead of relying on "default" for
1340 opcodes that mean there is no starting character; this means that when new
1341 ones are added and accidentally left out of pcre_study(), testing should
1342 pick them up.
1343
1344 14. The -s option of pcretest has been documented for ages as being an old
1345 synonym of -m (show memory usage). I have changed it to mean "force study
1346 for every regex", that is, assume /S for every regex. This is similar to -i
1347 and -d etc. It's slightly incompatible, but I'm hoping nobody is still
1348 using it. It makes it easier to run collections of tests with and without
1349 study enabled, and thereby test pcre_study() more easily. All the standard
1350 tests are now run with and without -s (but some patterns can be marked as
1351 "never study" - see 20 below).
1352
1353 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
1354 restoration of the capturing data to the outer values was not happening
1355 correctly.
1356
1357 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
1358 empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
1359 pattern had matched an empty string, and so incorrectly returned a no
1360 match.
1361
1362 17. There was optimizing code for the last branch of non-capturing parentheses,
1363 and also for the obeyed branch of a conditional subexpression, which used
1364 tail recursion to cut down on stack usage. Unfortunately, now that there is
1365 the possibility of (*THEN) occurring in these branches, tail recursion is
1366 no longer possible because the return has to be checked for (*THEN). These
1367 two optimizations have therefore been removed. [But see 8.20/11 above.]
1368
1369 18. If a pattern containing \R was studied, it was assumed that \R always
1370 matched two bytes, thus causing the minimum subject length to be
1371 incorrectly computed because \R can also match just one byte.
1372
1373 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
1374 was incorrectly computed.
1375
1376 20. If /S is present twice on a test pattern in pcretest input, it now
1377 *disables* studying, thereby overriding the use of -s on the command line
1378 (see 14 above). This is necessary for one or two tests to keep the output
1379 identical in both cases.
1380
1381 21. When (*ACCEPT) was used in an assertion that matched an empty string and
1382 PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
1383
1384 22. When an atomic group that contained a capturing parenthesis was
1385 successfully matched, but the branch in which it appeared failed, the
1386 capturing was not being forgotten if a higher numbered group was later
1387 captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
1388 group 1 to "a", when in fact it should be unset. This applied to multi-
1389 branched capturing and non-capturing groups, repeated or not, and also to
1390 positive assertions (capturing in negative assertions does not happen
1391 in PCRE) and also to nested atomic groups.
1392
1393 23. Add the ++ qualifier feature to pcretest, to show the remainder of the
1394 subject after a captured substring, to make it easier to tell which of a
1395 number of identical substrings has been captured.
1396
1397 24. The way atomic groups are processed by pcre_exec() has been changed so that
1398 if they are repeated, backtracking one repetition now resets captured
1399 values correctly. For example, if ((?>(a+)b)+aabab) is matched against
1400 "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
1401 "aaa". Previously, it would have been "a". As part of this code
1402 refactoring, the way recursive calls are handled has also been changed.
1403
1404 25. If an assertion condition captured any substrings, they were not passed
1405 back unless some other capturing happened later. For example, if
1406 (?(?=(a))a) was matched against "a", no capturing was returned.
1407
1408 26. When studying a pattern that contained subroutine calls or assertions,
1409 the code for finding the minimum length of a possible match was handling
1410 direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
1411 group 1 called group 2 while simultaneously a separate group 2 called group
1412 1). A stack overflow occurred in this case. I have fixed this by limiting
1413 the recursion depth to 10.
1414
1415 27. Updated RunTest.bat in the distribution to the version supplied by Tom
1416 Fortmann. This supports explicit test numbers on the command line, and has
1417 argument validation and error reporting.
1418
1419 28. An instance of \X with an unlimited repeat could fail if at any point the
1420 first character it looked at was a mark character.
1421
1422 29. Some minor code refactoring concerning Unicode properties and scripts
1423 should reduce the stack requirement of match() slightly.
1424
1425 30. Added the '=' option to pcretest to check the setting of unused capturing
1426 slots at the end of the pattern, which are documented as being -1, but are
1427 not included in the return count.
1428
1429 31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
1430 compiled something random. Now it gives a compile-time error (as does
1431 Perl).
1432
1433 32. A *MARK encountered during the processing of a positive assertion is now
1434 recorded and passed back (compatible with Perl).
1435
1436 33. If --only-matching or --colour was set on a pcregrep call whose pattern
1437 had alternative anchored branches, the search for a second match in a line
1438 was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
1439 matched the line "0102" twice. The same bug affected patterns that started
1440 with a backwards assertion. For example /\b01|\b02/ also matched "0102"
1441 twice.
1442
1443 34. Previously, PCRE did not allow quantification of assertions. However, Perl
1444 does, and because of capturing effects, quantifying parenthesized
1445 assertions may at times be useful. Quantifiers are now allowed for
1446 parenthesized assertions.
1447
1448 35. A minor code tidy in pcre_compile() when checking options for \R usage.
1449
1450 36. \g was being checked for fancy things in a character class, when it should
1451 just be a literal "g".
1452
1453 37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
1454 appearance of a nested POSIX class supersedes an apparent external class.
1455 For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
1456 unescaped square brackets may also appear as part of class names. For
1457 example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
1458 more like Perl. (But see 8.20/1 above.)
1459
1460 38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
1461 was because it thought it was \N{name}, which is not supported).
1462
1463 39. Add minix to OS list not supporting the -S option in pcretest.
1464
1465 40. PCRE tries to detect cases of infinite recursion at compile time, but it
1466 cannot analyze patterns in sufficient detail to catch mutual recursions
1467 such as ((?1))((?2)). There is now a runtime test that gives an error if a
1468 subgroup is called recursively as a subpattern for a second time at the
1469 same position in the subject string. In previous releases this might have
1470 been caught by the recursion limit, or it might have run out of stack.
1471
1472 41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
1473 happen only once. PCRE was, however incorrectly giving a compile time error
1474 "recursive call could loop indefinitely" because it cannot analyze the
1475 pattern in sufficient detail. The compile time test no longer happens when
1476 PCRE is compiling a conditional subpattern, but actual runaway loops are
1477 now caught at runtime (see 40 above).
1478
1479 42. It seems that Perl allows any characters other than a closing parenthesis
1480 to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
1481 has been changed to be the same.
1482
1483 43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
1484 as not to get warnings when autogen.sh is called. Also changed
1485 AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
1486
1487 44. To help people who use pcregrep to scan files containing exceedingly long
1488 lines, the following changes have been made:
1489
1490 (a) The default value of the buffer size parameter has been increased from
1491 8K to 20K. (The actual buffer used is three times this size.)
1492
1493 (b) The default can be changed by ./configure --with-pcregrep-bufsize when
1494 PCRE is built.
1495
1496 (c) A --buffer-size=n option has been added to pcregrep, to allow the size
1497 to be set at run time.
1498
1499 (d) Numerical values in pcregrep options can be followed by K or M, for
1500 example --buffer-size=50K.
1501
1502 (e) If a line being scanned overflows pcregrep's buffer, an error is now
1503 given and the return code is set to 2.
1504
1505 45. Add a pointer to the latest mark to the callout data block.
1506
1507 46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
1508 partial match of an empty string instead of no match. This was specific to
1509 the use of ".".
1510
1511 47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
1512 complete match instead of a partial match. This bug was dependent on both
1513 the PCRE_UTF8 and PCRE_DOTALL options being set.
1514
1515 48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
1516 starting byte set, because \b was not being ignored.
1517
1518
1519 Version 8.12 15-Jan-2011
1520 ------------------------
1521
1522 1. Fixed some typos in the markup of the man pages, and wrote a script that
1523 checks for such things as part of the documentation building process.
1524
1525 2. On a big-endian 64-bit system, pcregrep did not correctly process the
1526 --match-limit and --recursion-limit options (added for 8.11). In
1527 particular, this made one of the standard tests fail. (The integer value
1528 went into the wrong half of a long int.)
1529
1530 3. If the --colour option was given to pcregrep with -v (invert match), it
1531 did strange things, either producing crazy output, or crashing. It should,
1532 of course, ignore a request for colour when reporting lines that do not
1533 match.
1534
1535 4. Another pcregrep bug caused similar problems if --colour was specified with
1536 -M (multiline) and the pattern match finished with a line ending.
1537
1538 5. In pcregrep, when a pattern that ended with a literal newline sequence was
1539 matched in multiline mode, the following line was shown as part of the
1540 match. This seems wrong, so I have changed it.
1541
1542 6. Another pcregrep bug in multiline mode, when --colour was specified, caused
1543 the check for further matches in the same line (so they could be coloured)
1544 to overrun the end of the current line. If another match was found, it was
1545 incorrectly shown (and then shown again when found in the next line).
1546
1547 7. If pcregrep was compiled under Windows, there was a reference to the
1548 function pcregrep_exit() before it was defined. I am assuming this was
1549 the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
1550 reported by a user. I've moved the definition above the reference.
1551
1552
1553 Version 8.11 10-Dec-2010
1554 ------------------------
1555
1556 1. (*THEN) was not working properly if there were untried alternatives prior
1557 to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
1558 backtracked to try for "b" instead of moving to the next alternative branch
1559 at the same level (in this case, to look for "c"). The Perl documentation
1560 is clear that when (*THEN) is backtracked onto, it goes to the "next
1561 alternative in the innermost enclosing group".
1562
1563 2. (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
1564 such as (A(*COMMIT)B(*THEN)C|D) any failure after matching A should
1565 result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
1566 (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
1567 (*THEN).
1568
1569 3. If \s appeared in a character class, it removed the VT character from
1570 the class, even if it had been included by some previous item, for example
1571 in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
1572 of \s, but is part of the POSIX "space" class.)
1573
1574 4. A partial match never returns an empty string (because you can always
1575 match an empty string at the end of the subject); however the checking for
1576 an empty string was starting at the "start of match" point. This has been
1577 changed to the "earliest inspected character" point, because the returned
1578 data for a partial match starts at this character. This means that, for
1579 example, /(?<=abc)def/ gives a partial match for the subject "abc"
1580 (previously it gave "no match").
1581
1582 5. Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
1583 of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
1584 previously a full match would be given. However, setting PCRE_PARTIAL_HARD
1585 has an implication that the given string is incomplete (because a partial
1586 match is preferred over a full match). For this reason, these items now
1587 give a partial match in this situation. [Aside: previously, the one case
1588 /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
1589 match rather than a full match, which was wrong by the old rules, but is
1590 now correct.]
1591
1592 6. There was a bug in the handling of #-introduced comments, recognized when
1593 PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
1594 If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
1595 UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
1596 scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
1597 but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
1598 places in pcre_compile().
1599
1600 7. Related to (6) above, when pcre_compile() was skipping #-introduced
1601 comments when looking ahead for named forward references to subpatterns,
1602 the only newline sequence it recognized was NL. It now handles newlines
1603 according to the set newline convention.
1604
1605 8. SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
1606 former, but used strtoul(), whereas pcretest avoided strtoul() but did not
1607 cater for a lack of strerror(). These oversights have been fixed.
1608
1609 9. Added --match-limit and --recursion-limit to pcregrep.
1610
1611 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
1612
1613 11. When the -o option was used, pcregrep was setting a return code of 1, even
1614 when matches were found, and --line-buffered was not being honoured.
1615
1616 12. Added an optional parentheses number to the -o and --only-matching options
1617 of pcregrep.
1618
1619 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
1620 can match an empty string. The code to do it in pcretest and pcredemo
1621 needed fixing:
1622
1623 (a) When the newline convention was "crlf", pcretest got it wrong, skipping
1624 only one byte after an empty string match just before CRLF (this case
1625 just got forgotten; "any" and "anycrlf" were OK).
1626
1627 (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
1628 mode when an empty string match preceded an ASCII character followed by
1629 a non-ASCII character. (The code for advancing by one character rather
1630 than one byte was nonsense.)
1631
1632 (c) The pcredemo.c sample program did not have any code at all to handle
1633 the cases when CRLF is a valid newline sequence.
1634
1635 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
1636 as a starting offset was within the subject string. There is now a new
1637 error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
1638 negative or greater than the length of the string. In order to test this,
1639 pcretest is extended to allow the setting of negative starting offsets.
1640
1641 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
1642 starting offset points to the beginning of a UTF-8 character was
1643 unnecessarily clumsy. I tidied it up.
1644
1645 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
1646 bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
1647
1648 17. Nobody had reported that the --include_dir option, which was added in
1649 release 7.7 should have been called --include-dir (hyphen, not underscore)
1650 for compatibility with GNU grep. I have changed it to --include-dir, but
1651 left --include_dir as an undocumented synonym, and the same for
1652 --exclude-dir, though that is not available in GNU grep, at least as of
1653 release 2.5.4.
1654
1655 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
1656 characters from a string of bytes) have been redefined so as not to use
1657 loops, in order to improve performance in some environments. At the same
1658 time, I abstracted some of the common code into auxiliary macros to save
1659 repetition (this should not affect the compiled code).
1660
1661 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
1662 compile-time error is now given if \c is not followed by an ASCII
1663 character, that is, a byte less than 128. (In EBCDIC mode, the code is
1664 different, and any byte value is allowed.)
1665
1666 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
1667 START_OPTIMIZE option, which is now allowed at compile time - but just
1668 passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
1669 to pcregrep and other applications that have no direct access to PCRE
1670 options. The new /Y option in pcretest sets this option when calling
1671 pcre_compile().
1672
1673 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
1674 back references. Groups containing recursive back references were forced to
1675 be atomic by that change, but in the case of named groups, the amount of
1676 memory required was incorrectly computed, leading to "Failed: internal
1677 error: code overflow". This has been fixed.
1678
1679 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
1680 pcretest.c, to avoid build problems in some Borland environments.
1681
1682
1683 Version 8.10 25-Jun-2010
1684 ------------------------
1685
1686 1. Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
1687 THEN.
1688
1689 2. (*ACCEPT) was not working when inside an atomic group.
1690
1691 3. Inside a character class, \B is treated as a literal by default, but
1692 faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
1693 causes the error). The code is unchanged, but I tidied the documentation.
1694
1695 4. Inside a character class, PCRE always treated \R and \X as literals,
1696 whereas Perl faults them if its -w option is set. I have changed PCRE so
1697 that it faults them when PCRE_EXTRA is set.
1698
1699 5. Added support for \N, which always matches any character other than
1700 newline. (It is the same as "." when PCRE_DOTALL is not set.)
1701
1702 6. When compiling pcregrep with newer versions of gcc which may have
1703 FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
1704 declared with attribute warn_unused_result" were given. Just casting the
1705 result to (void) does not stop the warnings; a more elaborate fudge is
1706 needed. I've used a macro to implement this.
1707
1708 7. Minor change to pcretest.c to avoid a compiler warning.
1709
1710 8. Added four artifical Unicode properties to help with an option to make
1711 \s etc use properties (see next item). The new properties are: Xan
1712 (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
1713
1714 9. Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
1715 use Unicode properties. (*UCP) at the start of a pattern can be used to set
1716 this option. Modified pcretest to add /W to test this facility. Added
1717 REG_UCP to make it available via the POSIX interface.
1718
1719 10. Added --line-buffered to pcregrep.
1720
1721 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
1722 studied, and the match started with a letter with a code point greater than
1723 127 whose first byte was different to the first byte of the other case of
1724 the letter, the other case of this starting letter was not recognized
1725 (#976).
1726
1727 12. If a pattern that was studied started with a repeated Unicode property
1728 test, for example, \p{Nd}+, there was the theoretical possibility of
1729 setting up an incorrect bitmap of starting bytes, but fortunately it could
1730 not have actually happened in practice until change 8 above was made (it
1731 added property types that matched character-matching opcodes).
1732
1733 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
1734 possible starting bytes for non-anchored patterns.
1735
1736 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
1737 \R, and also a number of cases that involve Unicode properties, both
1738 explicit and implicit when PCRE_UCP is set.
1739
1740 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
1741 input, it could crash or give wrong results if characters with values
1742 greater than 0xc0 were present in the subject string. (Detail: it assumed
1743 UTF-8 input when processing these items.)
1744
1745 16. Added a lot of (int) casts to avoid compiler warnings in systems where
1746 size_t is 64-bit (#991).
1747
1748 17. Added a check for running out of memory when PCRE is compiled with
1749 --disable-stack-for-recursion (#990).
1750
1751 18. If the last data line in a file for pcretest does not have a newline on
1752 the end, a newline was missing in the output.
1753
1754 19. The default pcre_chartables.c file recognizes only ASCII characters (values
1755 less than 128) in its various bitmaps. However, there is a facility for
1756 generating tables according to the current locale when PCRE is compiled. It
1757 turns out that in some environments, 0x85 and 0xa0, which are Unicode space
1758 characters, are recognized by isspace() and therefore were getting set in
1759 these tables, and indeed these tables seem to approximate to ISO 8859. This
1760 caused a problem in UTF-8 mode when pcre_study() was used to create a list
1761 of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
1762 which of course cannot start UTF-8 characters. I have changed the code so
1763 that only real ASCII characters (less than 128) and the correct starting
1764 bytes for UTF-8 encodings are set for characters greater than 127 when in
1765 UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
1766 altogether.)
1767
1768 20. Added the /T option to pcretest so as to be able to run tests with non-
1769 standard character tables, thus making it possible to include the tests
1770 used for 19 above in the standard set of tests.
1771
1772 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
1773 reference to a subpattern the other side of a comment that contains an
1774 opening parenthesis caused either an internal compiling error, or a
1775 reference to the wrong subpattern.
1776
1777
1778 Version 8.02 19-Mar-2010
1779 ------------------------
1780
1781 1. The Unicode data tables have been updated to Unicode 5.2.0.
1782
1783 2. Added the option --libs-cpp to pcre-config, but only when C++ support is
1784 configured.
1785
1786 3. Updated the licensing terms in the pcregexp.pas file, as agreed with the
1787 original author of that file, following a query about its status.
1788
1789 4. On systems that do not have stdint.h (e.g. Solaris), check for and include
1790 inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
1791
1792 5. A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
1793 quantifier applied to a forward-referencing subroutine call, could compile
1794 incorrect code or give the error "internal error: previously-checked
1795 referenced subpattern not found".
1796
1797 6. Both MS Visual Studio and Symbian OS have problems with initializing
1798 variables to point to external functions. For these systems, therefore,
1799 pcre_malloc etc. are now initialized to local functions that call the
1800 relevant global functions.
1801
1802 7. There were two entries missing in the vectors called coptable and poptable
1803 in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
1804 I've fixed the data, and added a kludgy way of testing at compile time that
1805 the lengths are correct (equal to the number of opcodes).
1806
1807 8. Following on from 7, I added a similar kludge to check the length of the
1808 eint vector in pcreposix.c.
1809
1810 9. Error texts for pcre_compile() are held as one long string to avoid too
1811 much relocation at load time. To find a text, the string is searched,
1812 counting zeros. There was no check for running off the end of the string,
1813 which could happen if a new error number was added without updating the
1814 string.
1815
1816 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
1817
1818 11. \K was not working if it appeared in an atomic group or in a group that
1819 was called as a "subroutine", or in an assertion. Perl 5.11 documents that
1820 \K is "not well defined" if used in an assertion. PCRE now accepts it if
1821 the assertion is positive, but not if it is negative.
1822
1823 12. Change 11 fortuitously reduced the size of the stack frame used in the
1824 "match()" function of pcre_exec.c by one pointer. Forthcoming
1825 implementation of support for (*MARK) will need an extra pointer on the
1826 stack; I have reserved it now, so that the stack frame size does not
1827 decrease.
1828
1829 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
1830 item in branch that calls a recursion is a subroutine call - as in the
1831 second branch in the above example - was incorrectly given the compile-
1832 time error "recursive call could loop indefinitely" because pcre_compile()
1833 was not correctly checking the subroutine for matching a non-empty string.
1834
1835 14. The checks for overrunning compiling workspace could trigger after an
1836 overrun had occurred. This is a "should never occur" error, but it can be
1837 triggered by pathological patterns such as hundreds of nested parentheses.
1838 The checks now trigger 100 bytes before the end of the workspace.
1839
1840 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
1841
1842
1843 Version 8.01 19-Jan-2010
1844 ------------------------
1845
1846 1. If a pattern contained a conditional subpattern with only one branch (in
1847 particular, this includes all (*DEFINE) patterns), a call to pcre_study()
1848 computed the wrong minimum data length (which is of course zero for such
1849 subpatterns). This could cause incorrect "no match" results.
1850
1851 2. For patterns such as (?i)a(?-i)b|c where an option setting at the start of
1852 the pattern is reset in the first branch, pcre_compile() failed with
1853 "internal error: code overflow at offset...". This happened only when
1854 the reset was to the original external option setting. (An optimization
1855 abstracts leading options settings into an external setting, which was the
1856 cause of this.)
1857
1858 3. A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
1859 of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
1860 assertion pattern did not match (meaning that the assertion was true), it
1861 was incorrectly treated as false if the SKIP had been reached during the
1862 matching. This also applied to assertions used as conditions.
1863
1864 4. If an item that is not supported by pcre_dfa_exec() was encountered in an
1865 assertion subpattern, including such a pattern used as a condition,
1866 unpredictable results occurred, instead of the error return
1867 PCRE_ERROR_DFA_UITEM.
1868
1869 5. The C++ GlobalReplace function was not working like Perl for the special
1870 situation when an empty string is matched. It now does the fancy magic
1871 stuff that is necessary.
1872
1873 6. In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
1874 removed. (These were left over from very, very early versions of PCRE.)
1875
1876 7. Some cosmetic changes to the code to make life easier when compiling it
1877 as part of something else:
1878
1879 (a) Change DEBUG to PCRE_DEBUG.
1880
1881 (b) In pcre_compile(), rename the member of the "branch_chain" structure
1882 called "current" as "current_branch", to prevent a collision with the
1883 Linux macro when compiled as a kernel module.
1884
1885 (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
1886 prevent a collision with the Linux macro when compiled as a kernel
1887 module.
1888
1889 8. In pcre_compile() there are some checks for integer overflows that used to
1890 cast potentially large values to (double). This has been changed to that
1891 when building, a check for int64_t is made, and if it is found, it is used
1892 instead, thus avoiding the use of floating point arithmetic. (There is no
1893 other use of FP in PCRE.) If int64_t is not found, the fallback is to
1894 double.
1895
1896 9. Added two casts to avoid signed/unsigned warnings from VS Studio Express
1897 2005 (difference between two addresses compared to an unsigned value).
1898
1899 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
1900 custom one, because of the following reported problem in Windows:
1901
1902 - libbz2 uses the Pascal calling convention (WINAPI) for the functions
1903 under Win32.
1904 - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
1905 therefore missing the function definition.
1906 - The compiler thus generates a "C" signature for the test function.
1907 - The linker fails to find the "C" function.
1908 - PCRE fails to configure if asked to do so against libbz2.
1909
1910 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
1911 messages were output:
1912
1913 Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
1914 rerunning libtoolize, to keep the correct libtool macros in-tree.
1915 Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
1916
1917 I have done both of these things.
1918
1919 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
1920 most of the time, it *can* run out if it is given a pattern that contains a
1921 runaway infinite recursion. I updated the discussion in the pcrestack man
1922 page.
1923
1924 13. Now that we have gone to the x.xx style of version numbers, the minor
1925 version may start with zero. Using 08 or 09 is a bad idea because users
1926 might check the value of PCRE_MINOR in their code, and 08 or 09 may be
1927 interpreted as invalid octal numbers. I've updated the previous comment in
1928 configure.ac, and also added a check that gives an error if 08 or 09 are
1929 used.
1930
1931 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
1932 causing partial matching to fail when the end of the subject matched \W
1933 in a UTF-8 pattern where \W was quantified with a minimum of 3.
1934
1935 15. There were some discrepancies between the declarations in pcre_internal.h
1936 of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
1937 their definitions. The declarations used "const uschar *" and the
1938 definitions used USPTR. Even though USPTR is normally defined as "const
1939 unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
1940 reported that: "This difference in casting confuses some C++ compilers, for
1941 example, SunCC recognizes above declarations as different functions and
1942 generates broken code for hbpcre." I have changed the declarations to use
1943 USPTR.
1944
1945 16. GNU libtool is named differently on some systems. The autogen.sh script now
1946 tries several variants such as glibtoolize (MacOSX) and libtoolize1x
1947 (FreeBSD).
1948
1949 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
1950 (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
1951 comment: "Figure out how to create a longlong from a string: strtoll and
1952 equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
1953 instance, but it only takes 2 args instead of 3!"
1954
1955 18. A subtle bug concerned with back references has been fixed by a change of
1956 specification, with a corresponding code fix. A pattern such as
1957 ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
1958 refers, was giving matches when it shouldn't. For example, xa=xaaa would
1959 match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
1960 same bug. Such groups have to be quantified to be useful, or contained
1961 inside another quantified group. (If there's no repetition, the reference
1962 can never match.) The problem arises because, having left the group and
1963 moved on to the rest of the pattern, a later failure that backtracks into
1964 the group uses the captured value from the final iteration of the group
1965 rather than the correct earlier one. I have fixed this in PCRE by forcing
1966 any group that contains a reference to itself to be an atomic group; that
1967 is, there cannot be any backtracking into it once it has completed. This is
1968 similar to recursive and subroutine calls.
1969
1970
1971 Version 8.00 19-Oct-09
1972 ----------------------
1973
1974 1. The table for translating pcre_compile() error codes into POSIX error codes
1975 was out-of-date, and there was no check on the pcre_compile() error code
1976 being within the table. This could lead to an OK return being given in
1977 error.
1978
1979 2. Changed the call to open a subject file in pcregrep from fopen(pathname,
1980 "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
1981 in a Windows environment.
1982
1983 3. The pcregrep --count option prints the count for each file even when it is
1984 zero, as does GNU grep. However, pcregrep was also printing all files when
1985 --files-with-matches was added. Now, when both options are given, it prints
1986 counts only for those files that have at least one match. (GNU grep just
1987 prints the file name in this circumstance, but including the count seems
1988 more useful - otherwise, why use --count?) Also ensured that the
1989 combination -clh just lists non-zero counts, with no names.
1990
1991 4. The long form of the pcregrep -F option was incorrectly implemented as
1992 --fixed_strings instead of --fixed-strings. This is an incompatible change,
1993 but it seems right to fix it, and I didn't think it was worth preserving
1994 the old behaviour.
1995
1996 5. The command line items --regex=pattern and --regexp=pattern were not
1997 recognized by pcregrep, which required --regex pattern or --regexp pattern
1998 (with a space rather than an '='). The man page documented the '=' forms,
1999 which are compatible with GNU grep; these now work.
2000
2001 6. No libpcreposix.pc file was created for pkg-config; there was just
2002 libpcre.pc and libpcrecpp.pc. The omission has been rectified.
2003
2004 7. Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
2005 when UCP support is not needed, by modifying the Python script that
2006 generates it from Unicode data files. This should not matter if the module
2007 is correctly used as a library, but I received one complaint about 50K of
2008 unwanted data. My guess is that the person linked everything into his
2009 program rather than using a library. Anyway, it does no harm.
2010
2011 8. A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
2012 was a minimum greater than 1 for a wide character in a possessive
2013 repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
2014 which had an unlimited repeat of a nested, fixed maximum repeat of a wide
2015 character. Chaos in the form of incorrect output or a compiling loop could
2016 result.
2017
2018 9. The restrictions on what a pattern can contain when partial matching is
2019 requested for pcre_exec() have been removed. All patterns can now be
2020 partially matched by this function. In addition, if there are at least two
2021 slots in the offset vector, the offset of the earliest inspected character
2022 for the match and the offset of the end of the subject are set in them when
2023 PCRE_ERROR_PARTIAL is returned.
2024
2025 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
2026 synonymous with PCRE_PARTIAL, for backwards compatibility, and
2027 PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
2028 and may be more useful for multi-segment matching.
2029
2030 11. Partial matching with pcre_exec() is now more intuitive. A partial match
2031 used to be given if ever the end of the subject was reached; now it is
2032 given only if matching could not proceed because another character was
2033 needed. This makes a difference in some odd cases such as Z(*FAIL) with the
2034 string "Z", which now yields "no match" instead of "partial match". In the
2035 case of pcre_dfa_exec(), "no match" is given if every matching path for the
2036 final character ended with (*FAIL).
2037
2038 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
2039 if the pattern had a "must contain" character that was already found in the
2040 earlier partial match, unless partial matching was again requested. For
2041 example, with the pattern /dog.(body)?/, the "must contain" character is
2042 "g". If the first part-match was for the string "dog", restarting with
2043 "sbody" failed. This bug has been fixed.
2044
2045 13. The string returned by pcre_dfa_exec() after a partial match has been
2046 changed so that it starts at the first inspected character rather than the
2047 first character of the match. This makes a difference only if the pattern
2048 starts with a lookbehind assertion or \b or \B (\K is not supported by
2049 pcre_dfa_exec()). It's an incompatible change, but it makes the two
2050 matching functions compatible, and I think it's the right thing to do.
2051
2052 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
2053 so that the demonstration program is easily available in environments where
2054 PCRE has not been installed from source.
2055
2056 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
2057 libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
2058 library.
2059
2060 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
2061 It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
2062 is not the first non-POSIX option to be added. Clearly some people find
2063 these options useful.
2064
2065 17. If a caller to the POSIX matching function regexec() passes a non-zero
2066 value for nmatch with a NULL value for pmatch, the value of
2067 nmatch is forced to zero.
2068
2069 18. RunGrepTest did not have a test for the availability of the -u option of
2070 the diff command, as RunTest does. It now checks in the same way as
2071 RunTest, and also checks for the -b option.
2072
2073 19. If an odd number of negated classes containing just a single character
2074 interposed, within parentheses, between a forward reference to a named
2075 subpattern and the definition of the subpattern, compilation crashed with
2076 an internal error, complaining that it could not find the referenced
2077 subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
2078 [The bug was that it was starting one character too far in when skipping
2079 over the character class, thus treating the ] as data rather than
2080 terminating the class. This meant it could skip too much.]
2081
2082 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
2083 /g option in pcretest when the pattern contains \K, which makes it possible
2084 to have an empty string match not at the start, even when the pattern is
2085 anchored. Updated pcretest and pcredemo to use this option.
2086
2087 21. If the maximum number of capturing subpatterns in a recursion was greater
2088 than the maximum at the outer level, the higher number was returned, but
2089 with unset values at the outer level. The correct (outer level) value is
2090 now given.
2091
2092 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
2093 PCRE did not set those parentheses (unlike Perl). I have now found a way to
2094 make it do so. The string so far is captured, making this feature
2095 compatible with Perl.
2096
2097 23. The tests have been re-organized, adding tests 11 and 12, to make it
2098 possible to check the Perl 5.10 features against Perl 5.10.
2099
2100 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
2101 pattern matches a fixed length string. PCRE did not allow this; now it
2102 does. Neither allows recursion.
2103
2104 25. I finally figured out how to implement a request to provide the minimum
2105 length of subject string that was needed in order to match a given pattern.
2106 (It was back references and recursion that I had previously got hung up
2107 on.) This code has now been added to pcre_study(); it finds a lower bound
2108 to the length of subject needed. It is not necessarily the greatest lower
2109 bound, but using it to avoid searching strings that are too short does give
2110 some useful speed-ups. The value is available to calling programs via
2111 pcre_fullinfo().
2112
2113 26. While implementing 25, I discovered to my embarrassment that pcretest had
2114 not been passing the result of pcre_study() to pcre_dfa_exec(), so the
2115 study optimizations had never been tested with that matching function.
2116 Oops. What is worse, even when it was passed study data, there was a bug in
2117 pcre_dfa_exec() that meant it never actually used it. Double oops. There
2118 were also very few tests of studied patterns with pcre_dfa_exec().
2119
2120 27. If (?| is used to create subpatterns with duplicate numbers, they are now
2121 allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
2122 on the other side of the coin, they are no longer allowed to have different
2123 names, because these cannot be distinguished in PCRE, and this has caused
2124 confusion. (This is a difference from Perl.)
2125
2126 28. When duplicate subpattern names are present (necessarily with different
2127 numbers, as required by 27 above), and a test is made by name in a
2128 conditional pattern, either for a subpattern having been matched, or for
2129 recursion in such a pattern, all the associated numbered subpatterns are
2130 tested, and the overall condition is true if the condition is true for any
2131 one of them. This is the way Perl works, and is also more like the way
2132 testing by number works.
2133
2134
2135 Version 7.9 11-Apr-09
2136 ---------------------
2137
2138 1. When building with support for bzlib/zlib (pcregrep) and/or readline
2139 (pcretest), all targets were linked against these libraries. This included
2140 libpcre, libpcreposix, and libpcrecpp, even though they do not use these
2141 libraries. This caused unwanted dependencies to be created. This problem
2142 has been fixed, and now only pcregrep is linked with bzlib/zlib and only
2143 pcretest is linked with readline.
2144
2145 2. The "typedef int BOOL" in pcre_internal.h that was included inside the
2146 "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
2147 moved outside it again, because FALSE and TRUE are already defined in AIX,
2148 but BOOL is not.
2149
2150 3. The pcre_config() function was treating the PCRE_MATCH_LIMIT and
2151 PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
2152
2153 4. The pcregrep documentation said spaces were inserted as well as colons (or
2154 hyphens) following file names and line numbers when outputting matching
2155 lines. This is not true; no spaces are inserted. I have also clarified the
2156 wording for the --colour (or --color) option.
2157
2158 5. In pcregrep, when --colour was used with -o, the list of matching strings
2159 was not coloured; this is different to GNU grep, so I have changed it to be
2160 the same.
2161
2162 6. When --colo(u)r was used in pcregrep, only the first matching substring in
2163 each matching line was coloured. Now it goes on to look for further matches
2164 of any of the test patterns, which is the same behaviour as GNU grep.
2165
2166 7. A pattern that could match an empty string could cause pcregrep to loop; it
2167 doesn't make sense to accept an empty string match in pcregrep, so I have
2168 locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
2169 seems to be how GNU grep behaves. [But see later change 40 for release
2170 8.33.]
2171
2172 8. The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
2173 start or after a newline", because the conditional assertion was not being
2174 correctly handled. The rule now is that both the assertion and what follows
2175 in the first alternative must satisfy the test.
2176
2177 9. If auto-callout was enabled in a pattern with a conditional group whose
2178 condition was an assertion, PCRE could crash during matching, both with
2179 pcre_exec() and pcre_dfa_exec().
2180
2181 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
2182 used for matching.
2183
2184 11. Unicode property support in character classes was not working for
2185 characters (bytes) greater than 127 when not in UTF-8 mode.
2186
2187 12. Added the -M command line option to pcretest.
2188
2189 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
2190
2191 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
2192
2193 16. Added comments and documentation about mis-use of no_arg in the C++
2194 wrapper.
2195
2196 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
2197 from Martin Jerabek that uses macro names for all relevant character and
2198 string constants.
2199
2200 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
2201 SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
2202 SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
2203 these, but not everybody uses configure.
2204
2205 19. A conditional group that had only one branch was not being correctly
2206 recognized as an item that could match an empty string. This meant that an
2207 enclosing group might also not be so recognized, causing infinite looping
2208 (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
2209 with the subject "ab", where knowledge that the repeated group can match
2210 nothing is needed in order to break the loop.
2211
2212 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
2213 exec(), but without supplying a callout function, matching went wrong.
2214
2215 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
2216 leak if the size of the offset vector was greater than 30. When the vector
2217 is smaller, the saved offsets during recursion go onto a local stack
2218 vector, but for larger vectors malloc() is used. It was failing to free
2219 when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
2220 error, in fact).
2221
2222 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
2223 heapframe that is used only when UTF-8 support is enabled. This caused no
2224 problem, but was untidy.
2225
2226 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
2227 CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
2228 included within another project.
2229
2230 24. Steven Van Ingelgem's patches to add more options to the CMake support,
2231 slightly modified by me:
2232
2233 (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
2234 not building pcregrep.
2235
2236 (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
2237 if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
2238
2239 25. Forward references, both numeric and by name, in patterns that made use of
2240 duplicate group numbers, could behave incorrectly or give incorrect errors,
2241 because when scanning forward to find the reference group, PCRE was not
2242 taking into account the duplicate group numbers. A pattern such as
2243 ^X(?3)(a)(?|(b)|(q))(Y) is an example.
2244
2245 26. Changed a few more instances of "const unsigned char *" to USPTR, making
2246 the feature of a custom pointer more persuasive (as requested by a user).
2247
2248 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
2249 pcretest.c, inside #ifndefs, because it seems they are sometimes already
2250 pre-defined.
2251
2252 28. Added support for (*UTF8) at the start of a pattern.
2253
2254 29. Arrange for flags added by the "release type" setting in CMake to be shown
2255 in the configuration summary.
2256
2257
2258 Version 7.8 05-Sep-08
2259 ---------------------
2260
2261 1. Replaced UCP searching code with optimized version as implemented for Ad
2262 Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
2263 stage table and inline lookup instead of a function, giving speed ups of 2
2264 to 5 times on some simple patterns that I tested. Permission was given to
2265 distribute the MultiStage2.py script that generates the tables (it's not in
2266 the tarball, but is in the Subversion repository).
2267
2268 2. Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
2269 scripts.
2270
2271 3. Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
2272 a group with a zero qualifier. The result of the study could be incorrect,
2273 or the function might crash, depending on the pattern.
2274
2275 4. Caseless matching was not working for non-ASCII characters in back
2276 references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
2277 It now works when Unicode Property Support is available.
2278
2279 5. In pcretest, an escape such as \x{de} in the data was always generating
2280 a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
2281 non-UTF-8 mode. If the value is greater than 255, it gives a warning about
2282 truncation.
2283
2284 6. Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
2285
2286 7. Added two (int) casts to pcregrep when printing the difference of two
2287 pointers, in case they are 64-bit values.
2288
2289 8. Added comments about Mac OS X stack usage to the pcrestack man page and to
2290 test 2 if it fails.
2291
2292 9. Added PCRE_CALL_CONVENTION just before the names of all exported functions,
2293 and a #define of that name to empty if it is not externally set. This is to
2294 allow users of MSVC to set it if necessary.
2295
2296 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
2297 the convenience functions in the pcre_get.c source file.
2298
2299 11. An option change at the start of a pattern that had top-level alternatives
2300 could cause overwriting and/or a crash. This command provoked a crash in
2301 some environments:
2302
2303 printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
2304
2305 This potential security problem was recorded as CVE-2008-2371.
2306
2307 12. For a pattern where the match had to start at the beginning or immediately
2308 after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
2309 pcre_dfa_exec() could read past the end of the passed subject if there was
2310 no match. To help with detecting such bugs (e.g. with valgrind), I modified
2311 pcretest so that it places the subject at the end of its malloc-ed buffer.
2312
2313 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
2314 exec() might read past the end of the data buffer in UTF-8 mode.
2315
2316 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
2317 the data contained the byte 0x85 as part of a UTF-8 character within its
2318 first line. This applied both to normal and DFA matching.
2319
2320 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
2321 /^[^d]*?$/8 failed to match "abc".
2322
2323 16. Added a missing copyright notice to pcrecpp_internal.h.
2324
2325 17. Make it more clear in the documentation that values returned from
2326 pcre_exec() in ovector are byte offsets, not character counts.
2327
2328 18. Tidied a few places to stop certain compilers from issuing warnings.
2329
2330 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
2331 supplied by Stefan Weber. I made a further small update for 7.8 because
2332 there is a change of source arrangements: the pcre_searchfuncs.c module is
2333 replaced by pcre_ucd.c.
2334
2335
2336 Version 7.7 07-May-08
2337 ---------------------
2338
2339 1. Applied Craig's patch to sort out a long long problem: "If we can't convert
2340 a string to a long long, pretend we don't even have a long long." This is
2341 done by checking for the strtoq, strtoll, and _strtoi64 functions.
2342
2343 2. Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
2344 pre-7.6 versions, which defined a global no_arg variable instead of putting
2345 it in the RE class. (See also #8 below.)
2346
2347 3. Remove a line of dead code, identified by coverity and reported by Nuno
2348 Lopes.
2349
2350 4. Fixed two related pcregrep bugs involving -r with --include or --exclude:
2351
2352 (1) The include/exclude patterns were being applied to the whole pathnames
2353 of files, instead of just to the final components.
2354
2355 (2) If there was more than one level of directory, the subdirectories were
2356 skipped unless they satisfied the include/exclude conditions. This is
2357 inconsistent with GNU grep (and could even be seen as contrary to the
2358 pcregrep specification - which I improved to make it absolutely clear).
2359 The action now is always to scan all levels of directory, and just
2360 apply the include/exclude patterns to regular files.
2361
2362 5. Added the --include_dir and --exclude_dir patterns to pcregrep, and used
2363 --exclude_dir in the tests to avoid scanning .svn directories.
2364
2365 6. Applied Craig's patch to the QuoteMeta function so that it escapes the
2366 NUL character as backslash + 0 rather than backslash + NUL, because PCRE
2367 doesn't support NULs in patterns.
2368
2369 7. Added some missing "const"s to declarations of static tables in
2370 pcre_compile.c and pcre_dfa_exec.c.
2371
2372 8. Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
2373 caused by fix #2 above. (Subsequently also a second patch to fix the
2374 first patch. And a third patch - this was a messy problem.)
2375
2376 9. Applied Craig's patch to remove the use of push_back().
2377
2378 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
2379 matching function regexec().
2380
2381 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
2382 which, however, unlike Perl's \g{...}, are subroutine calls, not back
2383 references. PCRE supports relative numbers with this syntax (I don't think
2384 Oniguruma does).
2385
2386 12. Previously, a group with a zero repeat such as (...){0} was completely
2387 omitted from the compiled regex. However, this means that if the group
2388 was called as a subroutine from elsewhere in the pattern, things went wrong
2389 (an internal error was given). Such groups are now left in the compiled
2390 pattern, with a new opcode that causes them to be skipped at execution
2391 time.
2392
2393 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
2394 to the way PCRE behaves:
2395
2396 (a) A lone ] character is dis-allowed (Perl treats it as data).
2397
2398 (b) A back reference to an unmatched subpattern matches an empty string
2399 (Perl fails the current match path).
2400
2401 (c) A data ] in a character class must be notated as \] because if the
2402 first data character in a class is ], it defines an empty class. (In
2403 Perl it is not possible to have an empty class.) The empty class []
2404 never matches; it forces failure and is equivalent to (*FAIL) or (?!).
2405 The negative empty class [^] matches any one character, independently
2406 of the DOTALL setting.
2407
2408 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
2409 non-existent subpattern following a character class starting with ']' and
2410 containing () gave an internal compiling error instead of "reference to
2411 non-existent subpattern". Fortunately, when the pattern did exist, the
2412 compiled code was correct. (When scanning forwards to check for the
2413 existence of the subpattern, it was treating the data ']' as terminating
2414 the class, so got the count wrong. When actually compiling, the reference
2415 was subsequently set up correctly.)
2416
2417 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
2418 it was being rejected as not supported by pcre_dfa_exec(), even though
2419 other assertions are supported. I have made pcre_dfa_exec() support
2420 (*FAIL).
2421
2422 16. The implementation of 13c above involved the invention of a new opcode,
2423 OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
2424 cannot be changed at match time, I realized I could make a small
2425 improvement to matching performance by compiling OP_ALLANY instead of
2426 OP_ANY for "." when DOTALL was set, and then removing the runtime tests
2427 on the OP_ANY path.
2428
2429 17. Compiling pcretest on Windows with readline support failed without the
2430 following two fixes: (1) Make the unistd.h include conditional on
2431 HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
2432
2433 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
2434 ncurses library to be included for pcretest when ReadLine support is
2435 requested, but also to allow for it to be overridden. This patch came from
2436 Daniel Bergström.
2437
2438 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
2439 as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
2440 any errors with the current Unicode tables. Thanks to Peter Kankowski for
2441 spotting this.
2442
2443
2444 Version 7.6 28-Jan-08
2445 ---------------------
2446
2447 1. A character class containing a very large number of characters with
2448 codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
2449 overflow.
2450
2451 2. Patch to cut out the "long long" test in pcrecpp_unittest when
2452 HAVE_LONG_LONG is not defined.
2453
2454 3. Applied Christian Ehrlicher's patch to update the CMake build files to
2455 bring them up to date and include new features. This patch includes:
2456
2457 - Fixed PH's badly added libz and libbz2 support.
2458 - Fixed a problem with static linking.
2459 - Added pcredemo. [But later removed - see 7 below.]
2460 - Fixed dftables problem and added an option.
2461 - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
2462 HAVE_LONG_LONG.
2463 - Added readline support for pcretest.
2464 - Added an listing of the option settings after cmake has run.
2465
2466 4. A user submitted a patch to Makefile that makes it easy to create
2467 "pcre.dll" under mingw when using Configure/Make. I added stuff to
2468 Makefile.am that cause it to include this special target, without
2469 affecting anything else. Note that the same mingw target plus all
2470 the other distribution libraries and programs are now supported
2471 when configuring with CMake (see 6 below) instead of with
2472 Configure/Make.
2473
2474 5. Applied Craig's patch that moves no_arg into the RE class in the C++ code.
2475 This is an attempt to solve the reported problem "pcrecpp::no_arg is not
2476 exported in the Windows port". It has not yet been confirmed that the patch
2477 solves the problem, but it does no harm.
2478
2479 6. Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
2480 NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
2481 with CMake, and also correct the comment about stack recursion.
2482
2483 7. Remove the automatic building of pcredemo from the ./configure system and
2484 from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
2485 of a program that users should build themselves after PCRE is installed, so
2486 building it automatically is not really right. What is more, it gave
2487 trouble in some build environments.
2488
2489 8. Further tidies to CMakeLists.txt from Sheri and Christian.
2490
2491
2492 Version 7.5 10-Jan-08
2493 ---------------------
2494
2495 1. Applied a patch from Craig: "This patch makes it possible to 'ignore'
2496 values in parens when parsing an RE using the C++ wrapper."
2497
2498 2. Negative specials like \S did not work in character classes in UTF-8 mode.
2499 Characters greater than 255 were excluded from the class instead of being
2500 included.
2501
2502 3. The same bug as (2) above applied to negated POSIX classes such as
2503 [:^space:].
2504
2505 4. PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
2506 defined or documented. It seems to have been a typo for PCRE_STATIC, so
2507 I have changed it.
2508
2509 5. The construct (?&) was not diagnosed as a syntax error (it referenced the
2510 first named subpattern) and a construct such as (?&a) would reference the
2511 first named subpattern whose name started with "a" (in other words, the
2512 length check was missing). Both these problems are fixed. "Subpattern name
2513 expected" is now given for (?&) (a zero-length name), and this patch also
2514 makes it give the same error for \k'' (previously it complained that that
2515 was a reference to a non-existent subpattern).
2516
2517 6. The erroneous patterns (?+-a) and (?-+a) give different error messages;
2518 this is right because (?- can be followed by option settings as well as by
2519 digits. I have, however, made the messages clearer.
2520
2521 7. Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
2522 than the number used in the conditional) now cause a compile-time error.
2523 This is actually not compatible with Perl, which accepts such patterns, but
2524 treats the conditional as always being FALSE (as PCRE used to), but it
2525 seems to me that giving a diagnostic is better.
2526
2527 8. Change "alphameric" to the more common word "alphanumeric" in comments
2528 and messages.
2529
2530 9. Fix two occurrences of "backslash" in comments that should have been
2531 "backspace".
2532
2533 10. Remove two redundant lines of code that can never be obeyed (their function
2534 was moved elsewhere).
2535
2536 11. The program that makes PCRE's Unicode character property table had a bug
2537 which caused it to generate incorrect table entries for sequences of
2538 characters that have the same character type, but are in different scripts.
2539 It amalgamated them into a single range, with the script of the first of
2540 them. In other words, some characters were in the wrong script. There were
2541 thirteen such cases, affecting characters in the following ranges:
2542
2543 U+002b0 - U+002c1
2544 U+0060c - U+0060d
2545 U+0061e - U+00612
2546 U+0064b - U+0065e
2547 U+0074d - U+0076d
2548 U+01800 - U+01805
2549 U+01d00 - U+01d77
2550 U+01d9b - U+01dbf
2551 U+0200b - U+0200f
2552 U+030fc - U+030fe
2553 U+03260 - U+0327f
2554 U+0fb46 - U+0fbb1
2555 U+10450 - U+1049d
2556
2557 12. The -o option (show only the matching part of a line) for pcregrep was not
2558 compatible with GNU grep in that, if there was more than one match in a
2559 line, it showed only the first of them. It now behaves in the same way as
2560 GNU grep.
2561
2562 13. If the -o and -v options were combined for pcregrep, it printed a blank
2563 line for every non-matching line. GNU grep prints nothing, and pcregrep now
2564 does the same. The return code can be used to tell if there were any
2565 non-matching lines.
2566
2567 14. Added --file-offsets and --line-offsets to pcregrep.
2568
2569 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
2570 infinitely looping recursion. The bug was that positive lookaheads were not
2571 being skipped when checking for a possible empty match (negative lookaheads
2572 and both kinds of lookbehind were skipped).
2573
2574 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
2575 inclusion of <windows.h> to before rather than after the definition of
2576 INVALID_FILE_ATTRIBUTES (patch from David Byron).
2577
2578 17. Specifying a possessive quantifier with a specific limit for a Unicode
2579 character property caused pcre_compile() to compile bad code, which led at
2580 runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
2581 are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
2582 caused the error; without that there was no problem.
2583
2584 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
2585
2586 19. Added --enable-pcretest-libreadline.
2587
2588 20. In pcrecpp.cc, the variable 'count' was incremented twice in
2589 RE::GlobalReplace(). As a result, the number of replacements returned was
2590 double what it should be. I removed one of the increments, but Craig sent a
2591 later patch that removed the other one (the right fix) and added unit tests
2592 that check the return values (which was not done before).
2593
2594 21. Several CMake things:
2595
2596 (1) Arranged that, when cmake is used on Unix, the libraries end up with
2597 the names libpcre and libpcreposix, not just pcre and pcreposix.
2598
2599 (2) The above change means that pcretest and pcregrep are now correctly
2600 linked with the newly-built libraries, not previously installed ones.
2601
2602 (3) Added PCRE_SUPPORT_LIBREADLINE, PCRE_SUPPORT_LIBZ, PCRE_SUPPORT_LIBBZ2.
2603
2604 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
2605 crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
2606 UTF-8 newline character). The key issue is that the pattern starts .*;
2607 this means that the match must be either at the beginning, or after a
2608 newline. The bug was in the code for advancing after a failed match and
2609 checking that the new position followed a newline. It was not taking
2610 account of UTF-8 characters correctly.
2611
2612 23. PCRE was behaving differently from Perl in the way it recognized POSIX
2613 character classes. PCRE was not treating the sequence [:...:] as a
2614 character class unless the ... were all letters. Perl, however, seems to
2615 allow any characters between [: and :], though of course it rejects as
2616 unknown any "names" that contain non-letters, because all the known class
2617 names consist only of letters. Thus, Perl gives an error for [[:1234:]],
2618 for example, whereas PCRE did not - it did not recognize a POSIX character
2619 class. This seemed a bit dangerous, so the code has been changed to be
2620 closer to Perl. The behaviour is not identical to Perl, because PCRE will
2621 diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
2622 treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
2623 Perl does, and where it didn't before.
2624
2625 24. Rewrite so as to remove the single use of %n from pcregrep because in some
2626 Windows environments %n is disabled by default.
2627
2628
2629 Version 7.4 21-Sep-07
2630 ---------------------
2631
2632 1. Change 7.3/28 was implemented for classes by looking at the bitmap. This
2633 means that a class such as [\s] counted as "explicit reference to CR or
2634 LF". That isn't really right - the whole point of the change was to try to
2635 help when there was an actual mention of one of the two characters. So now
2636 the change happens only if \r or \n (or a literal CR or LF) character is
2637 encountered.
2638
2639 2. The 32-bit options word was also used for 6 internal flags, but the numbers
2640 of both had grown to the point where there were only 3 bits left.
2641 Fortunately, there was spare space in the data structure, and so I have
2642 moved the internal flags into a new 16-bit field to free up more option
2643 bits.
2644
2645 3. The appearance of (?J) at the start of a pattern set the DUPNAMES option,
2646 but did not set the internal JCHANGED flag - either of these is enough to
2647 control the way the "get" function works - but the PCRE_INFO_JCHANGED
2648 facility is supposed to tell if (?J) was ever used, so now (?J) at the
2649 start sets both bits.
2650
2651 4. Added options (at build time, compile time, exec time) to change \R from
2652 matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
2653
2654 5. doc/pcresyntax.html was missing from the distribution.
2655
2656 6. Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
2657 compatibility, even though it is no longer used.
2658
2659 7. Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
2660 strtoull to pcrecpp.cc to select the available functions in WIN32 when the
2661 windows.h file is present (where different names are used). [This was
2662 reversed later after testing - see 16 below.]
2663
2664 8. Changed all #include <config.h> to #include "config.h". There were also
2665 some further <pcre.h> cases that I changed to "pcre.h".
2666
2667 9. When pcregrep was used with the --colour option, it missed the line ending
2668 sequence off the lines that it output.
2669
2670 10. It was pointed out to me that arrays of string pointers cause lots of
2671 relocations when a shared library is dynamically loaded. A technique of
2672 using a single long string with a table of offsets can drastically reduce
2673 these. I have refactored PCRE in four places to do this. The result is
2674 dramatic:
2675
2676 Originally: 290
2677 After changing UCP table: 187
2678 After changing error message table: 43
2679 After changing table of "verbs" 36
2680 After changing table of Posix names 22
2681
2682 Thanks to the folks working on Gregex for glib for this insight.
2683
2684 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
2685 unicode-properties was also set.
2686
2687 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
2688
2689 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
2690 checked only for CRLF.
2691
2692 14. Added casts to pcretest.c to avoid compiler warnings.
2693
2694 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
2695
2696 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
2697 and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
2698 entirely. This removes changes made in 7 above.
2699
2700 17. The CMake files have been updated, and there is now more information about
2701 building with CMake in the NON-UNIX-USE document.
2702
2703
2704 Version 7.3 28-Aug-07
2705 ---------------------
2706
2707 1. In the rejigging of the build system that eventually resulted in 7.1, the
2708 line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
2709 brackets there is not right, since it causes compilers to look for an
2710 installed pcre.h, not the version that is in the source that is being
2711 compiled (which of course may be different). I have changed it back to:
2712
2713 #include "pcre.h"
2714
2715 I have a vague recollection that the change was concerned with compiling in
2716 different directories, but in the new build system, that is taken care of
2717 by the VPATH setting the Makefile.
2718
2719 2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
2720 when the subject happened to end in the byte 0x85 (e.g. if the last
2721 character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
2722 characters but of course it shouldn't be taken as a newline when it is part
2723 of another character. The bug was that, for an unlimited repeat of . in
2724 not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
2725 characters when looking for a newline.
2726
2727 3. A small performance improvement in the DOTALL UTF-8 mode .* case.
2728
2729 4. Debugging: adjusted the names of opcodes for different kinds of parentheses
2730 in debug output.
2731
2732 5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
2733 long printing in the pcrecpp unittest when running under MinGW.
2734
2735 6. ESC_K was left out of the EBCDIC table.
2736
2737 7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
2738 parentheses; I made it 1000, which seemed large enough. Unfortunately, the
2739 limit also applies to "virtual nesting" when a pattern is recursive, and in
2740 this case 1000 isn't so big. I have been able to remove this limit at the
2741 expense of backing off one optimization in certain circumstances. Normally,
2742 when pcre_exec() would call its internal match() function recursively and
2743 immediately return the result unconditionally, it uses a "tail recursion"
2744 feature to save stack. However, when a subpattern that can match an empty
2745 string has an unlimited repetition quantifier, it no longer makes this
2746 optimization. That gives it a stack frame in which to save the data for
2747 checking that an empty string has been matched. Previously this was taken
2748 from the 1000-entry workspace that had been reserved. So now there is no
2749 explicit limit, but more stack is used.
2750
2751 8. Applied Daniel's patches to solve problems with the import/export magic
2752 syntax that is required for Windows, and which was going wrong for the
2753 pcreposix and pcrecpp parts of the library. These were overlooked when this
2754 problem was solved for the main library.
2755
2756 9. There were some crude static tests to avoid integer overflow when computing
2757 the size of patterns that contain repeated groups with explicit upper
2758 limits. As the maximum quantifier is 65535, the maximum group length was
2759 set at 30,000 so that the product of these two numbers did not overflow a
2760 32-bit integer. However, it turns out that people want to use groups that
2761 are longer than 30,000 bytes (though not repeat them that many times).
2762 Change 7.0/17 (the refactoring of the way the pattern size is computed) has
2763 made it possible to implement the integer overflow checks in a much more
2764 dynamic way, which I have now done. The artificial limitation on group
2765 length has been removed - we now have only the limit on the total length of
2766 the compiled pattern, which depends on the LINK_SIZE setting.
2767
2768 10. Fixed a bug in the documentation for get/copy named substring when
2769 duplicate names are permitted. If none of the named substrings are set, the
2770 functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
2771 empty string.
2772
2773 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
2774 instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
2775 because the ] is interpreted as the first data character and the
2776 terminating ] is not found. PCRE has been made compatible with Perl in this
2777 regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
2778 cause memory overwriting.
2779
2780 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
2781 string has been matched (to stop an infinite loop). It was not recognizing
2782 a conditional subpattern that could match an empty string if that
2783 subpattern was within another subpattern. For example, it looped when
2784 trying to match (((?(1)X|))*) but it was OK with ((?(1)X|)*) where the
2785 condition was not nested. This bug has been fixed.
2786
2787 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
2788 past the start of the subject in the presence of bytes with the top bit
2789 set, for example "\x8aBCD".
2790
2791 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
2792 (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
2793
2794 14. Optimized (?!) to (*FAIL).
2795
2796 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
2797 This restricts code points to be within the range 0 to 0x10FFFF, excluding
2798 the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
2799 full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
2800 does: it's just the validity check that is more restrictive.
2801
2802 16. Inserted checks for integer overflows during escape sequence (backslash)
2803 processing, and also fixed erroneous offset values for syntax errors during
2804 backslash processing.
2805
2806 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
2807 for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
2808
2809 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
2810 caused an overrun.
2811
2812 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
2813 something other than just ASCII characters) inside a group that had an
2814 unlimited repeat caused a loop at compile time (while checking to see
2815 whether the group could match an empty string).
2816
2817 20. Debugging a pattern containing \p or \P could cause a crash. For example,
2818 [\P{Any}] did so. (Error in the code for printing property names.)
2819
2820 21. An orphan \E inside a character class could cause a crash.
2821
2822 22. A repeated capturing bracket such as (A)? could cause a wild memory
2823 reference during compilation.
2824
2825 23. There are several functions in pcre_compile() that scan along a compiled
2826 expression for various reasons (e.g. to see if it's fixed length for look
2827 behind). There were bugs in these functions when a repeated \p or \P was
2828 present in the pattern. These operators have additional parameters compared
2829 with \d, etc, and these were not being taken into account when moving along
2830 the compiled data. Specifically:
2831
2832 (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
2833 length.
2834
2835 (b) An item such as \pL+ within a repeated group could cause crashes or
2836 loops.
2837
2838 (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
2839 "reference to non-existent subpattern" error.
2840
2841 (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
2842
2843 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
2844 characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
2845
2846 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
2847
2848 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
2849 character were causing crashes (broken optimization).
2850
2851 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
2852 \p or \P) caused a compile-time loop.
2853
2854 28. More problems have arisen in unanchored patterns when CRLF is a valid line
2855 break. For example, the unstudied pattern [\r\n]A does not match the string
2856 "\r\nA" because change 7.0/46 below moves the current point on by two
2857 characters after failing to match at the start. However, the pattern \nA
2858 *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
2859 the same is true. There doesn't seem any very clean way out of this, but
2860 what I have chosen to do makes the common cases work: PCRE now takes note
2861 of whether there can be an explicit match for \r or \n anywhere in the
2862 pattern, and if so, 7.0/46 no longer applies. As part of this change,
2863 there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
2864 pattern has explicit CR or LF references.
2865
2866 29. Added (*CR) etc for changing newline setting at start of pattern.
2867
2868
2869 Version 7.2 19-Jun-07
2870 ---------------------
2871
2872 1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
2873 which is apparently normally available under Windows.
2874
2875 2. Re-jig the pcregrep tests with different newline settings in an attempt
2876 to make them independent of the local environment's newline setting.
2877
2878 3. Add code to configure.ac to remove -g from the CFLAGS default settings.
2879
2880 4. Some of the "internals" tests were previously cut out when the link size
2881 was not 2, because the output contained actual offsets. The recent new
2882 "Z" feature of pcretest means that these can be cut out, making the tests
2883 usable with all link sizes.
2884
2885 5. Implemented Stan Switzer's goto replacement for longjmp() when not using
2886 stack recursion. This gives a massive performance boost under BSD, but just
2887 a small improvement under Linux. However, it saves one field in the frame
2888 in all cases.
2889
2890 6. Added more features from the forthcoming Perl 5.10:
2891
2892 (a) (?-n) (where n is a string of digits) is a relative subroutine or
2893 recursion call. It refers to the nth most recently opened parentheses.
2894
2895 (b) (?+n) is also a relative subroutine call; it refers to the nth next
2896 to be opened parentheses.
2897
2898 (c) Conditions that refer to capturing parentheses can be specified
2899 relatively, for example, (?(-2)... or (?(+3)...
2900
2901 (d) \K resets the start of the current match so that everything before
2902 is not part of it.
2903
2904 (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
2905
2906 (f) \g{name} is another synonym - part of Perl 5.10's unification of
2907 reference syntax.
2908
2909 (g) (?| introduces a group in which the numbering of parentheses in each
2910 alternative starts with the same number.
2911
2912 (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
2913
2914 7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
2915 PCRE_INFO_JCHANGED.
2916
2917 8. A pattern such as (.*(.)?)* caused pcre_exec() to fail by either not
2918 terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
2919 for detecting groups that can match an empty string.
2920
2921 9. A pattern with a very large number of alternatives (more than several
2922 hundred) was running out of internal workspace during the pre-compile
2923 phase, where pcre_compile() figures out how much memory will be needed. A
2924 bit of new cunning has reduced the workspace needed for groups with
2925 alternatives. The 1000-alternative test pattern now uses 12 bytes of
2926 workspace instead of running out of the 4096 that are available.
2927
2928 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
2929
2930 11. Applied patch from Google to remove an optimization that didn't quite work.
2931 The report of the bug said:
2932
2933 pcrecpp::RE("a*").FullMatch("aaa") matches, while
2934 pcrecpp::RE("a*?").FullMatch("aaa") does not, and
2935 pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
2936
2937 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
2938 it matched the wrong number of bytes.
2939
2940
2941 Version 7.1 24-Apr-07
2942 ---------------------
2943
2944 1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
2945 that is more "standard", making use of automake and other Autotools. There
2946 is some re-arrangement of the files and adjustment of comments consequent
2947 on this.
2948
2949 2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
2950 for recursive directory scanning broke on some systems because the files
2951 are not scanned in any specific order and on different systems the order
2952 was different. A call to "sort" has been inserted into RunGrepTest for the
2953 approprate test as a short-term fix. In the longer term there may be an
2954 alternative.
2955
2956 3. I had an email from Eric Raymond about problems translating some of PCRE's
2957 man pages to HTML (despite the fact that I distribute HTML pages, some
2958 people do their own conversions for various reasons). The problems
2959 concerned the use of low-level troff macros .br and .in. I have therefore
2960 removed all such uses from the man pages (some were redundant, some could
2961 be replaced by .nf/.fi pairs). The 132html script that I use to generate
2962 HTML has been updated to handle .nf/.fi and to complain if it encounters
2963 .br or .in.
2964
2965 4. Updated comments in configure.ac that get placed in config.h.in and also
2966 arranged for config.h to be included in the distribution, with the name
2967 config.h.generic, for the benefit of those who have to compile without
2968 Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
2969
2970 5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
2971 Weber: (1) pcre_internal.h was missing some function renames; (2) updated
2972 makevp.bat for the current PCRE, using the additional files
2973 makevp_c.txt, makevp_l.txt, and pcregexp.pas.
2974
2975 6. A Windows user reported a minor discrepancy with test 2, which turned out
2976 to be caused by a trailing space on an input line that had got lost in his
2977 copy. The trailing space was an accident, so I've just removed it.
2978
2979 7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
2980 that is needed.
2981
2982 8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
2983 as "const" (a) because they are and (b) because it helps the PHP
2984 maintainers who have recently made a script to detect big data structures
2985 in the php code that should be moved to the .rodata section. I remembered
2986 to update Builducptable as well, so it won't revert if ucptable.h is ever
2987 re-created.
2988
2989 9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
2990 pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
2991 order to be able to cut out the UTF-8 tables in the latter when UTF-8
2992 support is not required. This saves 1.5-2K of code, which is important in
2993 some applications.
2994
2995 Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
2996 so as not to refer to the tables, even though these functions will never be
2997 called when UTF-8 support is disabled. Otherwise there are problems with a
2998 shared library.
2999
3000 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
3001
3002 (a) It was defining its arguments as char * instead of void *.
3003
3004 (b) It was assuming that all moves were upwards in memory; this was true
3005 a long time ago when I wrote it, but is no longer the case.
3006
3007 The emulated memove() is provided for those environments that have neither
3008 memmove() nor bcopy(). I didn't think anyone used it these days, but that
3009 is clearly not the case, as these two bugs were recently reported.
3010
3011 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
3012 and Detrail to create the HTML documentation, the .txt form of the man
3013 pages, and it removes trailing spaces from listed files. It also creates
3014 pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
3015 case, it wraps all the #defines with #ifndefs. This script should be run
3016 before "make dist".
3017
3018 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
3019 with Unicode property support.
3020
3021 (a) For a maximizing quantifier, if the two different cases of the
3022 character were of different lengths in their UTF-8 codings (there are
3023 some cases like this - I found 11), and the matching function had to
3024 back up over a mixture of the two cases, it incorrectly assumed they
3025 were both the same length.
3026
3027 (b) When PCRE was configured to use the heap rather than the stack for
3028 recursion during matching, it was not correctly preserving the data for
3029 the other case of a UTF-8 character when checking ahead for a match
3030 while processing a minimizing repeat. If the check also involved
3031 matching a wide character, but failed, corruption could cause an
3032 erroneous result when trying to check for a repeat of the original
3033 character.
3034
3035 13. Some tidying changes to the testing mechanism:
3036
3037 (a) The RunTest script now detects the internal link size and whether there
3038 is UTF-8 and UCP support by running ./pcretest -C instead of relying on
3039 values substituted by "configure". (The RunGrepTest script already did
3040 this for UTF-8.) The configure.ac script no longer substitutes the
3041 relevant variables.
3042
3043 (b) The debugging options /B and /D in pcretest show the compiled bytecode
3044 with length and offset values. This means that the output is different
3045 for different internal link sizes. Test 2 is skipped for link sizes
3046 other than 2 because of this, bypassing the problem. Unfortunately,
3047 there was also a test in test 3 (the locale tests) that used /B and
3048 failed for link sizes other than 2. Rather than cut the whole test out,
3049 I have added a new /Z option to pcretest that replaces the length and
3050 offset values with spaces. This is now used to make test 3 independent
3051 of link size. (Test 2 will be tidied up later.)
3052
3053 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
3054 segmentation fault instead of returning the appropriate error message.
3055
3056 15. In multiline mode when the newline sequence was set to "any", the pattern
3057 ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
3058 This doesn't seem right; it now treats the CRLF combination as the line
3059 ending, and so does not match in that case. It's only a pattern such as ^$
3060 that would hit this one: something like ^ABC$ would have failed after \r
3061 and then tried again after \r\n.
3062
3063 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
3064 in an attempt to make files that differ only in their line terminators
3065 compare equal. This works on Linux.
3066
3067 17. Under certain error circumstances pcregrep might try to free random memory
3068 as it exited. This is now fixed, thanks to valgrind.
3069
3070 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
3071 "abc\r\n\r\n", it found an unwanted second match after the second \r. This
3072 was because its rules for how to advance for /g after matching an empty
3073 string at the end of a line did not allow for this case. They now check for
3074 it specially.
3075
3076 20. pcretest is supposed to handle patterns and data of any length, by
3077 extending its buffers when necessary. It was getting this wrong when the
3078 buffer for a data line had to be extended.
3079
3080 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
3081 CRLF as a newline sequence.
3082
3083 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
3084 out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
3085 I have nevertheless tidied it up.
3086
3087 23. Added some casts to kill warnings from HP-UX ia64 compiler.
3088
3089 24. Added a man page for pcre-config.
3090
3091
3092 Version 7.0 19-Dec-06
3093 ---------------------
3094
3095 1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
3096 moving to gcc 4.1.1.
3097
3098 2. The -S option for pcretest uses setrlimit(); I had omitted to #include
3099 sys/time.h, which is documented as needed for this function. It doesn't
3100 seem to matter on Linux, but it showed up on some releases of OS X.
3101
3102 3. It seems that there are systems where bytes whose values are greater than
3103 127 match isprint() in the "C" locale. The "C" locale should be the
3104 default when a C program starts up. In most systems, only ASCII printing
3105 characters match isprint(). This difference caused the output from pcretest
3106 to vary, making some of the tests fail. I have changed pcretest so that:
3107
3108 (a) When it is outputting text in the compiled version of a pattern, bytes
3109 other than 32-126 are always shown as hex escapes.
3110
3111 (b) When it is outputting text that is a matched part of a subject string,
3112 it does the same, unless a different locale has been set for the match
3113 (using the /L modifier). In this case, it uses isprint() to decide.
3114
3115 4. Fixed a major bug that caused incorrect computation of the amount of memory
3116 required for a compiled pattern when options that changed within the
3117 pattern affected the logic of the preliminary scan that determines the
3118 length. The relevant options are -x, and -i in UTF-8 mode. The result was
3119 that the computed length was too small. The symptoms of this bug were
3120 either the PCRE error "internal error: code overflow" from pcre_compile(),
3121 or a glibc crash with a message such as "pcretest: free(): invalid next
3122 size (fast)". Examples of patterns that provoked this bug (shown in
3123 pcretest format) are:
3124
3125 /(?-x: )/x
3126 /(?x)(?-x: \s*#\s*)/
3127 /((?i)[\x{c0}])/8
3128 /(?i:[\x{c0}])/8
3129
3130 HOWEVER: Change 17 below makes this fix obsolete as the memory computation
3131 is now done differently.
3132
3133 5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
3134 wrapper classes; (b) implement a new function in the C++ scanner that is
3135 more efficient than the old way of doing things because it avoids levels of
3136 recursion in the regex matching; (c) add a paragraph to the documentation
3137 for the FullMatch() function.
3138
3139 6. The escape sequence \n was being treated as whatever was defined as
3140 "newline". Not only was this contrary to the documentation, which states
3141 that \n is character 10 (hex 0A), but it also went horribly wrong when
3142 "newline" was defined as CRLF. This has been fixed.
3143
3144 7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
3145 was being set to -1 for the "end of line" case (supposedly a value that no
3146 character can have). Though this value is never used (the check for end of
3147 line is "zero bytes in current character"), it caused compiler complaints.
3148 I've changed it to 0xffffffff.
3149
3150 8. In pcre_version.c, the version string was being built by a sequence of
3151 C macros that, in the event of PCRE_PRERELEASE being defined as an empty
3152 string (as it is for production releases) called a macro with an empty
3153 argument. The C standard says the result of this is undefined. The gcc
3154 compiler treats it as an empty string (which was what was wanted) but it is
3155 reported that Visual C gives an error. The source has been hacked around to
3156 avoid this problem.
3157
3158 9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
3159 builds of pcretest, and changed the call to _setmode() to use _O_BINARY
3160 instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
3161 of them did).
3162
3163 10. Originally, pcretest opened its input and output without "b"; then I was
3164 told that "b" was needed in some environments, so it was added for release
3165 5.0 to both the input and output. (It makes no difference on Unix-like
3166 systems.) Later I was told that it is wrong for the input on Windows. I've
3167 now abstracted the modes into two macros, to make it easier to fiddle with
3168 them, and removed "b" from the input mode under Windows.
3169
3170 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
3171
3172 12. Added -help and --help to pcretest as an official way of being reminded
3173 of the options.
3174
3175 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
3176 and pcrecpp.cc because they annoy compilers at high warning levels.
3177
3178 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
3179
3180 15. Fixed an occurrence of == in configure.ac that should have been = (shell
3181 scripts are not C programs :-) and which was not noticed because it works
3182 on Linux.
3183
3184 16. pcretest is supposed to handle any length of pattern and data line (as one
3185 line or as a continued sequence of lines) by extending its input buffer if
3186 necessary. This feature was broken for very long pattern lines, leading to
3187 a string of junk being passed to pcre_compile() if the pattern was longer
3188 than about 50K.
3189
3190 17. I have done a major re-factoring of the way pcre_compile() computes the
3191 amount of memory needed for a compiled pattern. Previously, there was code
3192 that made a preliminary scan of the pattern in order to do this. That was
3193 OK when PCRE was new, but as the facilities have expanded, it has become
3194 harder and harder to keep it in step with the real compile phase, and there
3195 have been a number of bugs (see for example, 4 above). I have now found a
3196 cunning way of running the real compile function in a "fake" mode that
3197 enables it to compute how much memory it would need, while actually only
3198 ever using a few hundred bytes of working memory and without too many
3199 tests of the mode. This should make future maintenance and development
3200 easier. A side effect of this work is that the limit of 200 on the nesting
3201 depth of parentheses has been removed (though this was never a serious
3202 limitation, I suspect). However, there is a downside: pcre_compile() now
3203 runs more slowly than before (30% or more, depending on the pattern). I
3204 hope this isn't a big issue. There is no effect on runtime performance.
3205
3206 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
3207 newline (only possible for the last line of a file) and it was a
3208 pattern that set a locale (followed by /Lsomething), pcretest crashed.
3209
3210 19. Added additional timing features to pcretest. (1) The -tm option now times
3211 matching only, not compiling. (2) Both -t and -tm can be followed, as a
3212 separate command line item, by a number that specifies the number of
3213 repeats to use when timing. The default is 50000; this gives better
3214 precision, but takes uncomfortably long for very large patterns.
3215
3216 20. Extended pcre_study() to be more clever in cases where a branch of a
3217 subpattern has no definite first character. For example, (a*|b*)[cd] would
3218 previously give no result from pcre_study(). Now it recognizes that the
3219 first character must be a, b, c, or d.
3220
3221 21. There was an incorrect error "recursive call could loop indefinitely" if
3222 a subpattern (or the entire pattern) that was being tested for matching an
3223 empty string contained only one non-empty item after a nested subpattern.
3224 For example, the pattern (?>\x{100}*)\d(?R) provoked this error
3225 incorrectly, because the \d was being skipped in the check.
3226
3227 22. The pcretest program now has a new pattern option /B and a command line
3228 option -b, which is equivalent to adding /B to every pattern. This causes
3229 it to show the compiled bytecode, without the additional information that
3230 -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
3231 is the same as /B/I).
3232
3233 23. A new optimization is now able automatically to treat some sequences such
3234 as a*b as a*+b. More specifically, if something simple (such as a character
3235 or a simple class like \d) has an unlimited quantifier, and is followed by
3236 something that cannot possibly match the quantified thing, the quantifier
3237 is automatically "possessified".
3238
3239 24. A recursive reference to a subpattern whose number was greater than 39
3240 went wrong under certain circumstances in UTF-8 mode. This bug could also
3241 have affected the operation of pcre_study().
3242
3243 25. Realized that a little bit of performance could be had by replacing
3244 (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
3245
3246 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
3247
3248 27. Possessive quantifiers such as a++ were previously implemented by turning
3249 them into atomic groups such as ($>a+). Now they have their own opcodes,
3250 which improves performance. This includes the automatically created ones
3251 from 23 above.
3252
3253 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
3254 lookahead was broken if it was not anchored. PCRE was mistakenly expecting
3255 the first matched character to be a colon. This applied both to named and
3256 numbered groups.
3257
3258 29. The ucpinternal.h header file was missing its idempotency #ifdef.
3259
3260 30. I was sent a "project" file called libpcre.a.dev which I understand makes
3261 building PCRE on Windows easier, so I have included it in the distribution.
3262
3263 31. There is now a check in pcretest against a ridiculously large number being
3264 returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
3265 loop, the loop is abandoned.
3266
3267 32. Forward references to subpatterns in conditions such as (?(2)...) where
3268 subpattern 2 is defined later cause pcre_compile() to search forwards in
3269 the pattern for the relevant set of parentheses. This search went wrong
3270 when there were unescaped parentheses in a character class, parentheses
3271 escaped with \Q...\E, or parentheses in a #-comment in /x mode.
3272
3273 33. "Subroutine" calls and backreferences were previously restricted to
3274 referencing subpatterns earlier in the regex. This restriction has now
3275 been removed.
3276
3277 34. Added a number of extra features that are going to be in Perl 5.10. On the
3278 whole, these are just syntactic alternatives for features that PCRE had
3279 previously implemented using the Python syntax or my own invention. The
3280 other formats are all retained for compatibility.
3281
3282 (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
3283 as (?P<name>...). The new forms, as well as being in Perl 5.10, are
3284 also .NET compatible.
3285
3286 (b) A recursion or subroutine call to a named group can now be defined as
3287 (?&name) as well as (?P>name).
3288
3289 (c) A backreference to a named group can now be defined as \k<name> or
3290 \k'name' as well as (?P=name). The new forms, as well as being in Perl
3291 5.10, are also .NET compatible.
3292
3293 (d) A conditional reference to a named group can now use the syntax
3294 (?(<name>) or (?('name') as well as (?(name).
3295
3296 (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
3297 groups (named and numbered) that are never evaluated inline, but can be
3298 called as "subroutines" from elsewhere. In effect, the DEFINE condition
3299 is always false. There may be only one alternative in such a group.
3300
3301 (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
3302 as the simple (?(R). The condition is true only if the most recent
3303 recursion is that of the given number or name. It does not search out
3304 through the entire recursion stack.
3305
3306 (g) The escape \gN or \g{N} has been added, where N is a positive or
3307 negative number, specifying an absolute or relative reference.
3308
3309 35. Tidied to get rid of some further signed/unsigned compiler warnings and
3310 some "unreachable code" warnings.
3311
3312 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
3313 things, this adds five new scripts.
3314
3315 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
3316 There were also incompatibilities regarding the handling of \Q..\E inside
3317 character classes, for example with patterns like [\Qa\E-\Qz\E] where the
3318 hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
3319
3320 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
3321 matches an empty string, and forcibly breaks the loop. There were bugs in
3322 this code in non-simple cases. For a pattern such as ^(a()*)* matched
3323 against aaaa the result was just "a" rather than "aaaa", for example. Two
3324 separate and independent bugs (that affected different cases) have been
3325 fixed.
3326
3327 39. Refactored the code to abolish the use of different opcodes for small
3328 capturing bracket numbers. This is a tidy that I avoided doing when I
3329 removed the limit on the number of capturing brackets for 3.5 back in 2001.
3330 The new approach is not only tidier, it makes it possible to reduce the
3331 memory needed to fix the previous bug (38).
3332
3333 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
3334 sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
3335 processing dot, circumflex, or dollar metacharacters, or #-comments in /x
3336 mode.
3337
3338 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
3339 report.
3340
3341 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
3342 copy construction and assignment in the C++ wrapper.
3343
3344 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
3345 couple of bugs that could have given wrong results in the "--newline=crlf"
3346 case.
3347
3348 44. Added a number of casts and did some reorganization of signed/unsigned int
3349 variables following suggestions from Dair Grant. Also renamed the variable
3350 "this" as "item" because it is a C++ keyword.
3351
3352 45. Arranged for dftables to add
3353
3354 #include "pcre_internal.h"
3355
3356 to pcre_chartables.c because without it, gcc 4.x may remove the array
3357 definition from the final binary if PCRE is built into a static library and
3358 dead code stripping is activated.
3359
3360 46. For an unanchored pattern, if a match attempt fails at the start of a
3361 newline sequence, and the newline setting is CRLF or ANY, and the next two
3362 characters are CRLF, advance by two characters instead of one.
3363
3364
3365 Version 6.7 04-Jul-06
3366 ---------------------
3367
3368 1. In order to handle tests when input lines are enormously long, pcretest has
3369 been re-factored so that it automatically extends its buffers when
3370 necessary. The code is crude, but this _is_ just a test program. The
3371 default size has been increased from 32K to 50K.
3372
3373 2. The code in pcre_study() was using the value of the re argument before
3374 testing it for NULL. (Of course, in any sensible call of the function, it
3375 won't be NULL.)
3376
3377 3. The memmove() emulation function in pcre_internal.h, which is used on
3378 systems that lack both memmove() and bcopy() - that is, hardly ever -
3379 was missing a "static" storage class specifier.
3380
3381 4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
3382 containing an extended class (one that cannot be represented by a bitmap
3383 because it contains high-valued characters or Unicode property items, e.g.
3384 [\pZ]). Almost always one would set UTF-8 mode when processing such a
3385 pattern, but PCRE should not loop if you do not (it no longer does).
3386 [Detail: two cases were found: (a) a repeated subpattern containing an
3387 extended class; (b) a recursive reference to a subpattern that followed a
3388 previous extended class. It wasn't skipping over the extended class
3389 correctly when UTF-8 mode was not set.]
3390
3391 5. A negated single-character class was not being recognized as fixed-length
3392 in lookbehind assertions such as (?<=[^f]), leading to an incorrect
3393 compile error "lookbehind assertion is not fixed length".
3394
3395 6. The RunPerlTest auxiliary script was showing an unexpected difference
3396 between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
3397 write a Perl script that can interpret lines of an input file either as
3398 byte characters or as UTF-8, which is what "perltest" was being required to
3399 do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
3400 can't do is switch easily at run time between having the "use utf8;" pragma
3401 or not. In the end, I fudged it by using the RunPerlTest script to insert
3402 "use utf8;" explicitly for the UTF-8 tests.
3403
3404 7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
3405 the end of the subject string, contrary to the documentation and to what
3406 Perl does. This was true of both matching functions. Now it matches only at
3407 the start of the subject and immediately after *internal* newlines.
3408
3409 8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
3410 a pointer to an int instead of a pointer to an unsigned long int. This
3411 caused problems on 64-bit systems.
3412
3413 9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
3414 instance of the 'standard' template library not being so standard".
3415
3416 10. There was no check on the number of named subpatterns nor the maximum
3417 length of a subpattern name. The product of these values is used to compute
3418 the size of the memory block for a compiled pattern. By supplying a very
3419 long subpattern name and a large number of named subpatterns, the size
3420 computation could be caused to overflow. This is now prevented by limiting
3421 the length of names to 32 characters, and the number of named subpatterns
3422 to 10,000.
3423
3424 11. Subpatterns that are repeated with specific counts have to be replicated in
3425 the compiled pattern. The size of memory for this was computed from the
3426 length of the subpattern and the repeat count. The latter is limited to
3427 65535, but there was no limit on the former, meaning that integer overflow
3428 could in principle occur. The compiled length of a repeated subpattern is
3429 now limited to 30,000 bytes in order to prevent this.
3430
3431 12. Added the optional facility to have named substrings with the same name.
3432
3433 13. Added the ability to use a named substring as a condition, using the
3434 Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
3435 are numbers (not recommended). Forward references are permitted.
3436
3437 14. Added forward references in named backreferences (if you see what I mean).
3438
3439 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
3440 pattern could run off the end of the subject. For example, the pattern
3441 "(?s)(.{1,5})"8 did this with the subject "ab".
3442
3443 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
3444 PCRE_CASELESS was set when matching characters that were quantified with ?
3445 or *.
3446
3447 17. A character class other than a single negated character that had a minimum
3448 but no maximum quantifier - for example [ab]{6,} - was not handled
3449 correctly by pce_dfa_exec(). It would match only one character.
3450
3451 18. A valid (though odd) pattern that looked like a POSIX character
3452 class but used an invalid character after [ (for example [[,abc,]]) caused
3453 pcre_compile() to give the error "Failed: internal error: code overflow" or
3454 in some cases to crash with a glibc free() error. This could even happen if
3455 the pattern terminated after [[ but there just happened to be a sequence of
3456 letters, a binary zero, and a closing ] in the memory that followed.
3457
3458 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
3459 over the years. Originally (before any Unicode support), just the bottom 8
3460 bits were taken. Thus, for example, \500 really meant \100. Nowadays the
3461 output from "man perlunicode" includes this:
3462
3463 The regular expression compiler produces polymorphic opcodes. That
3464 is, the pattern adapts to the data and automatically switches to
3465 the Unicode character scheme when presented with Unicode data--or
3466 instead uses a traditional byte scheme when presented with byte
3467 data.
3468
3469 Sadly, a wide octal escape does not cause a switch, and in a string with
3470 no other multibyte characters, these octal escapes are treated as before.
3471 Thus, in Perl, the pattern /\500/ actually matches \100 but the pattern
3472 /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
3473 Unicode string.
3474
3475 I have not perpetrated such confusion in PCRE. Up till now, it took just
3476 the bottom 8 bits, as in old Perl. I have now made octal escapes with
3477 values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
3478 translate to the appropriate multibyte character.
3479
3480 29. Applied some refactoring to reduce the number of warnings from Microsoft
3481 and Borland compilers. This has included removing the fudge introduced
3482 seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
3483 a warning about an unused variable.
3484
3485 21. PCRE has not included VT (character 0x0b) in the set of whitespace
3486 characters since release 4.0, because Perl (from release 5.004) does not.
3487 [Or at least, is documented not to: some releases seem to be in conflict
3488 with the documentation.] However, when a pattern was studied with
3489 pcre_study() and all its branches started with \s, PCRE still included VT
3490 as a possible starting character. Of course, this did no harm; it just
3491 caused an unnecessary match attempt.
3492
3493 22. Removed a now-redundant internal flag bit that recorded the fact that case
3494 dependency changed within the pattern. This was once needed for "required
3495 byte" processing, but is no longer used. This recovers a now-scarce options
3496 bit. Also moved the least significant internal flag bit to the most-
3497 significant bit of the word, which was not previously used (hangover from
3498 the days when it was an int rather than a uint) to free up another bit for
3499 the future.
3500
3501 23. Added support for CRLF line endings as well as CR and LF. As well as the
3502 default being selectable at build time, it can now be changed at runtime
3503 via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
3504 specify that it is scanning data with non-default line endings.
3505
3506 24. Changed the definition of CXXLINK to make it agree with the definition of
3507 LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
3508
3509 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
3510 recursions. This makes a big different to stack usage for some patterns.
3511
3512 26. If a subpattern containing a named recursion or subroutine reference such
3513 as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
3514 the space required for the compiled pattern went wrong and gave too small a
3515 value. Depending on the environment, this could lead to "Failed: internal
3516 error: code overflow at offset 49" or "glibc detected double free or
3517 corruption" errors.
3518
3519 27. Applied patches from Google (a) to support the new newline modes and (b) to
3520 advance over multibyte UTF-8 characters in GlobalReplace.
3521
3522 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
3523 difference for some implementation of PCRE in some Windows version.
3524
3525 29. Added some extra testing facilities to pcretest:
3526
3527 \q<number> in a data line sets the "match limit" value
3528 \Q<number> in a data line sets the "match recursion limt" value
3529 -S <number> sets the stack size, where <number> is in megabytes
3530
3531 The -S option isn't available for Windows.
3532
3533
3534 Version 6.6 06-Feb-06
3535 ---------------------
3536
3537 1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
3538 in pcreposix.h. I have copied the definition from pcre.h.
3539
3540 2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
3541 because pcre.h is no longer a built file.
3542
3543 3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
3544 not normally included in the compiled code.
3545
3546
3547 Version 6.5 01-Feb-06
3548 ---------------------
3549
3550 1. When using the partial match feature with pcre_dfa_exec(), it was not
3551 anchoring the second and subsequent partial matches at the new starting
3552 point. This could lead to incorrect results. For example, with the pattern
3553 /1234/, partially matching against "123" and then "a4" gave a match.
3554
3555 2. Changes to pcregrep:
3556
3557 (a) All non-match returns from pcre_exec() were being treated as failures
3558 to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
3559 error message is output. Some extra information is given for the
3560 PCRE_ERROR_MATCHLIMIT and PCRE_ERROR_RECURSIONLIMIT errors, which are
3561 probably the only errors that are likely to be caused by users (by
3562 specifying a regex that has nested indefinite repeats, for instance).
3563 If there are more than 20 of these errors, pcregrep is abandoned.
3564
3565 (b) A binary zero was treated as data while matching, but terminated the
3566 output line if it was written out. This has been fixed: binary zeroes
3567 are now no different to any other data bytes.
3568
3569 (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
3570 used to set a locale for matching. The --locale=xxxx long option has
3571 been added (no short equivalent) to specify a locale explicitly on the
3572 pcregrep command, overriding the environment variables.
3573
3574 (d) When -B was used with -n, some line numbers in the output were one less
3575 than they should have been.
3576
3577 (e) Added the -o (--only-matching) option.
3578
3579 (f) If -A or -C was used with -c (count only), some lines of context were
3580 accidentally printed for the final match.
3581
3582 (g) Added the -H (--with-filename) option.
3583
3584 (h) The combination of options -rh failed to suppress file names for files
3585 that were found from directory arguments.
3586
3587 (i) Added the -D (--devices) and -d (--directories) options.
3588
3589 (j) Added the -F (--fixed-strings) option.
3590
3591 (k) Allow "-" to be used as a file name for -f as well as for a data file.
3592
3593 (l) Added the --colo(u)r option.
3594
3595 (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
3596 is not present by default.
3597
3598 3. A nasty bug was discovered in the handling of recursive patterns, that is,
3599 items such as (?R) or (?1), when the recursion could match a number of
3600 alternatives. If it matched one of the alternatives, but subsequently,
3601 outside the recursion, there was a failure, the code tried to back up into
3602 the recursion. However, because of the way PCRE is implemented, this is not
3603 possible, and the result was an incorrect result from the match.
3604
3605 In order to prevent this happening, the specification of recursion has
3606 been changed so that all such subpatterns are automatically treated as
3607 atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
3608
3609 4. I had overlooked the fact that, in some locales, there are characters for
3610 which isalpha() is true but neither isupper() nor islower() are true. In
3611 the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
3612 and ordfeminine) are like this. This affected the treatment of \w and \W
3613 when they appeared in character classes, but not when they appeared outside
3614 a character class. The bit map for "word" characters is now created
3615 separately from the results of isalnum() instead of just taking it from the
3616 upper, lower, and digit maps. (Plus the underscore character, of course.)
3617
3618 5. The above bug also affected the handling of POSIX character classes such as
3619 [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
3620 permanent tables. Instead, the bit maps for such a class were previously
3621 created as the appropriate unions of the upper, lower, and digit bitmaps.
3622 Now they are created by subtraction from the [[:word:]] class, which has
3623 its own bitmap.
3624
3625 6. The [[:blank:]] character class matches horizontal, but not vertical space.
3626 It is created by subtracting the vertical space characters (\x09, \x0a,
3627 \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
3628 subtraction was done in the overall bitmap for a character class, meaning
3629 that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
3630 be recognized. This bug has been fixed.
3631
3632 7. Patches from the folks at Google:
3633
3634 (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
3635 real life, but is still worth protecting against".
3636
3637 (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
3638 regular expressions".
3639
3640 (c) pcre_scanner.cc: avoid use of std::count() because not all systems
3641 have it.
3642
3643 (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
3644 "configure" and the latter not, in order to fix a problem somebody had
3645 with compiling the Arg class on HP-UX.
3646
3647 (e) Improve the error-handling of the C++ wrapper a little bit.
3648
3649 (f) New tests for checking recursion limiting.
3650
3651 8. The pcre_memmove() function, which is used only if the environment does not
3652 have a standard memmove() function (and is therefore rarely compiled),
3653 contained two bugs: (a) use of int instead of size_t, and (b) it was not
3654 returning a result (though PCRE never actually uses the result).
3655
3656 9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
3657 large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
3658 returned instead of calling malloc() with an overflowing number that would
3659 most likely cause subsequent chaos.
3660
3661 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
3662
3663 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
3664 with this option is matched, the nmatch and pmatch options of regexec() are
3665 ignored.
3666
3667 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
3668 provided in case anyone wants to the the POSIX interface with UTF-8
3669 strings.
3670
3671 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
3672 C++ linking (needed for some HP-UX environments).
3673
3674 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
3675 (unused parameter) and in the pcre_printint() function (omitted "default"
3676 switch label when the default is to do nothing).
3677
3678 15. Added some code to make it possible, when PCRE is compiled as a C++
3679 library, to replace subject pointers for pcre_exec() with a smart pointer
3680 class, thus making it possible to process discontinuous strings.
3681
3682 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
3683 much the same function. They were added by different people who were trying
3684 to make PCRE easy to compile on non-Unix systems. It has been suggested
3685 that PCRE_EXPORT be abolished now that there is more automatic apparatus
3686 for compiling on Windows systems. I have therefore replaced it with
3687 PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
3688 defaults to "extern" for C or "extern C" for C++, which works fine on
3689 Unix-like systems. It is now possible to override the value of PCRE_DATA_
3690 SCOPE with something explicit in config.h. In addition:
3691
3692 (a) pcreposix.h still had just "extern" instead of either of these macros;
3693 I have replaced it with PCRE_DATA_SCOPE.
3694
3695 (b) Functions such as _pcre_xclass(), which are internal to the library,
3696 but external in the C sense, all had PCRE_EXPORT in their definitions.
3697 This is apparently wrong for the Windows case, so I have removed it.
3698 (It makes no difference on Unix-like systems.)
3699
3700 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
3701 of recursive calls to match(). This is different to MATCH_LIMIT because
3702 that limits the total number of calls to match(), not all of which increase
3703 the depth of recursion. Limiting the recursion depth limits the amount of
3704 stack (or heap if NO_RECURSE is set) that is used. The default can be set
3705 when PCRE is compiled, and changed at run time. A patch from Google adds
3706 this functionality to the C++ interface.
3707
3708 18. Changes to the handling of Unicode character properties:
3709
3710 (a) Updated the table to Unicode 4.1.0.
3711
3712 (b) Recognize characters that are not in the table as "Cn" (undefined).
3713
3714 (c) I revised the way the table is implemented to a much improved format
3715 which includes recognition of ranges. It now supports the ranges that
3716 are defined in UnicodeData.txt, and it also amalgamates other
3717 characters into ranges. This has reduced the number of entries in the
3718 table from around 16,000 to around 3,000, thus reducing its size
3719 considerably. I realized I did not need to use a tree structure after
3720 all - a binary chop search is just as efficient. Having reduced the
3721 number of entries, I extended their size from 6 bytes to 8 bytes to
3722 allow for more data.
3723
3724 (d) Added support for Unicode script names via properties such as \p{Han}.
3725
3726 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
3727 matching that character.
3728
3729 20. When matching a repeated Unicode property with a minimum greater than zero,
3730 (for example \pL{2,}), PCRE could look past the end of the subject if it
3731 reached it while seeking the minimum number of characters. This could
3732 happen only if some of the characters were more than one byte long, because
3733 there is a check for at least the minimum number of bytes.
3734
3735 21. Refactored the implementation of \p and \P so as to be more general, to
3736 allow for more different types of property in future. This has changed the
3737 compiled form incompatibly. Anybody with saved compiled patterns that use
3738 \p or \P will have to recompile them.
3739
3740 22. Added "Any" and "L&" to the supported property types.
3741
3742 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
3743 but give a compile time error if the value is greater than 0xff.
3744
3745 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
3746 accidentally not being installed or uninstalled.
3747
3748 25. The pcre.h file was built from pcre.h.in, but the only changes that were
3749 made were to insert the current release number. This seemed silly, because
3750 it made things harder for people building PCRE on systems that don't run
3751 "configure". I have turned pcre.h into a distributed file, no longer built
3752 by "configure", with the version identification directly included. There is
3753 no longer a pcre.h.in file.
3754
3755 However, this change necessitated a change to the pcre-config script as
3756 well. It is built from pcre-config.in, and one of the substitutions was the
3757 release number. I have updated configure.ac so that ./configure now finds
3758 the release number by grepping pcre.h.
3759
3760 26. Added the ability to run the tests under valgrind.
3761
3762
3763 Version 6.4 05-Sep-05
3764 ---------------------
3765
3766 1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
3767 "--" to be printed when multiple files were scanned, even when none of the
3768 -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
3769 consider it to be a bug, and have restored the previous behaviour.
3770
3771 2. A couple of code tidies to get rid of compiler warnings.
3772
3773 3. The pcretest program used to cheat by referring to symbols in the library
3774 whose names begin with _pcre_. These are internal symbols that are not
3775 really supposed to be visible externally, and in some environments it is
3776 possible to suppress them. The cheating is now confined to including
3777 certain files from the library's source, which is a bit cleaner.
3778
3779 4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
3780 file's purpose clearer.
3781
3782 5. Reorganized pcre_ucp_findchar().
3783
3784
3785 Version 6.3 15-Aug-05
3786 ---------------------
3787
3788 1. The file libpcre.pc.in did not have general read permission in the tarball.
3789
3790 2. There were some problems when building without C++ support:
3791
3792 (a) If C++ support was not built, "make install" and "make test" still
3793 tried to test it.
3794
3795 (b) There were problems when the value of CXX was explicitly set. Some
3796 changes have been made to try to fix these, and ...
3797
3798 (c) --disable-cpp can now be used to explicitly disable C++ support.
3799
3800 (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
3801 backslash in a target when C++ was disabled. This confuses some
3802 versions of "make", apparently. Using an intermediate variable solves
3803 this. (Same for CPP_LOBJ.)
3804
3805 3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
3806 (non-Windows) now includes $(CFLAGS) because these flags are sometimes
3807 necessary on certain architectures.
3808
3809 4. Added a setting of -export-symbols-regex to the link command to remove
3810 those symbols that are exported in the C sense, but actually are local
3811 within the library, and not documented. Their names all begin with
3812 "_pcre_". This is not a perfect job, because (a) we have to except some
3813 symbols that pcretest ("illegally") uses, and (b) the facility isn't always
3814 available (and never for static libraries). I have made a note to try to
3815 find a way round (a) in the future.
3816
3817
3818 Version 6.2 01-Aug-05
3819 ---------------------
3820
3821 1. There was no test for integer overflow of quantifier values. A construction
3822 such as {1111111111111111} would give undefined results. What is worse, if
3823 a minimum quantifier for a parenthesized subpattern overflowed and became
3824 negative, the calculation of the memory size went wrong. This could have
3825 led to memory overwriting.
3826
3827 2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
3828
3829 3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
3830 operating environments where this matters.
3831
3832 4. Applied Giuseppe Maxia's patch to add additional features for controlling
3833 PCRE options from within the C++ wrapper.
3834
3835 5. Named capturing subpatterns were not being correctly counted when a pattern
3836 was compiled. This caused two problems: (a) If there were more than 100
3837 such subpatterns, the calculation of the memory needed for the whole
3838 compiled pattern went wrong, leading to an overflow error. (b) Numerical
3839 back references of the form \12, where the number was greater than 9, were
3840 not recognized as back references, even though there were sufficient
3841 previous subpatterns.
3842
3843 6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
3844 versions of gcc, e.g. 2.95.4.
3845
3846
3847 Version 6.1 21-Jun-05
3848 ---------------------
3849
3850 1. There was one reference to the variable "posix" in pcretest.c that was not
3851 surrounded by "#if !defined NOPOSIX".
3852
3853 2. Make it possible to compile pcretest without DFA support, UTF8 support, or
3854 the cross-check on the old pcre_info() function, for the benefit of the
3855 cut-down version of PCRE that is currently imported into Exim.
3856
3857 3. A (silly) pattern starting with (?i)(?-i) caused an internal space
3858 allocation error. I've done the easy fix, which wastes 2 bytes for sensible
3859 patterns that start (?i) but I don't think that matters. The use of (?i) is
3860 just an example; this all applies to the other options as well.
3861
3862 4. Since libtool seems to echo the compile commands it is issuing, the output
3863 from "make" can be reduced a bit by putting "@" in front of each libtool
3864 compile command.
3865
3866 5. Patch from the folks at Google for configure.in to be a bit more thorough
3867 in checking for a suitable C++ installation before trying to compile the
3868 C++ stuff. This should fix a reported problem when a compiler was present,
3869 but no suitable headers.
3870
3871 6. The man pages all had just "PCRE" as their title. I have changed them to
3872 be the relevant file name. I have also arranged that these names are
3873 retained in the file doc/pcre.txt, which is a concatenation in text format
3874 of all the man pages except the little individual ones for each function.
3875
3876 7. The NON-UNIX-USE file had not been updated for the different set of source
3877 files that come with release 6. I also added a few comments about the C++
3878 wrapper.
3879
3880
3881 Version 6.0 07-Jun-05
3882 ---------------------
3883
3884 1. Some minor internal re-organization to help with my DFA experiments.
3885
3886 2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
3887 didn't matter for the library itself when fully configured, but did matter
3888 when compiling without UCP support, or within Exim, where the ucp files are
3889 not imported.
3890
3891 3. Refactoring of the library code to split up the various functions into
3892 different source modules. The addition of the new DFA matching code (see
3893 below) to a single monolithic source would have made it really too
3894 unwieldy, quite apart from causing all the code to be include in a
3895 statically linked application, when only some functions are used. This is
3896 relevant even without the DFA addition now that patterns can be compiled in
3897 one application and matched in another.
3898
3899 The downside of splitting up is that there have to be some external
3900 functions and data tables that are used internally in different modules of
3901 the library but which are not part of the API. These have all had their
3902 names changed to start with "_pcre_" so that they are unlikely to clash
3903 with other external names.
3904
3905 4. Added an alternate matching function, pcre_dfa_exec(), which matches using
3906 a different (DFA) algorithm. Although it is slower than the original
3907 function, it does have some advantages for certain types of matching
3908 problem.
3909
3910 5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
3911 including restarting after a partial match.
3912
3913 6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
3914 defined when compiling for Windows was sent to me. I have put it into the
3915 code, though I have no means of testing or verifying it.
3916
3917 7. Added the pcre_refcount() auxiliary function.
3918
3919 8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
3920 match before or at the first newline in the subject string. In pcretest,
3921 the /f option on a pattern can be used to set this.
3922
3923 9. A repeated \w when used in UTF-8 mode with characters greater than 256
3924 would behave wrongly. This has been present in PCRE since release 4.0.
3925
3926 10. A number of changes to the pcregrep command:
3927
3928 (a) Refactored how -x works; insert ^(...)$ instead of setting
3929 PCRE_ANCHORED and checking the length, in preparation for adding
3930 something similar for -w.
3931
3932 (b) Added the -w (match as a word) option.
3933
3934 (c) Refactored the way lines are read and buffered so as to have more
3935 than one at a time available.
3936
3937 (d) Implemented a pcregrep test script.
3938
3939 (e) Added the -M (multiline match) option. This allows patterns to match
3940 over several lines of the subject. The buffering ensures that at least
3941 8K, or the rest of the document (whichever is the shorter) is available
3942 for matching (and similarly the previous 8K for lookbehind assertions).
3943
3944 (f) Changed the --help output so that it now says
3945
3946 -w, --word-regex(p)
3947
3948 instead of two lines, one with "regex" and the other with "regexp"
3949 because that confused at least one person since the short forms are the
3950 same. (This required a bit of code, as the output is generated
3951 automatically from a table. It wasn't just a text change.)
3952
3953 (g) -- can be used to terminate pcregrep options if the next thing isn't an
3954 option but starts with a hyphen. Could be a pattern or a path name
3955 starting with a hyphen, for instance.
3956
3957 (h) "-" can be given as a file name to represent stdin.
3958
3959 (i) When file names are being printed, "(standard input)" is used for
3960 the standard input, for compatibility with GNU grep. Previously
3961 "<stdin>" was used.
3962
3963 (j) The option --label=xxx can be used to supply a name to be used for
3964 stdin when file names are being printed. There is no short form.
3965
3966 (k) Re-factored the options decoding logic because we are going to add
3967 two more options that take data. Such options can now be given in four
3968 different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
3969
3970 (l) Added the -A, -B, and -C options for requesting that lines of context
3971 around matches be printed.
3972
3973 (m) Added the -L option to print the names of files that do not contain
3974 any matching lines, that is, the complement of -l.
3975
3976 (n) The return code is 2 if any file cannot be opened, but pcregrep does
3977 continue to scan other files.
3978
3979 (o) The -s option was incorrectly implemented. For compatibility with other
3980 greps, it now suppresses the error message for a non-existent or non-
3981 accessible file (but not the return code). There is a new option called
3982 -q that suppresses the output of matching lines, which was what -s was
3983 previously doing.
3984
3985 (p) Added --include and --exclude options to specify files for inclusion
3986 and exclusion when recursing.
3987
3988 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
3989 Hopefully, it now does.
3990
3991 12. Missing cast in pcre_study().
3992
3993 13. Added an "uninstall" target to the makefile.
3994
3995 14. Replaced "extern" in the function prototypes in Makefile.in with
3996 "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
3997 world, but is set differently for Windows.
3998
3999 15. Added a second compiling function called pcre_compile2(). The only
4000 difference is that it has an extra argument, which is a pointer to an
4001 integer error code. When there is a compile-time failure, this is set
4002 non-zero, in addition to the error test pointer being set to point to an
4003 error message. The new argument may be NULL if no error number is required
4004 (but then you may as well call pcre_compile(), which is now just a
4005 wrapper). This facility is provided because some applications need a
4006 numeric error indication, but it has also enabled me to tidy up the way
4007 compile-time errors are handled in the POSIX wrapper.
4008
4009 16. Added VPATH=.libs to the makefile; this should help when building with one
4010 prefix path and installing with another. (Or so I'm told by someone who
4011 knows more about this stuff than I do.)
4012
4013 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
4014 passes PCRE_DOTALL to the pcre_compile() function, making the "." character
4015 match everything, including newlines. This is not POSIX-compatible, but
4016 somebody wanted the feature. From pcretest it can be activated by using
4017 both the P and the s flags.
4018
4019 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
4020
4021 19. libpcre.pc was being incorrectly installed as executable.
4022
4023 20. A couple of places in pcretest check for end-of-line by looking for '\n';
4024 it now also looks for '\r' so that it will work unmodified on Windows.
4025
4026 21. Added Google's contributed C++ wrapper to the distribution.
4027
4028 22. Added some untidy missing memory free() calls in pcretest, to keep
4029 Electric Fence happy when testing.
4030
4031
4032
4033 Version 5.0 13-Sep-04
4034 ---------------------
4035
4036 1. Internal change: literal characters are no longer packed up into items
4037 containing multiple characters in a single byte-string. Each character
4038 is now matched using a separate opcode. However, there may be more than one
4039 byte in the character in UTF-8 mode.
4040
4041 2. The pcre_callout_block structure has two new fields: pattern_position and
4042 next_item_length. These contain the offset in the pattern to the next match
4043 item, and its length, respectively.
4044
4045 3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
4046 insertion of callouts before each pattern item. Added the /C option to
4047 pcretest to make use of this.
4048
4049 4. On the advice of a Windows user, the lines
4050
4051 #if defined(_WIN32) || defined(WIN32)
4052 _setmode( _fileno( stdout ), 0x8000 );
4053 #endif /* defined(_WIN32) || defined(WIN32) */
4054
4055 have been added to the source of pcretest. This apparently does useful
4056 magic in relation to line terminators.
4057
4058 5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
4059 for the benefit of those environments where the "b" makes a difference.
4060
4061 6. The icc compiler has the same options as gcc, but "configure" doesn't seem
4062 to know about it. I have put a hack into configure.in that adds in code
4063 to set GCC=yes if CC=icc. This seems to end up at a point in the
4064 generated configure script that is early enough to affect the setting of
4065 compiler options, which is what is needed, but I have no means of testing
4066 whether it really works. (The user who reported this had patched the
4067 generated configure script, which of course I cannot do.)
4068
4069 LATER: After change 22 below (new libtool files), the configure script
4070 seems to know about icc (and also ecc). Therefore, I have commented out
4071 this hack in configure.in.
4072
4073 7. Added support for pkg-config (2 patches were sent in).
4074
4075 8. Negated POSIX character classes that used a combination of internal tables
4076 were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
4077 [[:^ascii]]. Typically, they would match almost any characters. The other
4078 POSIX classes were not broken in this way.
4079
4080 9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
4081 to find the match, as PCRE was deluded into thinking that the match had to
4082 start at the start point or following a newline. The same bug applied to
4083 patterns with negative forward assertions or any backward assertions
4084 preceding ".*" at the start, unless the pattern required a fixed first
4085 character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
4086
4087 10. In UTF-8 mode, when moving forwards in the subject after a failed match
4088 starting at the last subject character, bytes beyond the end of the subject
4089 string were read.
4090
4091 11. Renamed the variable "class" as "classbits" to make life easier for C++
4092 users. (Previously there was a macro definition, but it apparently wasn't
4093 enough.)
4094
4095 12. Added the new field "tables" to the extra data so that tables can be passed
4096 in at exec time, or the internal tables can be re-selected. This allows
4097 a compiled regex to be saved and re-used at a later time by a different
4098 program that might have everything at different addresses.
4099
4100 13. Modified the pcre-config script so that, when run on Solaris, it shows a
4101 -R library as well as a -L library.
4102
4103 14. The debugging options of pcretest (-d on the command line or D on a
4104 pattern) showed incorrect output for anything following an extended class
4105 that contained multibyte characters and which was followed by a quantifier.
4106
4107 15. Added optional support for general category Unicode character properties
4108 via the \p, \P, and \X escapes. Unicode property support implies UTF-8
4109 support. It adds about 90K to the size of the library. The meanings of the
4110 inbuilt class escapes such as \d and \s have NOT been changed.
4111
4112 16. Updated pcredemo.c to include calls to free() to release the memory for the
4113 compiled pattern.
4114
4115 17. The generated file chartables.c was being created in the source directory
4116 instead of in the building directory. This caused the build to fail if the
4117 source directory was different from the building directory, and was
4118 read-only.
4119
4120 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
4121 file. No doubt somebody will tell me if they don't make sense... Also added
4122 Dan Mooney's comments about building on OpenVMS.
4123
4124 19. Added support for partial matching via the PCRE_PARTIAL option for
4125 pcre_exec() and the \P data escape in pcretest.
4126
4127 20. Extended pcretest with 3 new pattern features:
4128
4129 (i) A pattern option of the form ">rest-of-line" causes pcretest to
4130 write the compiled pattern to the file whose name is "rest-of-line".
4131 This is a straight binary dump of the data, with the saved pointer to
4132 the character tables forced to be NULL. The study data, if any, is
4133 written too. After writing, pcretest reads a new pattern.
4134
4135 (ii) If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
4136 compiled pattern from the given file. There must not be any
4137 occurrences of "<" in the file name (pretty unlikely); if there are,
4138 pcretest will instead treat the initial "<" as a pattern delimiter.
4139 After reading in the pattern, pcretest goes on to read data lines as
4140 usual.
4141
4142 (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
4143 and 16-bit fields in a compiled pattern, to simulate a pattern that
4144 was compiled on a host of opposite endianness.
4145
4146 21. The pcre-exec() function can now cope with patterns that were compiled on
4147 hosts of opposite endianness, with this restriction:
4148
4149 As for any compiled expression that is saved and used later, the tables
4150 pointer field cannot be preserved; the extra_data field in the arguments
4151 to pcre_exec() should be used to pass in a tables address if a value
4152 other than the default internal tables were used at compile time.
4153
4154 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
4155 now diagnosed as an error. Previously, most of the time, a negative number
4156 would have been treated as zero, but if in addition "ovector" was passed as
4157 NULL, a crash could occur.
4158
4159 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
4160 new versions from the libtool 1.5 distribution (the last one is a copy of
4161 a file called libtool.m4). This seems to have fixed the need to patch
4162 "configure" to support Darwin 1.3 (which I used to do). However, I still
4163 had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
4164 workstation).
4165
4166 24. Changed the PCRE licence to be the more standard "BSD" licence.
4167
4168
4169 Version 4.5 01-Dec-03
4170 ---------------------
4171
4172 1. There has been some re-arrangement of the code for the match() function so
4173 that it can be compiled in a version that does not call itself recursively.
4174 Instead, it keeps those local variables that need separate instances for
4175 each "recursion" in a frame on the heap, and gets/frees frames whenever it
4176 needs to "recurse". Keeping track of where control must go is done by means
4177 of setjmp/longjmp. The whole thing is implemented by a set of macros that
4178 hide most of the details from the main code, and operates only if
4179 NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
4180 "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
4181 operating.
4182
4183 To make it easier for callers to provide specially tailored get/free
4184 functions for this usage, two new functions, pcre_stack_malloc, and
4185 pcre_stack_free, are used. They are always called in strict stacking order,
4186 and the size of block requested is always the same.
4187
4188 The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
4189 PCRE has been compiled to use the stack or the heap for recursion. The
4190 -C option of pcretest uses this to show which version is compiled.
4191
4192 A new data escape \S, is added to pcretest; it causes the amounts of store
4193 obtained and freed by both kinds of malloc/free at match time to be added
4194 to the output.
4195
4196 2. Changed the locale test to use "fr_FR" instead of "fr" because that's
4197 what's available on my current Linux desktop machine.
4198
4199 3. When matching a UTF-8 string, the test for a valid string at the start has
4200 been extended. If start_offset is not zero, PCRE now checks that it points
4201 to a byte that is the start of a UTF-8 character. If not, it returns
4202 PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
4203 this is necessary because there may be backward assertions in the pattern.
4204 When matching the same subject several times, it may save resources to use
4205 PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
4206
4207 4. The code for checking the validity of UTF-8 strings has been tightened so
4208 that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
4209 containing "overlong sequences".
4210
4211 5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
4212 I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
4213 should have been "&", but it just so happened that all the cases this let
4214 through by mistake were picked up later in the function.
4215
4216 6. I had used a variable called "isblank" - this is a C99 function, causing
4217 some compilers to warn. To avoid this, I renamed it (as "blankclass").
4218
4219 7. Cosmetic: (a) only output another newline at the end of pcretest if it is
4220 prompting; (b) run "./pcretest /dev/null" at the start of the test script
4221 so the version is shown; (c) stop "make test" echoing "./RunTest".
4222
4223 8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
4224
4225 9. The prototype for memmove() for systems that don't have it was using
4226 size_t, but the inclusion of the header that defines size_t was later. I've
4227 moved the #includes for the C headers earlier to avoid this.
4228
4229 10. Added some adjustments to the code to make it easier to compiler on certain
4230 special systems:
4231
4232 (a) Some "const" qualifiers were missing.
4233 (b) Added the macro EXPORT before all exported functions; by default this
4234 is defined to be empty.
4235 (c) Changed the dftables auxiliary program (that builds chartables.c) so
4236 that it reads its output file name as an argument instead of writing
4237 to the standard output and assuming this can be redirected.
4238
4239 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
4240 class containing characters with values greater than 255, PCRE compilation
4241 went into a loop.
4242
4243 12. A recursive reference to a subpattern that was within another subpattern
4244 that had a minimum quantifier of zero caused PCRE to crash. For example,
4245 (x(y(?2))z)? provoked this bug with a subject that got as far as the
4246 recursion. If the recursively-called subpattern itself had a zero repeat,
4247 that was OK.
4248
4249 13. In pcretest, the buffer for reading a data line was set at 30K, but the
4250 buffer into which it was copied (for escape processing) was still set at
4251 1024, so long lines caused crashes.
4252
4253 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
4254 "internal error: code overflow...". This applied to any character class
4255 that was followed by a possessive quantifier.
4256
4257 15. Modified the Makefile to add libpcre.la as a prerequisite for
4258 libpcreposix.la because I was told this is needed for a parallel build to
4259 work.
4260
4261 16. If a pattern that contained .* following optional items at the start was
4262 studied, the wrong optimizing data was generated, leading to matching
4263 errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
4264 matching string must start with a or b or c. The correct conclusion for
4265 this pattern is that a match can start with any character.
4266
4267
4268 Version 4.4 13-Aug-03
4269 ---------------------
4270
4271 1. In UTF-8 mode, a character class containing characters with values between
4272 127 and 255 was not handled correctly if the compiled pattern was studied.
4273 In fixing this, I have also improved the studying algorithm for such
4274 classes (slightly).
4275
4276 2. Three internal functions had redundant arguments passed to them. Removal
4277 might give a very teeny performance improvement.
4278
4279 3. Documentation bug: the value of the capture_top field in a callout is *one
4280 more than* the number of the hightest numbered captured substring.
4281
4282 4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
4283 in incorrectly linking with a previously installed version. They now link
4284 explicitly with libpcre.la.
4285
4286 5. configure.in no longer needs to recognize Cygwin specially.
4287
4288 6. A problem in pcre.in for Windows platforms is fixed.
4289
4290 7. If a pattern was successfully studied, and the -d (or /D) flag was given to
4291 pcretest, it used to include the size of the study block as part of its
4292 output. Unfortunately, the structure contains a field that has a different
4293 size on different hardware architectures. This meant that the tests that
4294 showed this size failed. As the block is currently always of a fixed size,
4295 this information isn't actually particularly useful in pcretest output, so
4296 I have just removed it.
4297
4298 8. Three pre-processor statements accidentally did not start in column 1.
4299 Sadly, there are *still* compilers around that complain, even though
4300 standard C has not required this for well over a decade. Sigh.
4301
4302 9. In pcretest, the code for checking callouts passed small integers in the
4303 callout_data field, which is a void * field. However, some picky compilers
4304 complained about the casts involved for this on 64-bit systems. Now
4305 pcretest passes the address of the small integer instead, which should get
4306 rid of the warnings.
4307
4308 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
4309 both compile and run time, and gives an error if an invalid UTF-8 sequence
4310 is found. There is a option for disabling this check in cases where the
4311 string is known to be correct and/or the maximum performance is wanted.
4312
4313 11. In response to a bug report, I changed one line in Makefile.in from
4314
4315 -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
4316 to
4317 -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
4318
4319 to look similar to other lines, but I have no way of telling whether this
4320 is the right thing to do, as I do not use Windows. No doubt I'll get told
4321 if it's wrong...
4322
4323
4324 Version 4.3 21-May-03
4325 ---------------------
4326
4327 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
4328 Makefile.
4329
4330 2. Some refactoring to improve the quality of the code:
4331
4332 (i) The utf8_table... variables are now declared "const".
4333
4334 (ii) The code for \cx, which used the "case flipping" table to upper case
4335 lower case letters, now just substracts 32. This is ASCII-specific,
4336 but the whole concept of \cx is ASCII-specific, so it seems
4337 reasonable.
4338
4339 (iii) PCRE was using its character types table to recognize decimal and
4340 hexadecimal digits in the pattern. This is silly, because it handles
4341 only 0-9, a-f, and A-F, but the character types table is locale-
4342 specific, which means strange things might happen. A private
4343 table is now used for this - though it costs 256 bytes, a table is
4344 much faster than multiple explicit tests. Of course, the standard
4345 character types table is still used for matching digits in subject
4346 strings against \d.
4347
4348 (iv) Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
4349 ending in _t are). So I've renamed it as ESC_tee.
4350
4351 3. The first argument for regexec() in the POSIX wrapper should have been
4352 defined as "const".
4353
4354 4. Changed pcretest to use malloc() for its buffers so that they can be
4355 Electric Fenced for debugging.
4356
4357 5. There were several places in the code where, in UTF-8 mode, PCRE would try
4358 to read one or more bytes before the start of the subject string. Often this
4359 had no effect on PCRE's behaviour, but in some circumstances it could
4360 provoke a segmentation fault.
4361
4362 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
4363 to try to read one or more bytes before the start of the subject string.
4364
4365 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
4366 UTF-8 support could misbehave in various ways if the subject string
4367 contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
4368 area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
4369 back over UTF-8 characters.)
4370
4371
4372 Version 4.2 14-Apr-03
4373 ---------------------
4374
4375 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
4376
4377 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
4378 [ON_WINDOWS]: new variable, "#" on non-Windows platforms
4379 [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
4380 [WIN_PREFIX]: new variable, "cyg" for Cygwin
4381 * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
4382 and BUILD_EXEEXT
4383 Note: automatic setting of the BUILD variables is not yet working
4384 set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
4385 compile-time but not at link-time
4386 [LINK]: use for linking executables only
4387 make different versions for Windows and non-Windows
4388 [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
4389 libraries
4390 [LINK_FOR_BUILD]: new variable
4391 [OBJEXT]: use throughout
4392 [EXEEXT]: use throughout
4393 <winshared>: new target
4394 <wininstall>: new target
4395 <dftables.o>: use native compiler
4396 <dftables>: use native linker
4397 <install>: handle Windows platform correctly
4398 <clean>: ditto
4399 <check>: ditto
4400 copy DLL to top builddir before testing
4401
4402 As part of these changes, -no-undefined was removed again. This was reported
4403 to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
4404 in any case.
4405
4406 3. Some tidies to get rid of compiler warnings:
4407
4408 . In the match_data structure, match_limit was an unsigned long int, whereas
4409 match_call_count was an int. I've made them both unsigned long ints.
4410
4411 . In pcretest the fact that a const uschar * doesn't automatically cast to
4412 a void * provoked a warning.
4413
4414 . Turning on some more compiler warnings threw up some "shadow" variables
4415 and a few more missing casts.
4416
4417 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
4418 option, a class that contained a single character with a value between 128
4419 and 255 (e.g. /[\xFF]/) caused PCRE to crash.
4420
4421 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
4422 option, a class that contained several characters, but with at least one
4423 whose value was between 128 and 255 caused PCRE to crash.
4424
4425
4426 Version 4.1 12-Mar-03
4427 ---------------------
4428
4429 1. Compiling with gcc -pedantic found a couple of places where casts were
4430 needed, and a string in dftables.c that was longer than standard compilers are
4431 required to support.
4432
4433 2. Compiling with Sun's compiler found a few more places where the code could
4434 be tidied up in order to avoid warnings.
4435
4436 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
4437 first of these names is deprecated in the latest Autoconf in favour of the name
4438 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
4439 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
4440 analogy I have changed it to CFLAGS_FOR_BUILD.
4441
4442 4. Added -no-undefined to the linking command in the Makefile, because this is
4443 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
4444 linking step for the pcreposix library.
4445
4446 5. PCRE was failing to diagnose the case of two named groups with the same
4447 name.
4448
4449 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
4450 literal character that is needed in the subject for a match, and scans along to
4451 ensure that it is present before embarking on the full matching process. This
4452 saves time in cases of nested unlimited repeats that are never going to match.
4453 Problem: the scan can take a lot of time if the subject is very long (e.g.
4454 megabytes), thus penalizing straightforward matches. It is now done only if the
4455 amount of subject to be scanned is less than 1000 bytes.
4456
4457 7. A lesser problem with the same optimization is that it was recording the
4458 first character of an anchored pattern as "needed", thus provoking a search
4459 right along the subject, even when the first match of the pattern was going to
4460 fail. The "needed" character is now not set for anchored patterns, unless it
4461 follows something in the pattern that is of non-fixed length. Thus, it still
4462 fulfils its original purpose of finding quick non-matches in cases of nested
4463 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
4464
4465
4466 Version 4.0 17-Feb-03
4467 ---------------------
4468
4469 1. If a comment in an extended regex that started immediately after a meta-item
4470 extended to the end of string, PCRE compiled incorrect data. This could lead to
4471 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
4472
4473 2. Moved to autoconf 2.53 and libtool 1.4.2.
4474
4475 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
4476 the special perltest8 script is no longer needed - all the tests can be run
4477 from a single perltest script.
4478
4479 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
4480 by \s. It has now been removed in PCRE. This means it isn't recognized as
4481 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
4482 class [:space:] *does* include VT, thereby creating a mess.
4483
4484 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
4485 space and tab.
4486
4487 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
4488 its new features into the main test script, reducing the number of scripts.
4489
4490 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
4491 were backward compatible, and made the (?i) apply to the whole pattern, as if
4492 /i were given. Now it behaves more logically, and applies the option setting
4493 only to what follows. PCRE has been changed to follow suit. However, if it
4494 finds options settings right at the start of the pattern, it extracts them into
4495 the global options, as before. Thus, they show up in the info data.
4496
4497 8. Added support for the \Q...\E escape sequence. Characters in between are
4498 treated as literals. This is slightly different from Perl in that $ and @ are
4499 also handled as literals inside the quotes. In Perl, they will cause variable
4500 interpolation. Note the following examples:
4501
4502 Pattern PCRE matches Perl matches
4503
4504 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
4505 \Qabc\$xyz\E abc\$xyz abc\$xyz
4506 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
4507
4508 For compatibility with Perl, \Q...\E sequences are recognized inside character
4509 classes as well as outside them.
4510
4511 9. Re-organized 3 code statements in pcretest to avoid "overflow in
4512 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
4513 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
4514 signed/unsigned warnings.
4515
4516 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
4517 option for pcretest, so I've replaced it by a simple function that does just
4518 that job.
4519
4520 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
4521 "pcregrep -".
4522
4523 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
4524 Java package. This provides some syntactic sugar for simple cases of what my
4525 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
4526 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
4527 item, you can use this simplified notation. Note that only makes sense with
4528 greedy quantifiers. Consequently, the use of the possessive quantifier forces
4529 greediness, whatever the setting of the PCRE_UNGREEDY option.
4530
4531 13. A change of greediness default within a pattern was not taking effect at
4532 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
4533 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
4534 was abstracted outside.
4535
4536 14. PCRE now supports the \G assertion. It is true when the current matching
4537 position is at the start point of the match. This differs from \A when the
4538 starting offset is non-zero. Used with the /g option of pcretest (or similar
4539 code), it works in the same way as it does for Perl's /g option. If all
4540 alternatives of a regex begin with \G, the expression is anchored to the start
4541 match position, and the "anchored" flag is set in the compiled expression.
4542
4543 15. Some bugs concerning the handling of certain option changes within patterns
4544 have been fixed. These applied to options other than (?ims). For example,
4545 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
4546 been the other way round. Some of this was related to change 7 above.
4547
4548 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
4549 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
4550 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
4551 POSIX classes only within a class (e.g. /[[:alpha:]]/).
4552
4553 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
4554 mode. Unlike ".", it always matches newline, whatever the setting of
4555 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
4556 assertions. Perl allows it, but it doesn't (in general) work because it can't
4557 calculate the length of the lookbehind. At least, that's the case for Perl
4558 5.8.0 - I've been told they are going to document that it doesn't work in
4559 future.
4560
4561 18. Added an error diagnosis for escapes that PCRE does not support: these are
4562 \L, \l, \N, \P, \p, \U, \u, and \X.
4563
4564 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
4565 reading past the end of the pattern in cases such as /[abcd/.
4566
4567 20. PCRE was getting more memory than necessary for patterns with classes that
4568 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
4569
4570 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
4571 compiling PCRE for use with Virtual Pascal.
4572
4573 22. Small fix to the Makefile to make it work properly if the build is done
4574 outside the source tree.
4575
4576 23. Added a new extension: a condition to go with recursion. If a conditional
4577 subpattern starts with (?(R) the "true" branch is used if recursion has
4578 happened, whereas the "false" branch is used only at the top level.
4579
4580 24. When there was a very long string of literal characters (over 255 bytes
4581 without UTF support, over 250 bytes with UTF support), the computation of how
4582 much memory was required could be incorrect, leading to segfaults or other
4583 strange effects.
4584
4585 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
4586 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
4587 there was a subsequent back reference to those brackets. This meant that, for
4588 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
4589 possible to check for precisely this case. All we can do is abandon the
4590 optimization if .* occurs inside capturing brackets when there are any back
4591 references whatsoever. (See below for a better fix that came later.)
4592
4593 26. The handling of the optimization for finding the first character of a
4594 non-anchored pattern, and for finding a character that is required later in the
4595 match were failing in some cases. This didn't break the matching; it just
4596 failed to optimize when it could. The way this is done has been re-implemented.
4597
4598 27. Fixed typo in error message for invalid (?R item (it said "(?p").
4599
4600 28. Added a new feature that provides some of the functionality that Perl
4601 provides with (?{...}). The facility is termed a "callout". The way it is done
4602 in PCRE is for the caller to provide an optional function, by setting
4603 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
4604 global variable. By default it is unset, which disables all calling out. To get
4605 the function called, the regex must include (?C) at appropriate points. This
4606 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
4607 This provides a means of identifying different callout points. When PCRE
4608 reaches such a point in the regex, if pcre_callout has been set, the external
4609 function is called. It is provided with data in a structure called
4610 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
4611 matching continues; if it returns a non-zero value, the match at the current
4612 point fails. However, backtracking will occur if possible. [This was changed
4613 later and other features added - see item 49 below.]
4614
4615 29. pcretest is upgraded to test the callout functionality. It provides a
4616 callout function that displays information. By default, it shows the start of
4617 the match and the current position in the text. There are some new data escapes
4618 to vary what happens:
4619
4620 \C+ in addition, show current contents of captured substrings
4621 \C- do not supply a callout function
4622 \C!n return 1 when callout number n is reached
4623 \C!n!m return 1 when callout number n is reached for the mth time
4624
4625 30. If pcregrep was called with the -l option and just a single file name, it
4626 output "<stdin>" if a match was found, instead of the file name.
4627
4628 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
4629 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
4630 pcre_exec(). This saves a malloc/free per call. The default value of
4631 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
4632 when configuring.
4633
4634 32. The default maximum size of a compiled pattern is 64K. There have been a
4635 few cases of people hitting this limit. The code now uses macros to handle the
4636 storing of links as offsets within the compiled pattern. It defaults to 2-byte
4637 links, but this can be changed to 3 or 4 bytes by --with-link-size when
4638 configuring. Tests 2 and 5 work only with 2-byte links because they output
4639 debugging information about compiled patterns.
4640
4641 33. Internal code re-arrangements:
4642
4643 (a) Moved the debugging function for printing out a compiled regex into
4644 its own source file (printint.c) and used #include to pull it into
4645 pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
4646 separate copies.
4647
4648 (b) Defined the list of op-code names for debugging as a macro in
4649 internal.h so that it is next to the definition of the opcodes.
4650
4651 (c) Defined a table of op-code lengths for simpler skipping along compiled
4652 code. This is again a macro in internal.h so that it is next to the
4653 definition of the opcodes.
4654
4655 34. Added support for recursive calls to individual subpatterns, along the
4656 lines of Robin Houston's patch (but implemented somewhat differently).
4657
4658 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
4659 allow it to read and process whole directories in Win32. This code was
4660 contributed by Lionel Fourquaux; it has not been tested by me.
4661
4662 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
4663 used to name a group. Names consist of alphanumerics and underscores, and must
4664 be unique. Back references use the syntax (?P=name) and recursive calls use
4665 (?P>name) which is a PCRE extension to the Python extension. Groups still have
4666 numbers. The function pcre_fullinfo() can be used after compilation to extract
4667 a name/number map. There are three relevant calls:
4668
4669 PCRE_INFO_NAMEENTRYSIZE yields the size of each entry in the map
4670 PCRE_INFO_NAMECOUNT yields the number of entries
4671 PCRE_INFO_NAMETABLE yields a pointer to the map.
4672
4673 The map is a vector of fixed-size entries. The size of each entry depends on
4674 the length of the longest name used. The first two bytes of each entry are the
4675 group number, most significant byte first. There follows the corresponding
4676 name, zero terminated. The names are in alphabetical order.
4677
4678 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
4679 case instead of 255. Making it the same both with and without UTF-8 support
4680 means that the same test output works with both.
4681
4682 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
4683 calling malloc() with a zero argument.
4684
4685 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
4686 optimization. I've improved things by keeping a bitmap of backreferences with
4687 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
4688 fact referenced, the optimization can be applied. It is unlikely that a
4689 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
4690 the match to follow \n) will appear inside brackets with a number greater than
4691 31, but if it does, any back reference > 31 suppresses the optimization.
4692
4693 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
4694 of disabling numbered capturing parentheses. Any opening parenthesis that is
4695 not followed by ? behaves as if it were followed by ?: but named parentheses
4696 can still be used for capturing (and they will acquire numbers in the usual
4697 way).
4698
4699 41. Redesigned the return codes from the match() function into yes/no/error so
4700 that errors can be passed back from deep inside the nested calls. A malloc
4701 failure while inside a recursive subpattern call now causes the
4702 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
4703
4704 42. It is now possible to set a limit on the number of times the match()
4705 function is called in a call to pcre_exec(). This facility makes it possible to
4706 limit the amount of recursion and backtracking, though not in a directly
4707 obvious way, because the match() function is used in a number of different
4708 circumstances. The count starts from zero for each position in the subject
4709 string (for non-anchored patterns). The default limit is, for compatibility, a
4710 large number, namely 10 000 000. You can change this in two ways:
4711
4712 (a) When configuring PCRE before making, you can use --with-match-limit=n
4713 to set a default value for the compiled library.
4714
4715 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
4716 a different value is set. See 45 below.
4717
4718 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
4719
4720 43. Added a new function pcre_config(int, void *) to enable run-time extraction
4721 of things that can be changed at compile time. The first argument specifies
4722 what is wanted and the second points to where the information is to be placed.
4723 The current list of available information is:
4724
4725 PCRE_CONFIG_UTF8
4726
4727 The output is an integer that is set to one if UTF-8 support is available;
4728 otherwise it is set to zero.
4729
4730 PCRE_CONFIG_NEWLINE
4731
4732 The output is an integer that it set to the value of the code that is used for
4733 newline. It is either LF (10) or CR (13).
4734
4735 PCRE_CONFIG_LINK_SIZE
4736
4737 The output is an integer that contains the number of bytes used for internal
4738 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
4739
4740 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
4741
4742 The output is an integer that contains the threshold above which the POSIX
4743 interface uses malloc() for output vectors. See item 31 above.
4744
4745 PCRE_CONFIG_MATCH_LIMIT
4746
4747 The output is an unsigned integer that contains the default limit of the number
4748 of match() calls in a pcre_exec() execution. See 42 above.
4749
4750 44. pcretest has been upgraded by the addition of the -C option. This causes it
4751 to extract all the available output from the new pcre_config() function, and to
4752 output it. The program then exits immediately.
4753
4754 45. A need has arisen to pass over additional data with calls to pcre_exec() in
4755 order to support additional features. One way would have been to define
4756 pcre_exec2() (for example) with extra arguments, but this would not have been
4757 extensible, and would also have required all calls to the original function to
4758 be mapped to the new one. Instead, I have chosen to extend the mechanism that
4759 is used for passing in "extra" data from pcre_study().
4760
4761 The pcre_extra structure is now exposed and defined in pcre.h. It currently
4762 contains the following fields:
4763
4764 flags a bitmap indicating which of the following fields are set
4765 study_data opaque data from pcre_study()
4766 match_limit a way of specifying a limit on match() calls for a specific
4767 call to pcre_exec()
4768 callout_data data for callouts (see 49 below)
4769
4770 The flag bits are also defined in pcre.h, and are
4771
4772 PCRE_EXTRA_STUDY_DATA
4773 PCRE_EXTRA_MATCH_LIMIT
4774 PCRE_EXTRA_CALLOUT_DATA
4775
4776 The pcre_study() function now returns one of these new pcre_extra blocks, with
4777 the actual study data pointed to by the study_data field, and the
4778 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
4779 before. That is, this change is entirely upwards-compatible and requires no
4780 change to existing code.
4781
4782 If you want to pass in additional data to pcre_exec(), you can either place it
4783 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
4784 block.
4785
4786 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
4787 data string contains the escape sequence \M, pcretest calls pcre_exec() several
4788 times with different match limits, until it finds the minimum value needed for
4789 pcre_exec() to complete. The value is then output. This can be instructive; for
4790 most simple matches the number is quite small, but for pathological cases it
4791 gets very large very quickly.
4792
4793 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
4794 returns the size of the data block pointed to by the study_data field in a
4795 pcre_extra block, that is, the value that was passed as the argument to
4796 pcre_malloc() when PCRE was getting memory in which to place the information
4797 created by pcre_study(). The fourth argument should point to a size_t variable.
4798 pcretest has been extended so that this information is shown after a successful
4799 pcre_study() call when information about the compiled regex is being displayed.
4800
4801 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
4802 because what follows is always an absolute path. (Later: it turns out that this
4803 is more than cosmetic for MinGW, because it doesn't like empty path
4804 components.)
4805
4806 49. Some changes have been made to the callout feature (see 28 above):
4807
4808 (i) A callout function now has three choices for what it returns:
4809
4810 0 => success, carry on matching
4811 > 0 => failure at this point, but backtrack if possible
4812 < 0 => serious error, return this value from pcre_exec()
4813
4814 Negative values should normally be chosen from the set of PCRE_ERROR_xxx
4815 values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
4816 "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
4817 use by callout functions. It will never be used by PCRE itself.
4818
4819 (ii) The pcre_extra structure (see 45 above) has a void * field called
4820 callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
4821 pcre_callout_block structure has a field of the same name. The contents of
4822 the field passed in the pcre_extra structure are passed to the callout
4823 function in the corresponding field in the callout block. This makes it
4824 easier to use the same callout-containing regex from multiple threads. For
4825 testing, the pcretest program has a new data escape
4826
4827 \C*n pass the number n (may be negative) as callout_data
4828
4829 If the callout function in pcretest receives a non-zero value as
4830 callout_data, it returns that value.
4831
4832 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
4833 there were some redundant $(CFLAGS) in commands that are now specified as
4834 $(LINK), which already includes $(CFLAGS).
4835
4836 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
4837 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
4838 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
4839 one-byte characters throughout. Note that case-insensitive matching applies
4840 only to characters whose values are less than 256. PCRE doesn't support the
4841 notion of cases for higher-valued characters.
4842
4843 (i) A character class whose characters are all within 0-255 is handled as
4844 a bit map, and the map is inverted for negative classes. Previously, a
4845 character > 255 always failed to match such a class; however it should
4846 match if the class was a negative one (e.g. [^ab]). This has been fixed.
4847
4848 (ii) A negated character class with a single character < 255 is coded as
4849 "not this character" (OP_NOT). This wasn't working properly when the test
4850 character was multibyte, either singly or repeated.
4851
4852 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
4853 mode, for example: \x{100}{2,3}.
4854
4855 (iv) The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
4856 singly or repeated) now correctly test multibyte characters. However,
4857 PCRE doesn't recognize any characters with values greater than 255 as
4858 digits, spaces, or word characters. Such characters always match \D, \S,
4859 and \W, and never match \d, \s, or \w.
4860
4861 (v) Classes may now contain characters and character ranges with values
4862 greater than 255. For example: [ab\x{100}-\x{400}].
4863
4864 (vi) pcregrep now has a --utf-8 option (synonym -u) which makes it call
4865 PCRE in UTF-8 mode.
4866
4867 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
4868 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
4869 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
4870 value.)
4871
4872 53. The single man page has become too large. I have therefore split it up into
4873 a number of separate man pages. These also give rise to individual HTML pages;
4874 these are now put in a separate directory, and there is an index.html page that
4875 lists them all. Some hyperlinking between the pages has been installed.
4876
4877 54. Added convenience functions for handling named capturing parentheses.
4878
4879 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
4880 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
4881 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
4882 are faulted.
4883
4884 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
4885 calling configure. These values are used when compiling the dftables.c program
4886 which is run to generate the source of the default character tables. They
4887 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
4888 you will need to set these values.
4889
4890 57. Updated the building process for Windows DLL, as provided by Fred Cox.
4891
4892
4893 Version 3.9 02-Jan-02
4894 ---------------------
4895
4896 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
4897
4898 2. If --disable-static was given, the building process failed when trying to
4899 build pcretest and pcregrep. (For some reason it was using libtool to compile
4900 them, which is not right, as they aren't part of the library.)
4901
4902
4903 Version 3.8 18-Dec-01
4904 ---------------------
4905
4906 1. The experimental UTF-8 code was completely screwed up. It was packing the
4907 bytes in the wrong order. How dumb can you get?
4908
4909
4910 Version 3.7 29-Oct-01
4911 ---------------------
4912
4913 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
4914 This caused pcretest, when used on the test data, to segfault. Unfortunately,
4915 this didn't happen under Solaris 8, where I normally test things.
4916
4917 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
4918 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
4919 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
4920 this fix an hour or so after the initial 3.7 release.)
4921
4922
4923 Version 3.6 23-Oct-01
4924 ---------------------
4925
4926 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
4927 offsets passed as NULL with zero offset count.
4928
4929 2. The config.guess and config.sub files had not been updated when I moved to
4930 the latest autoconf.
4931
4932
4933 Version 3.5 15-Aug-01
4934 ---------------------
4935
4936 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
4937 had been forgotten.
4938
4939 2. By using declared but undefined structures, we can avoid using "void"
4940 definitions in pcre.h while keeping the internal definitions of the structures
4941 private.
4942
4943 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
4944 user point of view, this means that both static and shared libraries are built
4945 by default, but this can be individually controlled. More of the work of
4946 handling this static/shared cases is now inside libtool instead of PCRE's make
4947 file.
4948
4949 4. The pcretest utility is now installed along with pcregrep because it is
4950 useful for users (to test regexs) and by doing this, it automatically gets
4951 relinked by libtool. The documentation has been turned into a man page, so
4952 there are now .1, .txt, and .html versions in /doc.
4953
4954 5. Upgrades to pcregrep:
4955 (i) Added long-form option names like gnu grep.
4956 (ii) Added --help to list all options with an explanatory phrase.
4957 (iii) Added -r, --recursive to recurse into sub-directories.
4958 (iv) Added -f, --file to read patterns from a file.
4959
4960 6. pcre_exec() was referring to its "code" argument before testing that
4961 argument for NULL (and giving an error if it was NULL).
4962
4963 7. Upgraded Makefile.in to allow for compiling in a different directory from
4964 the source directory.
4965
4966 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
4967 options bits, the pointer it was passed was to an int instead of to an unsigned
4968 long int. This mattered only on 64-bit systems.
4969
4970 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
4971 generated) instead of pcre.in, which it its source. Also made the same change
4972 in several of the .c files.
4973
4974 10. A new release of gcc defines printf() as a macro, which broke pcretest
4975 because it had an ifdef in the middle of a string argument for printf(). Fixed
4976 by using separate calls to printf().
4977
4978 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
4979 script, to force use of CR or LF instead of \n in the source. On non-Unix
4980 systems, the value can be set in config.h.
4981
4982 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
4983 absolute limit. Changed the text of the error message to make this clear, and
4984 likewise updated the man page.
4985
4986 13. The limit of 99 on the number of capturing subpatterns has been removed.
4987 The new limit is 65535, which I hope will not be a "real" limit.
4988
4989
4990 Version 3.4 22-Aug-00
4991 ---------------------
4992
4993 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
4994
4995 2. Diagnose condition (?(0) as an error instead of crashing on matching.
4996
4997
4998 Version 3.3 01-Aug-00
4999 ---------------------
5000
5001 1. If an octal character was given, but the value was greater than \377, it
5002 was not getting masked to the least significant bits, as documented. This could
5003 lead to crashes in some systems.
5004
5005 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
5006 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
5007
5008 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
5009 These just pass their arguments on to (pcre_free)(), but they are provided
5010 because some uses of PCRE bind it to non-C systems that can call its functions,
5011 but cannot call free() or pcre_free() directly.
5012
5013 4. Add "make test" as a synonym for "make check". Corrected some comments in
5014 the Makefile.
5015
5016 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
5017 Makefile.
5018
5019 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
5020 command called pgrep for grepping around the active processes.
5021
5022 7. Added the beginnings of support for UTF-8 character strings.
5023
5024 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
5025 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
5026 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
5027 out for the ar command.)
5028
5029
5030 Version 3.2 12-May-00
5031 ---------------------
5032
5033 This is purely a bug fixing release.
5034
5035 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
5036 of ZA. This was just one example of several cases that could provoke this bug,
5037 which was introduced by change 9 of version 2.00. The code for breaking
5038 infinite loops after an iteration that matches an empty string was't working
5039 correctly.
5040
5041 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
5042 when matched against abbab (for example). After matching an empty string, it
5043 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
5044 caused it to match further down the string than it should.
5045
5046 3. The code contained an inclusion of sys/types.h. It isn't clear why this
5047 was there because it doesn't seem to be needed, and it causes trouble on some
5048 systems, as it is not a Standard C header. It has been removed.
5049
5050 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
5051 were reported on the Macintosh. The changes were from
5052
5053 while ((c = *(++ptr)) != 0 && c != '\n');
5054 to
5055 while ((c = *(++ptr)) != 0 && c != '\n') ;
5056
5057 Totally extraordinary, but if that's what it takes...
5058
5059 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
5060 available. Added HAVE_BCOPY and an autoconf test for it; if neither
5061 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
5062 assumes the way PCRE uses memmove() (always moving upwards).
5063
5064 6. PCRE is being used in one environment where strchr() is not available. There
5065 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
5066 faster code anyway.
5067
5068
5069 Version 3.1 09-Feb-00
5070 ---------------------
5071
5072 The only change in this release is the fixing of some bugs in Makefile.in for
5073 the "install" target:
5074
5075 (1) It was failing to install pcreposix.h.
5076
5077 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
5078
5079
5080 Version 3.0 01-Feb-00
5081 ---------------------
5082
5083 1. Add support for the /+ modifier to perltest (to output $` like it does in
5084 pcretest).
5085
5086 2. Add support for the /g modifier to perltest.
5087
5088 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
5089 matches null strings.
5090
5091 4. Fix perltest so that it doesn't do unwanted things when fed an empty
5092 pattern. Perl treats empty patterns specially - it reuses the most recent
5093 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
5094 effect.
5095
5096 5. The POSIX interface was broken in that it was just handing over the POSIX
5097 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
5098 required a bigger vector, with some working space on the end. This means that
5099 the POSIX wrapper now has to get and free some memory, and copy the results.
5100
5101 6. Added some simple autoconf support, placing the test data and the
5102 documentation in separate directories, re-organizing some of the
5103 information files, and making it build pcre-config (a GNU standard). Also added
5104 libtool support for building PCRE as a shared library, which is now the
5105 default.
5106
5107 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
5108 09 are not valid octal constants. Single digits will be used for minor values
5109 less than 10.
5110
5111 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
5112 existing programs that set these in the POSIX interface can use PCRE without
5113 modification.
5114
5115 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
5116 return all that pcre_info() returns, plus additional data. The pcre_info()
5117 function is retained for compatibility, but is considered to be obsolete.
5118
5119 10. Added experimental recursion feature (?R) to handle one common case that
5120 Perl 5.6 will be able to do with (?p{...}).
5121
5122 11. Added support for POSIX character classes like [:alpha:], which Perl is
5123 adopting.
5124
5125
5126 Version 2.08 31-Aug-99
5127 ----------------------
5128
5129 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
5130 trying to match at the startoffset position, but instead was moving forward to
5131 the next newline as if a previous match had failed.
5132
5133 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
5134 and could get into a loop if a null string was matched other than at the start
5135 of the subject.
5136
5137 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
5138 be distinguished at compile time, and for completeness also added PCRE_DATE.
5139
5140 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
5141 in GnuWin32 environments.
5142
5143
5144 Version 2.07 29-Jul-99
5145 ----------------------
5146
5147 1. The documentation is now supplied in plain text form and HTML as well as in
5148 the form of man page sources.
5149
5150 2. C++ compilers don't like assigning (void *) values to other pointer types.
5151 In particular this affects malloc(). Although there is no problem in Standard
5152 C, I've put in casts to keep C++ compilers happy.
5153
5154 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
5155 should be (const char *).
5156
5157 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
5158 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
5159 However, I haven't made this a standard facility. The documentation doesn't
5160 mention it, and the Makefile doesn't support it.
5161
5162 5. The Makefile now contains an "install" target, with editable destinations at
5163 the top of the file. The pcretest program is not installed.
5164
5165 6. pgrep -V now gives the PCRE version number and date.
5166
5167 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
5168 causing the entire string to be ignored, instead of just the last character.
5169
5170 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
5171 non-matching string, it can take a very, very long time, even for strings of
5172 quite modest length, because of the nested recursion. PCRE now does better in
5173 some of these cases. It does this by remembering the last required literal
5174 character in the pattern, and pre-searching the subject to ensure it is present
5175 before running the real match. In other words, it applies a heuristic to detect
5176 some types of certain failure quickly, and in the above example, if presented
5177 with a string that has no trailing " it gives "no match" very quickly.
5178
5179 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
5180 other alternatives are tried instead.
5181
5182
5183 Version 2.06 09-Jun-99
5184 ----------------------
5185
5186 1. Change pcretest's output for amount of store used to show just the code
5187 space, because the remainder (the data block) varies in size between 32-bit and
5188 64-bit systems.
5189
5190 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
5191 start matching at. This allows lookbehinds to work when searching for multiple
5192 occurrences in a string.
5193
5194 3. Added additional options to pcretest for testing multiple occurrences:
5195
5196 /+ outputs the rest of the string that follows a match
5197 /g loops for multiple occurrences, using the new startoffset argument
5198 /G loops for multiple occurrences by passing an incremented pointer
5199
5200 4. PCRE wasn't doing the "first character" optimization for patterns starting
5201 with \b or \B, though it was doing it for other lookbehind assertions. That is,
5202 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
5203 the letter 'x'. On long subject strings, this gives a significant speed-up.
5204
5205
5206 Version 2.05 21-Apr-99
5207 ----------------------
5208
5209 1. Changed the type of magic_number from int to long int so that it works
5210 properly on 16-bit systems.
5211
5212 2. Fixed a bug which caused patterns starting with .* not to work correctly
5213 when the subject string contained newline characters. PCRE was assuming
5214 anchoring for such patterns in all cases, which is not correct because .* will
5215 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
5216 DOTALL is set at top level; otherwise it knows that patterns starting with .*
5217 must be retried after every newline in the subject.
5218
5219
5220 Version 2.04 18-Feb-99
5221 ----------------------
5222
5223 1. For parenthesized subpatterns with repeats whose minimum was zero, the
5224 computation of the store needed to hold the pattern was incorrect (too large).
5225 If such patterns were nested a few deep, this could multiply and become a real
5226 problem.
5227
5228 2. Added /M option to pcretest to show the memory requirement of a specific
5229 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
5230
5231 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
5232 compiled in such a way that the backtracking after subsequent failure was
5233 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
5234 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
5235
5236
5237 Version 2.03 02-Feb-99
5238 ----------------------
5239
5240 1. Fixed typo and small mistake in man page.
5241
5242 2. Added 4th condition (GPL supersedes if conflict) and created separate
5243 LICENCE file containing the conditions.
5244
5245 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
5246 Perl, that is the internal \ allows the delimiter to be included in the
5247 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
5248 the final delimiter, add \ to the end of the pattern (to test the error).
5249
5250 4. Added the convenience functions for extracting substrings after a successful
5251 match. Updated pcretest to make it able to test these functions.
5252
5253
5254 Version 2.02 14-Jan-99
5255 ----------------------
5256
5257 1. Initialized the working variables associated with each extraction so that
5258 their saving and restoring doesn't refer to uninitialized store.
5259
5260 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
5261 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
5262 fix the problem.
5263
5264 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
5265 calls, and wasn't printing the correct value for compiling calls. Increased the
5266 default value of LOOPREPEAT, and the number of significant figures in the
5267 times.
5268
5269 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
5270
5271 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
5272 a building problem on Windows NT with a FAT file system.
5273
5274
5275 Version 2.01 21-Oct-98
5276 ----------------------
5277
5278 1. Changed the API for pcre_compile() to allow for the provision of a pointer
5279 to character tables built by pcre_maketables() in the current locale. If NULL
5280 is passed, the default tables are used.
5281
5282
5283 Version 2.00 24-Sep-98
5284 ----------------------
5285
5286 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
5287 it any more.
5288
5289 2. Allow quantification of (?>) groups, and make it work correctly.
5290
5291 3. The first character computation wasn't working for (?>) groups.
5292
5293 4. Correct the implementation of \Z (it is permitted to match on the \n at the
5294 end of the subject) and add 5.005's \z, which really does match only at the
5295 very end of the subject.
5296
5297 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
5298
5299 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
5300 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
5301 localized options. All options to pcre_study() were also removed.
5302
5303 7. Add other new features from 5.005:
5304
5305 $(?<= positive lookbehind
5306 $(?<! negative lookbehind
5307 (?imsx-imsx) added the unsetting capability
5308 such a setting is global if at outer level; local otherwise
5309 (?imsx-imsx:) non-capturing groups with option setting
5310 (?(cond)re|re) conditional pattern matching
5311
5312 A backreference to itself in a repeated group matches the previous
5313 captured string.
5314
5315 8. General tidying up of studying (both automatic and via "study")
5316 consequential on the addition of new assertions.
5317
5318 9. As in 5.005, unlimited repeated groups that could match an empty substring
5319 are no longer faulted at compile time. Instead, the loop is forcibly broken at
5320 runtime if any iteration does actually match an empty substring.
5321
5322 10. Include the RunTest script in the distribution.
5323
5324 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
5325 discrepancies, some of which were old and were also with respect to 5.004. They
5326 have now been fixed.
5327
5328
5329 Version 1.09 28-Apr-98
5330 ----------------------
5331
5332 1. A negated single character class followed by a quantifier with a minimum
5333 value of one (e.g. [^x]{1,6} ) was not compiled correctly. This could lead to
5334 program crashes, or just wrong answers. This did not apply to negated classes
5335 containing more than one character, or to minima other than one.
5336
5337
5338 Version 1.08 27-Mar-98
5339 ----------------------
5340
5341 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
5342
5343 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
5344 latter must appear before anything that relies on it in the pattern.
5345
5346
5347 Version 1.07 16-Feb-98
5348 ----------------------
5349
5350 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
5351 repeat of a potentially empty string).
5352
5353
5354 Version 1.06 23-Jan-98
5355 ----------------------
5356
5357 1. Added Markus Oberhumer's little patches for C++.
5358
5359 2. Literal strings longer than 255 characters were broken.
5360
5361
5362 Version 1.05 23-Dec-97
5363 ----------------------
5364
5365 1. Negated character classes containing more than one character were failing if
5366 PCRE_CASELESS was set at run time.
5367
5368
5369 Version 1.04 19-Dec-97
5370 ----------------------
5371
5372 1. Corrected the man page, where some "const" qualifiers had been omitted.
5373
5374 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
5375 input syntax.
5376
5377 3. Fixed memory leak which occurred when a regex with back references was
5378 matched with an offsets vector that wasn't big enough. The temporary memory
5379 that is used in this case wasn't being freed if the match failed.
5380
5381 4. Tidied pcretest to ensure it frees memory that it gets.
5382
5383 5. Temporary memory was being obtained in the case where the passed offsets
5384 vector was exactly big enough.
5385
5386 6. Corrected definition of offsetof() from change 5 below.
5387
5388 7. I had screwed up change 6 below and broken the rules for the use of
5389 setjmp(). Now fixed.
5390
5391
5392 Version 1.03 18-Dec-97
5393 ----------------------
5394
5395 1. A erroneous regex with a missing opening parenthesis was correctly
5396 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
5397 on some systems.
5398
5399 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
5400 it was reported that one broken compiler failed on the former because "code" is
5401 also an independent variable.
5402
5403 3. The erroneous regex a[]b caused an array overrun reference.
5404
5405 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
5406 fail on data ending with that character. (It was going on too far, and checking
5407 the next character, typically a binary zero.) This was specific to the
5408 optimized code for single-character negative classes.
5409
5410 5. Added a contributed patch from the TIN world which does the following:
5411
5412 + Add an undef for memmove, in case the the system defines a macro for it.
5413
5414 + Add a definition of offsetof(), in case there isn't one. (I don't know
5415 the reason behind this - offsetof() is part of the ANSI standard - but
5416 it does no harm).
5417
5418 + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
5419 most of the places where whitespace preceded '#'. I have given up and
5420 allowed the remaining 2 cases to be at the margin.
5421
5422 + Rename some variables in pcre to eliminate shadowing. This seems very
5423 pedantic, but does no harm, of course.
5424
5425 6. Moved the call to setjmp() into its own function, to get rid of warnings
5426 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
5427
5428 7. Constructs such as \d{8,} were compiling into the equivalent of
5429 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
5430 outcome, but in this particular case used more store than had been allocated,
5431 which caused the bug to be discovered because it threw up an internal error.
5432
5433 8. The debugging code in both pcre and pcretest for outputting the compiled
5434 form of a regex was going wrong in the case of back references followed by
5435 curly-bracketed repeats.
5436
5437
5438 Version 1.02 12-Dec-97
5439 ----------------------
5440
5441 1. Typos in pcre.3 and comments in the source fixed.
5442
5443 2. Applied a contributed patch to get rid of places where it used to remove
5444 'const' from variables, and fixed some signed/unsigned and uninitialized
5445 variable warnings.
5446
5447 3. Added the "runtest" target to Makefile.
5448
5449 4. Set default compiler flag to -O2 rather than just -O.
5450
5451
5452 Version 1.01 19-Nov-97
5453 ----------------------
5454
5455 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
5456 like /([ab]*)*/, that is, for classes with more than one character in them.
5457
5458 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
5459 as /((?>a*))*/ (a PCRE_EXTRA facility).
5460
5461
5462 Version 1.00 18-Nov-97
5463 ----------------------
5464
5465 1. Added compile-time macros to support systems such as SunOS4 which don't have
5466 memmove() or strerror() but have other things that can be used instead.
5467
5468 2. Arranged that "make clean" removes the executables.
5469
5470
5471 Version 0.99 27-Oct-97
5472 ----------------------
5473
5474 1. Fixed bug in code for optimizing classes with only one character. It was
5475 initializing a 32-byte map regardless, which could cause it to run off the end
5476 of the memory it had got.
5477
5478 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
5479
5480
5481 Version 0.98 22-Oct-97
5482 ----------------------
5483
5484 1. Fixed bug in code for handling temporary memory usage when there are more
5485 back references than supplied space in the ovector. This could cause segfaults.
5486
5487
5488 Version 0.97 21-Oct-97
5489 ----------------------
5490
5491 1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
5492
5493 2. Optimized negated single characters not to use a bit map.
5494
5495 3. Brought error texts together as macro definitions; clarified some of them;
5496 fixed one that was wrong - it said "range out of order" when it meant "invalid
5497 escape sequence".
5498
5499 4. Changed some char * arguments to const char *.
5500
5501 5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
5502
5503 6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
5504 pcretest.
5505
5506
5507 Version 0.96 16-Oct-97
5508 ----------------------
5509
5510 1. Added a simple "pgrep" utility to the distribution.
5511
5512 2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
5513 unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
5514 where "ddd" means "one or more decimal digits".
5515
5516 3. Fixed serious bug. If a pattern had a back reference, but the call to
5517 pcre_exec() didn't supply a large enough ovector to record the related
5518 identifying subpattern, the match always failed. PCRE now remembers the number
5519 of the largest back reference, and gets some temporary memory in which to save
5520 the offsets during matching if necessary, in order to ensure that
5521 backreferences always work.
5522
5523 4. Increased the compatibility with Perl in a number of ways:
5524
5525 (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
5526 to request this handling. The option can be set at compile or exec time.
5527
5528 (b) $ matches before a terminating newline by default; an option
5529 PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
5530 mode). The option can be set at compile or exec time.
5531
5532 (c) The handling of \ followed by a digit other than 0 is now supposed to be
5533 the same as Perl's. If the decimal number it represents is less than 10
5534 or there aren't that many previous left capturing parentheses, an octal
5535 escape is read. Inside a character class, it's always an octal escape,
5536 even if it is a single digit.
5537
5538 (d) An escaped but undefined alphabetic character is taken as a literal,
5539 unless PCRE_EXTRA is set. Currently this just reserves the remaining
5540 escapes.
5541
5542 (e) {0} is now permitted. (The previous item is removed from the compiled
5543 pattern).
5544
5545 5. Changed all the names of code files so that the basic parts are no longer
5546 than 10 characters, and abolished the teeny "globals.c" file.
5547
5548 6. Changed the handling of character classes; they are now done with a 32-byte
5549 bit map always.
5550
5551 7. Added the -d and /D options to pcretest to make it possible to look at the
5552 internals of compilation without having to recompile pcre.
5553
5554
5555 Version 0.95 23-Sep-97
5556 ----------------------
5557
5558 1. Fixed bug in pre-pass concerning escaped "normal" characters such as \x5c or
5559 \x20 at the start of a run of normal characters. These were being treated as
5560 real characters, instead of the source characters being re-checked.
5561
5562
5563 Version 0.94 18-Sep-97
5564 ----------------------
5565
5566 1. The functions are now thread-safe, with the caveat that the global variables
5567 containing pointers to malloc() and free() or alternative functions are the
5568 same for all threads.
5569
5570 2. Get pcre_study() to generate a bitmap of initial characters for non-
5571 anchored patterns when this is possible, and use it if passed to pcre_exec().
5572
5573
5574 Version 0.93 15-Sep-97
5575 ----------------------
5576
5577 1. /(b)|(:+)/ was computing an incorrect first character.
5578
5579 2. Add pcre_study() to the API and the passing of pcre_extra to pcre_exec(),
5580 but not actually doing anything yet.
5581
5582 3. Treat "-" characters in classes that cannot be part of ranges as literals,
5583 as Perl does (e.g. [-az] or [az-]).
5584
5585 4. Set the anchored flag if a branch starts with .* or .*? because that tests
5586 all possible positions.
5587
5588 5. Split up into different modules to avoid including unneeded functions in a
5589 compiled binary. However, compile and exec are still in one module. The "study"
5590 function is split off.
5591
5592 6. The character tables are now in a separate module whose source is generated
5593 by an auxiliary program - but can then be edited by hand if required. There are
5594 now no calls to isalnum(), isspace(), isdigit(), isxdigit(), tolower() or
5595 toupper() in the code.
5596
5597 7. Turn the malloc/free funtions variables into pcre_malloc and pcre_free and
5598 make them global. Abolish the function for setting them, as the caller can now
5599 set them directly.
5600
5601
5602 Version 0.92 11-Sep-97
5603 ----------------------
5604
5605 1. A repeat with a fixed maximum and a minimum of 1 for an ordinary character
5606 (e.g. /a{1,3}/) was broken (I mis-optimized it).
5607
5608 2. Caseless matching was not working in character classes if the characters in
5609 the pattern were in upper case.
5610
5611 3. Make ranges like [W-c] work in the same way as Perl for caseless matching.
5612
5613 4. Make PCRE_ANCHORED public and accept as a compile option.
5614
5615 5. Add an options word to pcre_exec() and accept PCRE_ANCHORED and
5616 PCRE_CASELESS at run time. Add escapes \A and \I to pcretest to cause it to
5617 pass them.
5618
5619 6. Give an error if bad option bits passed at compile or run time.
5620
5621 7. Add PCRE_MULTILINE at compile and exec time, and (?m) as well. Add \M to
5622 pcretest to cause it to pass that flag.
5623
5624 8. Add pcre_info(), to get the number of identifying subpatterns, the stored
5625 options, and the first character, if set.
5626
5627 9. Recognize C+ or C{n,m} where n >= 1 as providing a fixed starting character.
5628
5629
5630 Version 0.91 10-Sep-97
5631 ----------------------
5632
5633 1. PCRE was failing to diagnose unlimited repeats of subpatterns that could
5634 match the empty string as in /(a*)*/. It was looping and ultimately crashing.
5635
5636 2. PCRE was looping on encountering an indefinitely repeated back reference to
5637 a subpattern that had matched an empty string, e.g. /(a|)\1*/. It now does what
5638 Perl does - treats the match as successful.
5639
5640 ****

Properties

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

  ViewVC Help
Powered by ViewVC 1.1.5