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

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1530 - (show annotations)
Thu Mar 5 08:53:37 2015 UTC (4 years, 7 months ago) by zherczeg
File size: 273427 byte(s)
Error occurred while calculating annotation data.
Remove computing the JIT read-only data size in advance and use on-demand memory allocation.
1 ChangeLog for PCRE
2 ------------------
3
4 Version 8.37 xx-xxx-2015
5 ------------------------
6
7 1. When an (*ACCEPT) is triggered inside capturing parentheses, it arranges
8 for those parentheses to be closed with whatever has been captured so far.
9 However, it was failing to mark any other groups between the hightest
10 capture so far and the currrent group as "unset". Thus, the ovector for
11 those groups contained whatever was previously there. An example is the
12 pattern /(x)|((*ACCEPT))/ when matched against "abcd".
13
14 2. If an assertion condition was quantified with a minimum of zero (an odd
15 thing to do, but it happened), SIGSEGV or other misbehaviour could occur.
16
17 3. If a pattern in pcretest input had the P (POSIX) modifier followed by an
18 unrecognized modifier, a crash could occur.
19
20 4. An attempt to do global matching in pcretest with a zero-length ovector
21 caused a crash.
22
23 5. Fixed a memory leak during matching that could occur for a subpattern
24 subroutine call (recursive or otherwise) if the number of captured groups
25 that had to be saved was greater than ten.
26
27 6. Catch a bad opcode during auto-possessification after compiling a bad UTF
28 string with NO_UTF_CHECK. This is a tidyup, not a bug fix, as passing bad
29 UTF with NO_UTF_CHECK is documented as having an undefined outcome.
30
31 7. A UTF pattern containing a "not" match of a non-ASCII character and a
32 subroutine reference could loop at compile time. Example: /[^\xff]((?1))/.
33
34 8. When a pattern is compiled, it remembers the highest back reference so that
35 when matching, if the ovector is too small, extra memory can be obtained to
36 use instead. A conditional subpattern whose condition is a check on a
37 capture having happened, such as, for example in the pattern
38 /^(?:(a)|b)(?(1)A|B)/, is another kind of back reference, but it was not
39 setting the highest backreference number. This mattered only if pcre_exec()
40 was called with an ovector that was too small to hold the capture, and there
41 was no other kind of back reference (a situation which is probably quite
42 rare). The effect of the bug was that the condition was always treated as
43 FALSE when the capture could not be consulted, leading to a incorrect
44 behaviour by pcre2_match(). This bug has been fixed.
45
46 9. A reference to a duplicated named group (either a back reference or a test
47 for being set in a conditional) that occurred in a part of the pattern where
48 PCRE_DUPNAMES was not set caused the amount of memory needed for the pattern
49 to be incorrectly calculated, leading to overwriting.
50
51 10. A mutually recursive set of back references such as (\2)(\1) caused a
52 segfault at study time (while trying to find the minimum matching length).
53 The infinite loop is now broken (with the minimum length unset, that is,
54 zero).
55
56 11. If an assertion that was used as a condition was quantified with a minimum
57 of zero, matching went wrong. In particular, if the whole group had
58 unlimited repetition and could match an empty string, a segfault was
59 likely. The pattern (?(?=0)?)+ is an example that caused this. Perl allows
60 assertions to be quantified, but not if they are being used as conditions,
61 so the above pattern is faulted by Perl. PCRE has now been changed so that
62 it also rejects such patterns.
63
64 12. A possessive capturing group such as (a)*+ with a minimum repeat of zero
65 failed to allow the zero-repeat case if pcre2_exec() was called with an
66 ovector too small to capture the group.
67
68 13. Fixed two bugs in pcretest that were discovered by fuzzing and reported by
69 Red Hat Product Security:
70
71 (a) A crash if /K and /F were both set with the option to save the compiled
72 pattern.
73
74 (b) Another crash if the option to print captured substrings in a callout
75 was combined with setting a null ovector, for example \O\C+ as a subject
76 string.
77
78 14. A pattern such as "((?2){0,1999}())?", which has a group containing a
79 forward reference repeated a large (but limited) number of times within a
80 repeated outer group that has a zero minimum quantifier, caused incorrect
81 code to be compiled, leading to the error "internal error:
82 previously-checked referenced subpattern not found" when an incorrect
83 memory address was read. This bug was reported as "heap overflow",
84 discovered by Kai Lu of Fortinet's FortiGuard Labs.
85
86 23. A pattern such as "((?+1)(\1))/" containing a forward reference subroutine
87 call within a group that also contained a recursive back reference caused
88 incorrect code to be compiled. This bug was reported as "heap overflow",
89 discovered by Kai Lu of Fortinet's FortiGuard Labs.
90
91 24. Computing the size of the JIT read-only data in advance has been a source
92 of various issues, and new ones are still appear unfortunately. To fix
93 existing and future issues, size computation is eliminated from the code,
94 and replaced by on-demand memory allocation.
95
96
97 Version 8.36 26-September-2014
98 ------------------------------
99
100 1. Got rid of some compiler warnings in the C++ modules that were shown up by
101 -Wmissing-field-initializers and -Wunused-parameter.
102
103 2. The tests for quantifiers being too big (greater than 65535) were being
104 applied after reading the number, and stupidly assuming that integer
105 overflow would give a negative number. The tests are now applied as the
106 numbers are read.
107
108 3. Tidy code in pcre_exec.c where two branches that used to be different are
109 now the same.
110
111 4. The JIT compiler did not generate match limit checks for certain
112 bracketed expressions with quantifiers. This may lead to exponential
113 backtracking, instead of returning with PCRE_ERROR_MATCHLIMIT. This
114 issue should be resolved now.
115
116 5. Fixed an issue, which occures when nested alternatives are optimized
117 with table jumps.
118
119 6. Inserted two casts and changed some ints to size_t in the light of some
120 reported 64-bit compiler warnings (Bugzilla 1477).
121
122 7. Fixed a bug concerned with zero-minimum possessive groups that could match
123 an empty string, which sometimes were behaving incorrectly in the
124 interpreter (though correctly in the JIT matcher). This pcretest input is
125 an example:
126
127 '\A(?:[^"]++|"(?:[^"]*+|"")*+")++'
128 NON QUOTED "QUOT""ED" AFTER "NOT MATCHED
129
130 the interpreter was reporting a match of 'NON QUOTED ' only, whereas the
131 JIT matcher and Perl both matched 'NON QUOTED "QUOT""ED" AFTER '. The test
132 for an empty string was breaking the inner loop and carrying on at a lower
133 level, when possessive repeated groups should always return to a higher
134 level as they have no backtrack points in them. The empty string test now
135 occurs at the outer level.
136
137 8. Fixed a bug that was incorrectly auto-possessifying \w+ in the pattern
138 ^\w+(?>\s*)(?<=\w) which caused it not to match "test test".
139
140 9. Give a compile-time error for \o{} (as Perl does) and for \x{} (which Perl
141 doesn't).
142
143 10. Change 8.34/15 introduced a bug that caused the amount of memory needed
144 to hold a pattern to be incorrectly computed (too small) when there were
145 named back references to duplicated names. This could cause "internal
146 error: code overflow" or "double free or corruption" or other memory
147 handling errors.
148
149 11. When named subpatterns had the same prefixes, back references could be
150 confused. For example, in this pattern:
151
152 /(?P<Name>a)?(?P<Name2>b)?(?(<Name>)c|d)*l/
153
154 the reference to 'Name' was incorrectly treated as a reference to a
155 duplicate name.
156
157 12. A pattern such as /^s?c/mi8 where the optional character has more than
158 one "other case" was incorrectly compiled such that it would only try to
159 match starting at "c".
160
161 13. When a pattern starting with \s was studied, VT was not included in the
162 list of possible starting characters; this should have been part of the
163 8.34/18 patch.
164
165 14. If a character class started [\Qx]... where x is any character, the class
166 was incorrectly terminated at the ].
167
168 15. If a pattern that started with a caseless match for a character with more
169 than one "other case" was studied, PCRE did not set up the starting code
170 unit bit map for the list of possible characters. Now it does. This is an
171 optimization improvement, not a bug fix.
172
173 16. The Unicode data tables have been updated to Unicode 7.0.0.
174
175 17. Fixed a number of memory leaks in pcregrep.
176
177 18. Avoid a compiler warning (from some compilers) for a function call with
178 a cast that removes "const" from an lvalue by using an intermediate
179 variable (to which the compiler does not object).
180
181 19. Incorrect code was compiled if a group that contained an internal recursive
182 back reference was optional (had quantifier with a minimum of zero). This
183 example compiled incorrect code: /(((a\2)|(a*)\g<-1>))*/ and other examples
184 caused segmentation faults because of stack overflows at compile time.
185
186 20. A pattern such as /((?(R)a|(?1)))+/, which contains a recursion within a
187 group that is quantified with an indefinite repeat, caused a compile-time
188 loop which used up all the system stack and provoked a segmentation fault.
189 This was not the same bug as 19 above.
190
191 21. Add PCRECPP_EXP_DECL declaration to operator<< in pcre_stringpiece.h.
192 Patch by Mike Frysinger.
193
194
195 Version 8.35 04-April-2014
196 --------------------------
197
198 1. A new flag is set, when property checks are present in an XCLASS.
199 When this flag is not set, PCRE can perform certain optimizations
200 such as studying these XCLASS-es.
201
202 2. The auto-possessification of character sets were improved: a normal
203 and an extended character set can be compared now. Furthermore
204 the JIT compiler optimizes more character set checks.
205
206 3. Got rid of some compiler warnings for potentially uninitialized variables
207 that show up only when compiled with -O2.
208
209 4. A pattern such as (?=ab\K) that uses \K in an assertion can set the start
210 of a match later then the end of the match. The pcretest program was not
211 handling the case sensibly - it was outputting from the start to the next
212 binary zero. It now reports this situation in a message, and outputs the
213 text from the end to the start.
214
215 5. Fast forward search is improved in JIT. Instead of the first three
216 characters, any three characters with fixed position can be searched.
217 Search order: first, last, middle.
218
219 6. Improve character range checks in JIT. Characters are read by an inprecise
220 function now, which returns with an unknown value if the character code is
221 above a certain threshold (e.g: 256). The only limitation is that the value
222 must be bigger than the threshold as well. This function is useful when
223 the characters above the threshold are handled in the same way.
224
225 7. The macros whose names start with RAWUCHAR are placeholders for a future
226 mode in which only the bottom 21 bits of 32-bit data items are used. To
227 make this more memorable for those maintaining the code, the names have
228 been changed to start with UCHAR21, and an extensive comment has been added
229 to their definition.
230
231 8. Add missing (new) files sljitNativeTILEGX.c and sljitNativeTILEGX-encoder.c
232 to the export list in Makefile.am (they were accidentally omitted from the
233 8.34 tarball).
234
235 9. The informational output from pcretest used the phrase "starting byte set"
236 which is inappropriate for the 16-bit and 32-bit libraries. As the output
237 for "first char" and "need char" really means "non-UTF-char", I've changed
238 "byte" to "char", and slightly reworded the output. The documentation about
239 these values has also been (I hope) clarified.
240
241 10. Another JIT related optimization: use table jumps for selecting the correct
242 backtracking path, when more than four alternatives are present inside a
243 bracket.
244
245 11. Empty match is not possible, when the minimum length is greater than zero,
246 and there is no \K in the pattern. JIT should avoid empty match checks in
247 such cases.
248
249 12. In a caseless character class with UCP support, when a character with more
250 than one alternative case was not the first character of a range, not all
251 the alternative cases were added to the class. For example, s and \x{17f}
252 are both alternative cases for S: the class [RST] was handled correctly,
253 but [R-T] was not.
254
255 13. The configure.ac file always checked for pthread support when JIT was
256 enabled. This is not used in Windows, so I have put this test inside a
257 check for the presence of windows.h (which was already tested for).
258
259 14. Improve pattern prefix search by a simplified Boyer-Moore algorithm in JIT.
260 The algorithm provides a way to skip certain starting offsets, and usually
261 faster than linear prefix searches.
262
263 15. Change 13 for 8.20 updated RunTest to check for the 'fr' locale as well
264 as for 'fr_FR' and 'french'. For some reason, however, it then used the
265 Windows-specific input and output files, which have 'french' screwed in.
266 So this could never have worked. One of the problems with locales is that
267 they aren't always the same. I have now updated RunTest so that it checks
268 the output of the locale test (test 3) against three different output
269 files, and it allows the test to pass if any one of them matches. With luck
270 this should make the test pass on some versions of Solaris where it was
271 failing. Because of the uncertainty, the script did not used to stop if
272 test 3 failed; it now does. If further versions of a French locale ever
273 come to light, they can now easily be added.
274
275 16. If --with-pcregrep-bufsize was given a non-integer value such as "50K",
276 there was a message during ./configure, but it did not stop. This now
277 provokes an error. The invalid example in README has been corrected.
278 If a value less than the minimum is given, the minimum value has always
279 been used, but now a warning is given.
280
281 17. If --enable-bsr-anycrlf was set, the special 16/32-bit test failed. This
282 was a bug in the test system, which is now fixed. Also, the list of various
283 configurations that are tested for each release did not have one with both
284 16/32 bits and --enable-bar-anycrlf. It now does.
285
286 18. pcretest was missing "-C bsr" for displaying the \R default setting.
287
288 19. Little endian PowerPC systems are supported now by the JIT compiler.
289
290 20. The fast forward newline mechanism could enter to an infinite loop on
291 certain invalid UTF-8 input. Although we don't support these cases
292 this issue can be fixed by a performance optimization.
293
294 21. Change 33 of 8.34 is not sufficient to ensure stack safety because it does
295 not take account if existing stack usage. There is now a new global
296 variable called pcre_stack_guard that can be set to point to an external
297 function to check stack availability. It is called at the start of
298 processing every parenthesized group.
299
300 22. A typo in the code meant that in ungreedy mode the max/min qualifier
301 behaved like a min-possessive qualifier, and, for example, /a{1,3}b/U did
302 not match "ab".
303
304 23. When UTF was disabled, the JIT program reported some incorrect compile
305 errors. These messages are silenced now.
306
307 24. Experimental support for ARM-64 and MIPS-64 has been added to the JIT
308 compiler.
309
310 25. Change all the temporary files used in RunGrepTest to be different to those
311 used by RunTest so that the tests can be run simultaneously, for example by
312 "make -j check".
313
314
315 Version 8.34 15-December-2013
316 -----------------------------
317
318 1. Add pcre[16|32]_jit_free_unused_memory to forcibly free unused JIT
319 executable memory. Patch inspired by Carsten Klein.
320
321 2. ./configure --enable-coverage defined SUPPORT_GCOV in config.h, although
322 this macro is never tested and has no effect, because the work to support
323 coverage involves only compiling and linking options and special targets in
324 the Makefile. The comment in config.h implied that defining the macro would
325 enable coverage support, which is totally false. There was also support for
326 setting this macro in the CMake files (my fault, I just copied it from
327 configure). SUPPORT_GCOV has now been removed.
328
329 3. Make a small performance improvement in strlen16() and strlen32() in
330 pcretest.
331
332 4. Change 36 for 8.33 left some unreachable statements in pcre_exec.c,
333 detected by the Solaris compiler (gcc doesn't seem to be able to diagnose
334 these cases). There was also one in pcretest.c.
335
336 5. Cleaned up a "may be uninitialized" compiler warning in pcre_exec.c.
337
338 6. In UTF mode, the code for checking whether a group could match an empty
339 string (which is used for indefinitely repeated groups to allow for
340 breaking an infinite loop) was broken when the group contained a repeated
341 negated single-character class with a character that occupied more than one
342 data item and had a minimum repetition of zero (for example, [^\x{100}]* in
343 UTF-8 mode). The effect was undefined: the group might or might not be
344 deemed as matching an empty string, or the program might have crashed.
345
346 7. The code for checking whether a group could match an empty string was not
347 recognizing that \h, \H, \v, \V, and \R must match a character.
348
349 8. Implemented PCRE_INFO_MATCH_EMPTY, which yields 1 if the pattern can match
350 an empty string. If it can, pcretest shows this in its information output.
351
352 9. Fixed two related bugs that applied to Unicode extended grapheme clusters
353 that were repeated with a maximizing qualifier (e.g. \X* or \X{2,5}) when
354 matched by pcre_exec() without using JIT:
355
356 (a) If the rest of the pattern did not match after a maximal run of
357 grapheme clusters, the code for backing up to try with fewer of them
358 did not always back up over a full grapheme when characters that do not
359 have the modifier quality were involved, e.g. Hangul syllables.
360
361 (b) If the match point in a subject started with modifier character, and
362 there was no match, the code could incorrectly back up beyond the match
363 point, and potentially beyond the first character in the subject,
364 leading to a segfault or an incorrect match result.
365
366 10. A conditional group with an assertion condition could lead to PCRE
367 recording an incorrect first data item for a match if no other first data
368 item was recorded. For example, the pattern (?(?=ab)ab) recorded "a" as a
369 first data item, and therefore matched "ca" after "c" instead of at the
370 start.
371
372 11. Change 40 for 8.33 (allowing pcregrep to find empty strings) showed up a
373 bug that caused the command "echo a | ./pcregrep -M '|a'" to loop.
374
375 12. The source of pcregrep now includes z/OS-specific code so that it can be
376 compiled for z/OS as part of the special z/OS distribution.
377
378 13. Added the -T and -TM options to pcretest.
379
380 14. The code in pcre_compile.c for creating the table of named capturing groups
381 has been refactored. Instead of creating the table dynamically during the
382 actual compiling pass, the information is remembered during the pre-compile
383 pass (on the stack unless there are more than 20 named groups, in which
384 case malloc() is used) and the whole table is created before the actual
385 compile happens. This has simplified the code (it is now nearly 150 lines
386 shorter) and prepared the way for better handling of references to groups
387 with duplicate names.
388
389 15. A back reference to a named subpattern when there is more than one of the
390 same name now checks them in the order in which they appear in the pattern.
391 The first one that is set is used for the reference. Previously only the
392 first one was inspected. This change makes PCRE more compatible with Perl.
393
394 16. Unicode character properties were updated from Unicode 6.3.0.
395
396 17. The compile-time code for auto-possessification has been refactored, based
397 on a patch by Zoltan Herczeg. It now happens after instead of during
398 compilation. The code is cleaner, and more cases are handled. The option
399 PCRE_NO_AUTO_POSSESS is added for testing purposes, and the -O and /O
400 options in pcretest are provided to set it. It can also be set by
401 (*NO_AUTO_POSSESS) at the start of a pattern.
402
403 18. The character VT has been added to the default ("C" locale) set of
404 characters that match \s and are generally treated as white space,
405 following this same change in Perl 5.18. There is now no difference between
406 "Perl space" and "POSIX space". Whether VT is treated as white space in
407 other locales depends on the locale.
408
409 19. The code for checking named groups as conditions, either for being set or
410 for being recursed, has been refactored (this is related to 14 and 15
411 above). Processing unduplicated named groups should now be as fast at
412 numerical groups, and processing duplicated groups should be faster than
413 before.
414
415 20. Two patches to the CMake build system, by Alexander Barkov:
416
417 (1) Replace the "source" command by "." in CMakeLists.txt because
418 "source" is a bash-ism.
419
420 (2) Add missing HAVE_STDINT_H and HAVE_INTTYPES_H to config-cmake.h.in;
421 without these the CMake build does not work on Solaris.
422
423 21. Perl has changed its handling of \8 and \9. If there is no previously
424 encountered capturing group of those numbers, they are treated as the
425 literal characters 8 and 9 instead of a binary zero followed by the
426 literals. PCRE now does the same.
427
428 22. Following Perl, added \o{} to specify codepoints in octal, making it
429 possible to specify values greater than 0777 and also making them
430 unambiguous.
431
432 23. Perl now gives an error for missing closing braces after \x{... instead of
433 treating the string as literal. PCRE now does the same.
434
435 24. RunTest used to grumble if an inappropriate test was selected explicitly,
436 but just skip it when running all tests. This make it awkward to run ranges
437 of tests when one of them was inappropriate. Now it just skips any
438 inappropriate tests, as it always did when running all tests.
439
440 25. If PCRE_AUTO_CALLOUT and PCRE_UCP were set for a pattern that contained
441 character types such as \d or \w, too many callouts were inserted, and the
442 data that they returned was rubbish.
443
444 26. In UCP mode, \s was not matching two of the characters that Perl matches,
445 namely NEL (U+0085) and MONGOLIAN VOWEL SEPARATOR (U+180E), though they
446 were matched by \h. The code has now been refactored so that the lists of
447 the horizontal and vertical whitespace characters used for \h and \v (which
448 are defined only in one place) are now also used for \s.
449
450 27. Add JIT support for the 64 bit TileGX architecture.
451 Patch by Jiong Wang (Tilera Corporation).
452
453 28. Possessive quantifiers for classes (both explicit and automatically
454 generated) now use special opcodes instead of wrapping in ONCE brackets.
455
456 29. Whereas an item such as A{4}+ ignored the possessivenes of the quantifier
457 (because it's meaningless), this was not happening when PCRE_CASELESS was
458 set. Not wrong, but inefficient.
459
460 30. Updated perltest.pl to add /u (force Unicode mode) when /W (use Unicode
461 properties for \w, \d, etc) is present in a test regex. Otherwise if the
462 test contains no characters greater than 255, Perl doesn't realise it
463 should be using Unicode semantics.
464
465 31. Upgraded the handling of the POSIX classes [:graph:], [:print:], and
466 [:punct:] when PCRE_UCP is set so as to include the same characters as Perl
467 does in Unicode mode.
468
469 32. Added the "forbid" facility to pcretest so that putting tests into the
470 wrong test files can sometimes be quickly detected.
471
472 33. There is now a limit (default 250) on the depth of nesting of parentheses.
473 This limit is imposed to control the amount of system stack used at compile
474 time. It can be changed at build time by --with-parens-nest-limit=xxx or
475 the equivalent in CMake.
476
477 34. Character classes such as [A-\d] or [a-[:digit:]] now cause compile-time
478 errors. Perl warns for these when in warning mode, but PCRE has no facility
479 for giving warnings.
480
481 35. Change 34 for 8.13 allowed quantifiers on assertions, because Perl does.
482 However, this was not working for (?!) because it is optimized to (*FAIL),
483 for which PCRE does not allow quantifiers. The optimization is now disabled
484 when a quantifier follows (?!). I can't see any use for this, but it makes
485 things uniform.
486
487 36. Perl no longer allows group names to start with digits, so I have made this
488 change also in PCRE. It simplifies the code a bit.
489
490 37. In extended mode, Perl ignores spaces before a + that indicates a
491 possessive quantifier. PCRE allowed a space before the quantifier, but not
492 before the possessive +. It now does.
493
494 38. The use of \K (reset reported match start) within a repeated possessive
495 group such as (a\Kb)*+ was not working.
496
497 40. Document that the same character tables must be used at compile time and
498 run time, and that the facility to pass tables to pcre_exec() and
499 pcre_dfa_exec() is for use only with saved/restored patterns.
500
501 41. Applied Jeff Trawick's patch CMakeLists.txt, which "provides two new
502 features for Builds with MSVC:
503
504 1. Support pcre.rc and/or pcreposix.rc (as is already done for MinGW
505 builds). The .rc files can be used to set FileDescription and many other
506 attributes.
507
508 2. Add an option (-DINSTALL_MSVC_PDB) to enable installation of .pdb files.
509 This allows higher-level build scripts which want .pdb files to avoid
510 hard-coding the exact files needed."
511
512 42. Added support for [[:<:]] and [[:>:]] as used in the BSD POSIX library to
513 mean "start of word" and "end of word", respectively, as a transition aid.
514
515 43. A minimizing repeat of a class containing codepoints greater than 255 in
516 non-UTF 16-bit or 32-bit modes caused an internal error when PCRE was
517 compiled to use the heap for recursion.
518
519 44. Got rid of some compiler warnings for unused variables when UTF but not UCP
520 is configured.
521
522
523 Version 8.33 28-May-2013
524 ------------------------
525
526 1. Added 'U' to some constants that are compared to unsigned integers, to
527 avoid compiler signed/unsigned warnings. Added (int) casts to unsigned
528 variables that are added to signed variables, to ensure the result is
529 signed and can be negated.
530
531 2. Applied patch by Daniel Richard G for quashing MSVC warnings to the
532 CMake config files.
533
534 3. Revise the creation of config.h.generic so that all boolean macros are
535 #undefined, whereas non-boolean macros are #ifndef/#endif-ed. This makes
536 overriding via -D on the command line possible.
537
538 4. Changing the definition of the variable "op" in pcre_exec.c from pcre_uchar
539 to unsigned int is reported to make a quite noticeable speed difference in
540 a specific Windows environment. Testing on Linux did also appear to show
541 some benefit (and it is clearly not harmful). Also fixed the definition of
542 Xop which should be unsigned.
543
544 5. Related to (4), changing the definition of the intermediate variable cc
545 in repeated character loops from pcre_uchar to pcre_uint32 also gave speed
546 improvements.
547
548 6. Fix forward search in JIT when link size is 3 or greater. Also removed some
549 unnecessary spaces.
550
551 7. Adjust autogen.sh and configure.ac to lose warnings given by automake 1.12
552 and later.
553
554 8. Fix two buffer over read issues in 16 and 32 bit modes. Affects JIT only.
555
556 9. Optimizing fast_forward_start_bits in JIT.
557
558 10. Adding support for callouts in JIT, and fixing some issues revealed
559 during this work. Namely:
560
561 (a) Unoptimized capturing brackets incorrectly reset on backtrack.
562
563 (b) Minimum length was not checked before the matching is started.
564
565 11. The value of capture_last that is passed to callouts was incorrect in some
566 cases when there was a capture on one path that was subsequently abandoned
567 after a backtrack. Also, the capture_last value is now reset after a
568 recursion, since all captures are also reset in this case.
569
570 12. The interpreter no longer returns the "too many substrings" error in the
571 case when an overflowing capture is in a branch that is subsequently
572 abandoned after a backtrack.
573
574 13. In the pathological case when an offset vector of size 2 is used, pcretest
575 now prints out the matched string after a yield of 0 or 1.
576
577 14. Inlining subpatterns in recursions, when certain conditions are fulfilled.
578 Only supported by the JIT compiler at the moment.
579
580 15. JIT compiler now supports 32 bit Macs thanks to Lawrence Velazquez.
581
582 16. Partial matches now set offsets[2] to the "bumpalong" value, that is, the
583 offset of the starting point of the matching process, provided the offsets
584 vector is large enough.
585
586 17. The \A escape now records a lookbehind value of 1, though its execution
587 does not actually inspect the previous character. This is to ensure that,
588 in partial multi-segment matching, at least one character from the old
589 segment is retained when a new segment is processed. Otherwise, if there
590 are no lookbehinds in the pattern, \A might match incorrectly at the start
591 of a new segment.
592
593 18. Added some #ifdef __VMS code into pcretest.c to help VMS implementations.
594
595 19. Redefined some pcre_uchar variables in pcre_exec.c as pcre_uint32; this
596 gives some modest performance improvement in 8-bit mode.
597
598 20. Added the PCRE-specific property \p{Xuc} for matching characters that can
599 be expressed in certain programming languages using Universal Character
600 Names.
601
602 21. Unicode validation has been updated in the light of Unicode Corrigendum #9,
603 which points out that "non characters" are not "characters that may not
604 appear in Unicode strings" but rather "characters that are reserved for
605 internal use and have only local meaning".
606
607 22. When a pattern was compiled with automatic callouts (PCRE_AUTO_CALLOUT) and
608 there was a conditional group that depended on an assertion, if the
609 assertion was false, the callout that immediately followed the alternation
610 in the condition was skipped when pcre_exec() was used for matching.
611
612 23. Allow an explicit callout to be inserted before an assertion that is the
613 condition for a conditional group, for compatibility with automatic
614 callouts, which always insert a callout at this point.
615
616 24. In 8.31, (*COMMIT) was confined to within a recursive subpattern. Perl also
617 confines (*SKIP) and (*PRUNE) in the same way, and this has now been done.
618
619 25. (*PRUNE) is now supported by the JIT compiler.
620
621 26. Fix infinite loop when /(?<=(*SKIP)ac)a/ is matched against aa.
622
623 27. Fix the case where there are two or more SKIPs with arguments that may be
624 ignored.
625
626 28. (*SKIP) is now supported by the JIT compiler.
627
628 29. (*THEN) is now supported by the JIT compiler.
629
630 30. Update RunTest with additional test selector options.
631
632 31. The way PCRE handles backtracking verbs has been changed in two ways.
633
634 (1) Previously, in something like (*COMMIT)(*SKIP), COMMIT would override
635 SKIP. Now, PCRE acts on whichever backtracking verb is reached first by
636 backtracking. In some cases this makes it more Perl-compatible, but Perl's
637 rather obscure rules do not always do the same thing.
638
639 (2) Previously, backtracking verbs were confined within assertions. This is
640 no longer the case for positive assertions, except for (*ACCEPT). Again,
641 this sometimes improves Perl compatibility, and sometimes does not.
642
643 32. A number of tests that were in test 2 because Perl did things differently
644 have been moved to test 1, because either Perl or PCRE has changed, and
645 these tests are now compatible.
646
647 32. Backtracking control verbs are now handled in the same way in JIT and
648 interpreter.
649
650 33. An opening parenthesis in a MARK/PRUNE/SKIP/THEN name in a pattern that
651 contained a forward subroutine reference caused a compile error.
652
653 34. Auto-detect and optimize limited repetitions in JIT.
654
655 35. Implement PCRE_NEVER_UTF to lock out the use of UTF, in particular,
656 blocking (*UTF) etc.
657
658 36. In the interpreter, maximizing pattern repetitions for characters and
659 character types now use tail recursion, which reduces stack usage.
660
661 37. The value of the max lookbehind was not correctly preserved if a compiled
662 and saved regex was reloaded on a host of different endianness.
663
664 38. Implemented (*LIMIT_MATCH) and (*LIMIT_RECURSION). As part of the extension
665 of the compiled pattern block, expand the flags field from 16 to 32 bits
666 because it was almost full.
667
668 39. Try madvise first before posix_madvise.
669
670 40. Change 7 for PCRE 7.9 made it impossible for pcregrep to find empty lines
671 with a pattern such as ^$. It has taken 4 years for anybody to notice! The
672 original change locked out all matches of empty strings. This has been
673 changed so that one match of an empty string per line is recognized.
674 Subsequent searches on the same line (for colouring or for --only-matching,
675 for example) do not recognize empty strings.
676
677 41. Applied a user patch to fix a number of spelling mistakes in comments.
678
679 42. Data lines longer than 65536 caused pcretest to crash.
680
681 43. Clarified the data type for length and startoffset arguments for pcre_exec
682 and pcre_dfa_exec in the function-specific man pages, where they were
683 explicitly stated to be in bytes, never having been updated. I also added
684 some clarification to the pcreapi man page.
685
686 44. A call to pcre_dfa_exec() with an output vector size less than 2 caused
687 a segmentation fault.
688
689
690 Version 8.32 30-November-2012
691 -----------------------------
692
693 1. Improved JIT compiler optimizations for first character search and single
694 character iterators.
695
696 2. Supporting IBM XL C compilers for PPC architectures in the JIT compiler.
697 Patch by Daniel Richard G.
698
699 3. Single character iterator optimizations in the JIT compiler.
700
701 4. Improved JIT compiler optimizations for character ranges.
702
703 5. Rename the "leave" variable names to "quit" to improve WinCE compatibility.
704 Reported by Giuseppe D'Angelo.
705
706 6. The PCRE_STARTLINE bit, indicating that a match can occur only at the start
707 of a line, was being set incorrectly in cases where .* appeared inside
708 atomic brackets at the start of a pattern, or where there was a subsequent
709 *PRUNE or *SKIP.
710
711 7. Improved instruction cache flush for POWER/PowerPC.
712 Patch by Daniel Richard G.
713
714 8. Fixed a number of issues in pcregrep, making it more compatible with GNU
715 grep:
716
717 (a) There is now no limit to the number of patterns to be matched.
718
719 (b) An error is given if a pattern is too long.
720
721 (c) Multiple uses of --exclude, --exclude-dir, --include, and --include-dir
722 are now supported.
723
724 (d) --exclude-from and --include-from (multiple use) have been added.
725
726 (e) Exclusions and inclusions now apply to all files and directories, not
727 just to those obtained from scanning a directory recursively.
728
729 (f) Multiple uses of -f and --file-list are now supported.
730
731 (g) In a Windows environment, the default for -d has been changed from
732 "read" (the GNU grep default) to "skip", because otherwise the presence
733 of a directory in the file list provokes an error.
734
735 (h) The documentation has been revised and clarified in places.
736
737 9. Improve the matching speed of capturing brackets.
738
739 10. Changed the meaning of \X so that it now matches a Unicode extended
740 grapheme cluster.
741
742 11. Patch by Daniel Richard G to the autoconf files to add a macro for sorting
743 out POSIX threads when JIT support is configured.
744
745 12. Added support for PCRE_STUDY_EXTRA_NEEDED.
746
747 13. In the POSIX wrapper regcomp() function, setting re_nsub field in the preg
748 structure could go wrong in environments where size_t is not the same size
749 as int.
750
751 14. Applied user-supplied patch to pcrecpp.cc to allow PCRE_NO_UTF8_CHECK to be
752 set.
753
754 15. The EBCDIC support had decayed; later updates to the code had included
755 explicit references to (e.g.) \x0a instead of CHAR_LF. There has been a
756 general tidy up of EBCDIC-related issues, and the documentation was also
757 not quite right. There is now a test that can be run on ASCII systems to
758 check some of the EBCDIC-related things (but is it not a full test).
759
760 16. The new PCRE_STUDY_EXTRA_NEEDED option is now used by pcregrep, resulting
761 in a small tidy to the code.
762
763 17. Fix JIT tests when UTF is disabled and both 8 and 16 bit mode are enabled.
764
765 18. If the --only-matching (-o) option in pcregrep is specified multiple
766 times, each one causes appropriate output. For example, -o1 -o2 outputs the
767 substrings matched by the 1st and 2nd capturing parentheses. A separating
768 string can be specified by --om-separator (default empty).
769
770 19. Improving the first n character searches.
771
772 20. Turn case lists for horizontal and vertical white space into macros so that
773 they are defined only once.
774
775 21. This set of changes together give more compatible Unicode case-folding
776 behaviour for characters that have more than one other case when UCP
777 support is available.
778
779 (a) The Unicode property table now has offsets into a new table of sets of
780 three or more characters that are case-equivalent. The MultiStage2.py
781 script that generates these tables (the pcre_ucd.c file) now scans
782 CaseFolding.txt instead of UnicodeData.txt for character case
783 information.
784
785 (b) The code for adding characters or ranges of characters to a character
786 class has been abstracted into a generalized function that also handles
787 case-independence. In UTF-mode with UCP support, this uses the new data
788 to handle characters with more than one other case.
789
790 (c) A bug that is fixed as a result of (b) is that codepoints less than 256
791 whose other case is greater than 256 are now correctly matched
792 caselessly. Previously, the high codepoint matched the low one, but not
793 vice versa.
794
795 (d) The processing of \h, \H, \v, and \ in character classes now makes use
796 of the new class addition function, using character lists defined as
797 macros alongside the case definitions of 20 above.
798
799 (e) Caseless back references now work with characters that have more than
800 one other case.
801
802 (f) General caseless matching of characters with more than one other case
803 is supported.
804
805 22. Unicode character properties were updated from Unicode 6.2.0
806
807 23. Improved CMake support under Windows. Patch by Daniel Richard G.
808
809 24. Add support for 32-bit character strings, and UTF-32
810
811 25. Major JIT compiler update (code refactoring and bugfixing).
812 Experimental Sparc 32 support is added.
813
814 26. Applied a modified version of Daniel Richard G's patch to create
815 pcre.h.generic and config.h.generic by "make" instead of in the
816 PrepareRelease script.
817
818 27. Added a definition for CHAR_NULL (helpful for the z/OS port), and use it in
819 pcre_compile.c when checking for a zero character.
820
821 28. Introducing a native interface for JIT. Through this interface, the compiled
822 machine code can be directly executed. The purpose of this interface is to
823 provide fast pattern matching, so several sanity checks are not performed.
824 However, feature tests are still performed. The new interface provides
825 1.4x speedup compared to the old one.
826
827 29. If pcre_exec() or pcre_dfa_exec() was called with a negative value for
828 the subject string length, the error given was PCRE_ERROR_BADOFFSET, which
829 was confusing. There is now a new error PCRE_ERROR_BADLENGTH for this case.
830
831 30. In 8-bit UTF-8 mode, pcretest failed to give an error for data codepoints
832 greater than 0x7fffffff (which cannot be represented in UTF-8, even under
833 the "old" RFC 2279). Instead, it ended up passing a negative length to
834 pcre_exec().
835
836 31. Add support for GCC's visibility feature to hide internal functions.
837
838 32. Running "pcretest -C pcre8" or "pcretest -C pcre16" gave a spurious error
839 "unknown -C option" after outputting 0 or 1.
840
841 33. There is now support for generating a code coverage report for the test
842 suite in environments where gcc is the compiler and lcov is installed. This
843 is mainly for the benefit of the developers.
844
845 34. If PCRE is built with --enable-valgrind, certain memory regions are marked
846 unaddressable using valgrind annotations, allowing valgrind to detect
847 invalid memory accesses. This is mainly for the benefit of the developers.
848
849 25. (*UTF) can now be used to start a pattern in any of the three libraries.
850
851 26. Give configure error if --enable-cpp but no C++ compiler found.
852
853
854 Version 8.31 06-July-2012
855 -------------------------
856
857 1. Fixing a wrong JIT test case and some compiler warnings.
858
859 2. Removed a bashism from the RunTest script.
860
861 3. Add a cast to pcre_exec.c to fix the warning "unary minus operator applied
862 to unsigned type, result still unsigned" that was given by an MS compiler
863 on encountering the code "-sizeof(xxx)".
864
865 4. Partial matching support is added to the JIT compiler.
866
867 5. Fixed several bugs concerned with partial matching of items that consist
868 of more than one character:
869
870 (a) /^(..)\1/ did not partially match "aba" because checking references was
871 done on an "all or nothing" basis. This also applied to repeated
872 references.
873
874 (b) \R did not give a hard partial match if \r was found at the end of the
875 subject.
876
877 (c) \X did not give a hard partial match after matching one or more
878 characters at the end of the subject.
879
880 (d) When newline was set to CRLF, a pattern such as /a$/ did not recognize
881 a partial match for the string "\r".
882
883 (e) When newline was set to CRLF, the metacharacter "." did not recognize
884 a partial match for a CR character at the end of the subject string.
885
886 6. If JIT is requested using /S++ or -s++ (instead of just /S+ or -s+) when
887 running pcretest, the text "(JIT)" added to the output whenever JIT is
888 actually used to run the match.
889
890 7. Individual JIT compile options can be set in pcretest by following -s+[+]
891 or /S+[+] with a digit between 1 and 7.
892
893 8. OP_NOT now supports any UTF character not just single-byte ones.
894
895 9. (*MARK) control verb is now supported by the JIT compiler.
896
897 10. The command "./RunTest list" lists the available tests without actually
898 running any of them. (Because I keep forgetting what they all are.)
899
900 11. Add PCRE_INFO_MAXLOOKBEHIND.
901
902 12. Applied a (slightly modified) user-supplied patch that improves performance
903 when the heap is used for recursion (compiled with --disable-stack-for-
904 recursion). Instead of malloc and free for each heap frame each time a
905 logical recursion happens, frames are retained on a chain and re-used where
906 possible. This sometimes gives as much as 30% improvement.
907
908 13. As documented, (*COMMIT) is now confined to within a recursive subpattern
909 call.
910
911 14. As documented, (*COMMIT) is now confined to within a positive assertion.
912
913 15. It is now possible to link pcretest with libedit as an alternative to
914 libreadline.
915
916 16. (*COMMIT) control verb is now supported by the JIT compiler.
917
918 17. The Unicode data tables have been updated to Unicode 6.1.0.
919
920 18. Added --file-list option to pcregrep.
921
922 19. Added binary file support to pcregrep, including the -a, --binary-files,
923 -I, and --text options.
924
925 20. The madvise function is renamed for posix_madvise for QNX compatibility
926 reasons. Fixed by Giuseppe D'Angelo.
927
928 21. Fixed a bug for backward assertions with REVERSE 0 in the JIT compiler.
929
930 22. Changed the option for creating symbolic links for 16-bit man pages from
931 -s to -sf so that re-installing does not cause issues.
932
933 23. Support PCRE_NO_START_OPTIMIZE in JIT as (*MARK) support requires it.
934
935 24. Fixed a very old bug in pcretest that caused errors with restarted DFA
936 matches in certain environments (the workspace was not being correctly
937 retained). Also added to pcre_dfa_exec() a simple plausibility check on
938 some of the workspace data at the beginning of a restart.
939
940 25. \s*\R was auto-possessifying the \s* when it should not, whereas \S*\R
941 was not doing so when it should - probably a typo introduced by SVN 528
942 (change 8.10/14).
943
944 26. When PCRE_UCP was not set, \w+\x{c4} was incorrectly auto-possessifying the
945 \w+ when the character tables indicated that \x{c4} was a word character.
946 There were several related cases, all because the tests for doing a table
947 lookup were testing for characters less than 127 instead of 255.
948
949 27. If a pattern contains capturing parentheses that are not used in a match,
950 their slots in the ovector are set to -1. For those that are higher than
951 any matched groups, this happens at the end of processing. In the case when
952 there were back references that the ovector was too small to contain
953 (causing temporary malloc'd memory to be used during matching), and the
954 highest capturing number was not used, memory off the end of the ovector
955 was incorrectly being set to -1. (It was using the size of the temporary
956 memory instead of the true size.)
957
958 28. To catch bugs like 27 using valgrind, when pcretest is asked to specify an
959 ovector size, it uses memory at the end of the block that it has got.
960
961 29. Check for an overlong MARK name and give an error at compile time. The
962 limit is 255 for the 8-bit library and 65535 for the 16-bit library.
963
964 30. JIT compiler update.
965
966 31. JIT is now supported on jailbroken iOS devices. Thanks for Ruiger
967 Rill for the patch.
968
969 32. Put spaces around SLJIT_PRINT_D in the JIT compiler. Required by CXX11.
970
971 33. Variable renamings in the PCRE-JIT compiler. No functionality change.
972
973 34. Fixed typos in pcregrep: in two places there was SUPPORT_LIBZ2 instead of
974 SUPPORT_LIBBZ2. This caused a build problem when bzip2 but not gzip (zlib)
975 was enabled.
976
977 35. Improve JIT code generation for greedy plus quantifier.
978
979 36. When /((?:a?)*)*c/ or /((?>a?)*)*c/ was matched against "aac", it set group
980 1 to "aa" instead of to an empty string. The bug affected repeated groups
981 that could potentially match an empty string.
982
983 37. Optimizing single character iterators in JIT.
984
985 38. Wide characters specified with \uxxxx in JavaScript mode are now subject to
986 the same checks as \x{...} characters in non-JavaScript mode. Specifically,
987 codepoints that are too big for the mode are faulted, and in a UTF mode,
988 disallowed codepoints are also faulted.
989
990 39. If PCRE was compiled with UTF support, in three places in the DFA
991 matcher there was code that should only have been obeyed in UTF mode, but
992 was being obeyed unconditionally. In 8-bit mode this could cause incorrect
993 processing when bytes with values greater than 127 were present. In 16-bit
994 mode the bug would be provoked by values in the range 0xfc00 to 0xdc00. In
995 both cases the values are those that cannot be the first data item in a UTF
996 character. The three items that might have provoked this were recursions,
997 possessively repeated groups, and atomic groups.
998
999 40. Ensure that libpcre is explicitly listed in the link commands for pcretest
1000 and pcregrep, because some OS require shared objects to be explicitly
1001 passed to ld, causing the link step to fail if they are not.
1002
1003 41. There were two incorrect #ifdefs in pcre_study.c, meaning that, in 16-bit
1004 mode, patterns that started with \h* or \R* might be incorrectly matched.
1005
1006
1007 Version 8.30 04-February-2012
1008 -----------------------------
1009
1010 1. Renamed "isnumber" as "is_a_number" because in some Mac environments this
1011 name is defined in ctype.h.
1012
1013 2. Fixed a bug in fixed-length calculation for lookbehinds that would show up
1014 only in quite long subpatterns.
1015
1016 3. Removed the function pcre_info(), which has been obsolete and deprecated
1017 since it was replaced by pcre_fullinfo() in February 2000.
1018
1019 4. For a non-anchored pattern, if (*SKIP) was given with a name that did not
1020 match a (*MARK), and the match failed at the start of the subject, a
1021 reference to memory before the start of the subject could occur. This bug
1022 was introduced by fix 17 of release 8.21.
1023
1024 5. A reference to an unset group with zero minimum repetition was giving
1025 totally wrong answers (in non-JavaScript-compatibility mode). For example,
1026 /(another)?(\1?)test/ matched against "hello world test". This bug was
1027 introduced in release 8.13.
1028
1029 6. Add support for 16-bit character strings (a large amount of work involving
1030 many changes and refactorings).
1031
1032 7. RunGrepTest failed on msys because \r\n was replaced by whitespace when the
1033 command "pattern=`printf 'xxx\r\njkl'`" was run. The pattern is now taken
1034 from a file.
1035
1036 8. Ovector size of 2 is also supported by JIT based pcre_exec (the ovector size
1037 rounding is not applied in this particular case).
1038
1039 9. The invalid Unicode surrogate codepoints U+D800 to U+DFFF are now rejected
1040 if they appear, or are escaped, in patterns.
1041
1042 10. Get rid of a number of -Wunused-but-set-variable warnings.
1043
1044 11. The pattern /(?=(*:x))(q|)/ matches an empty string, and returns the mark
1045 "x". The similar pattern /(?=(*:x))((*:y)q|)/ did not return a mark at all.
1046 Oddly, Perl behaves the same way. PCRE has been fixed so that this pattern
1047 also returns the mark "x". This bug applied to capturing parentheses,
1048 non-capturing parentheses, and atomic parentheses. It also applied to some
1049 assertions.
1050
1051 12. Stephen Kelly's patch to CMakeLists.txt allows it to parse the version
1052 information out of configure.ac instead of relying on pcre.h.generic, which
1053 is not stored in the repository.
1054
1055 13. Applied Dmitry V. Levin's patch for a more portable method for linking with
1056 -lreadline.
1057
1058 14. ZH added PCRE_CONFIG_JITTARGET; added its output to pcretest -C.
1059
1060 15. Applied Graycode's patch to put the top-level frame on the stack rather
1061 than the heap when not using the stack for recursion. This gives a
1062 performance improvement in many cases when recursion is not deep.
1063
1064 16. Experimental code added to "pcretest -C" to output the stack frame size.
1065
1066
1067 Version 8.21 12-Dec-2011
1068 ------------------------
1069
1070 1. Updating the JIT compiler.
1071
1072 2. JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
1073 are added as well.
1074
1075 3. Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
1076 PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
1077 calling _pcre_jit_exec. Some extra comments are added.
1078
1079 4. (*MARK) settings inside atomic groups that do not contain any capturing
1080 parentheses, for example, (?>a(*:m)), were not being passed out. This bug
1081 was introduced by change 18 for 8.20.
1082
1083 5. Supporting of \x, \U and \u in JavaScript compatibility mode based on the
1084 ECMA-262 standard.
1085
1086 6. Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
1087 erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
1088 This bug was probably introduced by change 9 of 8.13.
1089
1090 7. While fixing 6 above, I noticed that a number of other items were being
1091 incorrectly rejected as "not fixed length". This arose partly because newer
1092 opcodes had not been added to the fixed-length checking code. I have (a)
1093 corrected the bug and added tests for these items, and (b) arranged for an
1094 error to occur if an unknown opcode is encountered while checking for fixed
1095 length instead of just assuming "not fixed length". The items that were
1096 rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
1097 (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
1098 repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
1099
1100 8. A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
1101 being incorrectly compiled and would have given unpredicatble results.
1102
1103 9. A possessively repeated subpattern with minimum repeat count greater than
1104 one behaved incorrectly. For example, (A){2,}+ behaved as if it was
1105 (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
1106 the first (A) could occur when it should not.
1107
1108 10. Add a cast and remove a redundant test from the code.
1109
1110 11. JIT should use pcre_malloc/pcre_free for allocation.
1111
1112 12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
1113 best practice nowadays, and helps with cross-compiling. (If the exec_prefix
1114 is anything other than /usr, -L is still shown).
1115
1116 13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
1117
1118 14. Perl does not support \N without a following name in a [] class; PCRE now
1119 also gives an error.
1120
1121 15. If a forward reference was repeated with an upper limit of around 2000,
1122 it caused the error "internal error: overran compiling workspace". The
1123 maximum number of forward references (including repeats) was limited by the
1124 internal workspace, and dependent on the LINK_SIZE. The code has been
1125 rewritten so that the workspace expands (via pcre_malloc) if necessary, and
1126 the default depends on LINK_SIZE. There is a new upper limit (for safety)
1127 of around 200,000 forward references. While doing this, I also speeded up
1128 the filling in of repeated forward references.
1129
1130 16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
1131 incorrectly expecting the subject to contain another "a" after the start.
1132
1133 17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
1134 in the match, the SKIP should be ignored. This was not happening; instead
1135 the SKIP was being treated as NOMATCH. For patterns such as
1136 /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
1137 tested.
1138
1139 18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
1140 now much more compatible with Perl, in particular in cases where the result
1141 is a non-match for a non-anchored pattern. For example, if
1142 /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
1143 "m", where previously it did not return a name. A side effect of this
1144 change is that for partial matches, the last encountered mark name is
1145 returned, as for non matches. A number of tests that were previously not
1146 Perl-compatible have been moved into the Perl-compatible test files. The
1147 refactoring has had the pleasing side effect of removing one argument from
1148 the match() function, thus reducing its stack requirements.
1149
1150 19. If the /S+ option was used in pcretest to study a pattern using JIT,
1151 subsequent uses of /S (without +) incorrectly behaved like /S+.
1152
1153 21. Retrieve executable code size support for the JIT compiler and fixing
1154 some warnings.
1155
1156 22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
1157 not work when the shorter character appeared right at the end of the
1158 subject string.
1159
1160 23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
1161 systems.
1162
1163 24. Added PCRE_INFO_JITSIZE to pass on the value from (21) above, and also
1164 output it when the /M option is used in pcretest.
1165
1166 25. The CheckMan script was not being included in the distribution. Also, added
1167 an explicit "perl" to run Perl scripts from the PrepareRelease script
1168 because this is reportedly needed in Windows.
1169
1170 26. If study data was being save in a file and studying had not found a set of
1171 "starts with" bytes for the pattern, the data written to the file (though
1172 never used) was taken from uninitialized memory and so caused valgrind to
1173 complain.
1174
1175 27. Updated RunTest.bat as provided by Sheri Pierce.
1176
1177 28. Fixed a possible uninitialized memory bug in pcre_jit_compile.c.
1178
1179 29. Computation of memory usage for the table of capturing group names was
1180 giving an unnecessarily large value.
1181
1182
1183 Version 8.20 21-Oct-2011
1184 ------------------------
1185
1186 1. Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
1187 a POSIX class. After further experiments with Perl, which convinced me that
1188 Perl has bugs and confusions, a closing square bracket is no longer allowed
1189 in a POSIX name. This bug also affected patterns with classes that started
1190 with full stops.
1191
1192 2. If a pattern such as /(a)b|ac/ is matched against "ac", there is no
1193 captured substring, but while checking the failing first alternative,
1194 substring 1 is temporarily captured. If the output vector supplied to
1195 pcre_exec() was not big enough for this capture, the yield of the function
1196 was still zero ("insufficient space for captured substrings"). This cannot
1197 be totally fixed without adding another stack variable, which seems a lot
1198 of expense for a edge case. However, I have improved the situation in cases
1199 such as /(a)(b)x|abc/ matched against "abc", where the return code
1200 indicates that fewer than the maximum number of slots in the ovector have
1201 been set.
1202
1203 3. Related to (2) above: when there are more back references in a pattern than
1204 slots in the output vector, pcre_exec() uses temporary memory during
1205 matching, and copies in the captures as far as possible afterwards. It was
1206 using the entire output vector, but this conflicts with the specification
1207 that only 2/3 is used for passing back captured substrings. Now it uses
1208 only the first 2/3, for compatibility. This is, of course, another edge
1209 case.
1210
1211 4. Zoltan Herczeg's just-in-time compiler support has been integrated into the
1212 main code base, and can be used by building with --enable-jit. When this is
1213 done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
1214 runtime --no-jit option is given.
1215
1216 5. When the number of matches in a pcre_dfa_exec() run exactly filled the
1217 ovector, the return from the function was zero, implying that there were
1218 other matches that did not fit. The correct "exactly full" value is now
1219 returned.
1220
1221 6. If a subpattern that was called recursively or as a subroutine contained
1222 (*PRUNE) or any other control that caused it to give a non-standard return,
1223 invalid errors such as "Error -26 (nested recursion at the same subject
1224 position)" or even infinite loops could occur.
1225
1226 7. If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
1227 computing the minimum length on reaching *ACCEPT, and so ended up with the
1228 wrong value of 1 rather than 0. Further investigation indicates that
1229 computing a minimum subject length in the presence of *ACCEPT is difficult
1230 (think back references, subroutine calls), and so I have changed the code
1231 so that no minimum is registered for a pattern that contains *ACCEPT.
1232
1233 8. If (*THEN) was present in the first (true) branch of a conditional group,
1234 it was not handled as intended. [But see 16 below.]
1235
1236 9. Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
1237 Sheri Pierce.
1238
1239 10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
1240 the first byte in a match must be "a".
1241
1242 11. Change 17 for 8.13 increased the recursion depth for patterns like
1243 /a(?:.)*?a/ drastically. I've improved things by remembering whether a
1244 pattern contains any instances of (*THEN). If it does not, the old
1245 optimizations are restored. It would be nice to do this on a per-group
1246 basis, but at the moment that is not feasible.
1247
1248 12. In some environments, the output of pcretest -C is CRLF terminated. This
1249 broke RunTest's code that checks for the link size. A single white space
1250 character after the value is now allowed for.
1251
1252 13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
1253 For "fr", it uses the Windows-specific input and output files.
1254
1255 14. If (*THEN) appeared in a group that was called recursively or as a
1256 subroutine, it did not work as intended. [But see next item.]
1257
1258 15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
1259 pattern fragments (but not containing any | characters). If A and B are
1260 matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
1261 was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
1262 D. In other words, Perl considers parentheses that do not contain any |
1263 characters to be part of a surrounding alternative, whereas PCRE was
1264 treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
1265 differently. PCRE now behaves in the same way as Perl, except in the case
1266 of subroutine/recursion calls such as (?1) which have in any case always
1267 been different (but PCRE had them first :-).
1268
1269 16. Related to 15 above: Perl does not treat the | in a conditional group as
1270 creating alternatives. Such a group is treated in the same way as an
1271 ordinary group without any | characters when processing (*THEN). PCRE has
1272 been changed to match Perl's behaviour.
1273
1274 17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
1275 RunGrepTest script failed.
1276
1277 18. Change 22 for version 13 caused atomic groups to use more stack. This is
1278 inevitable for groups that contain captures, but it can lead to a lot of
1279 stack use in large patterns. The old behaviour has been restored for atomic
1280 groups that do not contain any capturing parentheses.
1281
1282 19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
1283 suppress the check for a minimum subject length at run time. (If it was
1284 given to pcre_exec() or pcre_dfa_exec() it did work.)
1285
1286 20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
1287 fail to work when decoding hex characters in data strings in EBCDIC
1288 environments.
1289
1290 21. It appears that in at least one Mac OS environment, the isxdigit() function
1291 is implemented as a macro that evaluates to its argument more than once,
1292 contravening the C 90 Standard (I haven't checked a later standard). There
1293 was an instance in pcretest which caused it to go wrong when processing
1294 \x{...} escapes in subject strings. The has been rewritten to avoid using
1295 things like p++ in the argument of isxdigit().
1296
1297
1298 Version 8.13 16-Aug-2011
1299 ------------------------
1300
1301 1. The Unicode data tables have been updated to Unicode 6.0.0.
1302
1303 2. Two minor typos in pcre_internal.h have been fixed.
1304
1305 3. Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
1306 pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
1307 in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
1308
1309 4. There were a number of related bugs in the code for matching backrefences
1310 caselessly in UTF-8 mode when codes for the characters concerned were
1311 different numbers of bytes. For example, U+023A and U+2C65 are an upper
1312 and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
1313 (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
1314 code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
1315 2-byte code at the end of the subject (it thought there wasn't enough data
1316 left).
1317
1318 5. Comprehensive information about what went wrong is now returned by
1319 pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
1320 as the output vector has at least 2 elements. The offset of the start of
1321 the failing character and a reason code are placed in the vector.
1322
1323 6. When the UTF-8 string check fails for pcre_compile(), the offset that is
1324 now returned is for the first byte of the failing character, instead of the
1325 last byte inspected. This is an incompatible change, but I hope it is small
1326 enough not to be a problem. It makes the returned offset consistent with
1327 pcre_exec() and pcre_dfa_exec().
1328
1329 7. pcretest now gives a text phrase as well as the error number when
1330 pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
1331 failure, the offset and reason code are output.
1332
1333 8. When \R was used with a maximizing quantifier it failed to skip backwards
1334 over a \r\n pair if the subsequent match failed. Instead, it just skipped
1335 back over a single character (\n). This seems wrong (because it treated the
1336 two characters as a single entity when going forwards), conflicts with the
1337 documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
1338 behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
1339 has been changed.
1340
1341 9. Some internal refactoring has changed the processing so that the handling
1342 of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
1343 time (the PCRE_DOTALL option was changed this way some time ago: version
1344 7.7 change 16). This has made it possible to abolish the OP_OPT op code,
1345 which was always a bit of a fudge. It also means that there is one less
1346 argument for the match() function, which reduces its stack requirements
1347 slightly. This change also fixes an incompatibility with Perl: the pattern
1348 (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
1349
1350 10. More internal refactoring has drastically reduced the number of recursive
1351 calls to match() for possessively repeated groups such as (abc)++ when
1352 using pcre_exec().
1353
1354 11. While implementing 10, a number of bugs in the handling of groups were
1355 discovered and fixed:
1356
1357 (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
1358 (a|)*(?1) gave a compile-time internal error.
1359 ((a|)+)+ did not notice that the outer group could match an empty string.
1360 (^a|^)+ was not marked as anchored.
1361 (.*a|.*)+ was not marked as matching at start or after a newline.
1362
1363 12. Yet more internal refactoring has removed another argument from the match()
1364 function. Special calls to this function are now indicated by setting a
1365 value in a variable in the "match data" data block.
1366
1367 13. Be more explicit in pcre_study() instead of relying on "default" for
1368 opcodes that mean there is no starting character; this means that when new
1369 ones are added and accidentally left out of pcre_study(), testing should
1370 pick them up.
1371
1372 14. The -s option of pcretest has been documented for ages as being an old
1373 synonym of -m (show memory usage). I have changed it to mean "force study
1374 for every regex", that is, assume /S for every regex. This is similar to -i
1375 and -d etc. It's slightly incompatible, but I'm hoping nobody is still
1376 using it. It makes it easier to run collections of tests with and without
1377 study enabled, and thereby test pcre_study() more easily. All the standard
1378 tests are now run with and without -s (but some patterns can be marked as
1379 "never study" - see 20 below).
1380
1381 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
1382 restoration of the capturing data to the outer values was not happening
1383 correctly.
1384
1385 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
1386 empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
1387 pattern had matched an empty string, and so incorrectly returned a no
1388 match.
1389
1390 17. There was optimizing code for the last branch of non-capturing parentheses,
1391 and also for the obeyed branch of a conditional subexpression, which used
1392 tail recursion to cut down on stack usage. Unfortunately, now that there is
1393 the possibility of (*THEN) occurring in these branches, tail recursion is
1394 no longer possible because the return has to be checked for (*THEN). These
1395 two optimizations have therefore been removed. [But see 8.20/11 above.]
1396
1397 18. If a pattern containing \R was studied, it was assumed that \R always
1398 matched two bytes, thus causing the minimum subject length to be
1399 incorrectly computed because \R can also match just one byte.
1400
1401 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
1402 was incorrectly computed.
1403
1404 20. If /S is present twice on a test pattern in pcretest input, it now
1405 *disables* studying, thereby overriding the use of -s on the command line
1406 (see 14 above). This is necessary for one or two tests to keep the output
1407 identical in both cases.
1408
1409 21. When (*ACCEPT) was used in an assertion that matched an empty string and
1410 PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
1411
1412 22. When an atomic group that contained a capturing parenthesis was
1413 successfully matched, but the branch in which it appeared failed, the
1414 capturing was not being forgotten if a higher numbered group was later
1415 captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
1416 group 1 to "a", when in fact it should be unset. This applied to multi-
1417 branched capturing and non-capturing groups, repeated or not, and also to
1418 positive assertions (capturing in negative assertions does not happen
1419 in PCRE) and also to nested atomic groups.
1420
1421 23. Add the ++ qualifier feature to pcretest, to show the remainder of the
1422 subject after a captured substring, to make it easier to tell which of a
1423 number of identical substrings has been captured.
1424
1425 24. The way atomic groups are processed by pcre_exec() has been changed so that
1426 if they are repeated, backtracking one repetition now resets captured
1427 values correctly. For example, if ((?>(a+)b)+aabab) is matched against
1428 "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
1429 "aaa". Previously, it would have been "a". As part of this code
1430 refactoring, the way recursive calls are handled has also been changed.
1431
1432 25. If an assertion condition captured any substrings, they were not passed
1433 back unless some other capturing happened later. For example, if
1434 (?(?=(a))a) was matched against "a", no capturing was returned.
1435
1436 26. When studying a pattern that contained subroutine calls or assertions,
1437 the code for finding the minimum length of a possible match was handling
1438 direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
1439 group 1 called group 2 while simultaneously a separate group 2 called group
1440 1). A stack overflow occurred in this case. I have fixed this by limiting
1441 the recursion depth to 10.
1442
1443 27. Updated RunTest.bat in the distribution to the version supplied by Tom
1444 Fortmann. This supports explicit test numbers on the command line, and has
1445 argument validation and error reporting.
1446
1447 28. An instance of \X with an unlimited repeat could fail if at any point the
1448 first character it looked at was a mark character.
1449
1450 29. Some minor code refactoring concerning Unicode properties and scripts
1451 should reduce the stack requirement of match() slightly.
1452
1453 30. Added the '=' option to pcretest to check the setting of unused capturing
1454 slots at the end of the pattern, which are documented as being -1, but are
1455 not included in the return count.
1456
1457 31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
1458 compiled something random. Now it gives a compile-time error (as does
1459 Perl).
1460
1461 32. A *MARK encountered during the processing of a positive assertion is now
1462 recorded and passed back (compatible with Perl).
1463
1464 33. If --only-matching or --colour was set on a pcregrep call whose pattern
1465 had alternative anchored branches, the search for a second match in a line
1466 was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
1467 matched the line "0102" twice. The same bug affected patterns that started
1468 with a backwards assertion. For example /\b01|\b02/ also matched "0102"
1469 twice.
1470
1471 34. Previously, PCRE did not allow quantification of assertions. However, Perl
1472 does, and because of capturing effects, quantifying parenthesized
1473 assertions may at times be useful. Quantifiers are now allowed for
1474 parenthesized assertions.
1475
1476 35. A minor code tidy in pcre_compile() when checking options for \R usage.
1477
1478 36. \g was being checked for fancy things in a character class, when it should
1479 just be a literal "g".
1480
1481 37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
1482 appearance of a nested POSIX class supersedes an apparent external class.
1483 For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
1484 unescaped square brackets may also appear as part of class names. For
1485 example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
1486 more like Perl. (But see 8.20/1 above.)
1487
1488 38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
1489 was because it thought it was \N{name}, which is not supported).
1490
1491 39. Add minix to OS list not supporting the -S option in pcretest.
1492
1493 40. PCRE tries to detect cases of infinite recursion at compile time, but it
1494 cannot analyze patterns in sufficient detail to catch mutual recursions
1495 such as ((?1))((?2)). There is now a runtime test that gives an error if a
1496 subgroup is called recursively as a subpattern for a second time at the
1497 same position in the subject string. In previous releases this might have
1498 been caught by the recursion limit, or it might have run out of stack.
1499
1500 41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
1501 happen only once. PCRE was, however incorrectly giving a compile time error
1502 "recursive call could loop indefinitely" because it cannot analyze the
1503 pattern in sufficient detail. The compile time test no longer happens when
1504 PCRE is compiling a conditional subpattern, but actual runaway loops are
1505 now caught at runtime (see 40 above).
1506
1507 42. It seems that Perl allows any characters other than a closing parenthesis
1508 to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
1509 has been changed to be the same.
1510
1511 43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
1512 as not to get warnings when autogen.sh is called. Also changed
1513 AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
1514
1515 44. To help people who use pcregrep to scan files containing exceedingly long
1516 lines, the following changes have been made:
1517
1518 (a) The default value of the buffer size parameter has been increased from
1519 8K to 20K. (The actual buffer used is three times this size.)
1520
1521 (b) The default can be changed by ./configure --with-pcregrep-bufsize when
1522 PCRE is built.
1523
1524 (c) A --buffer-size=n option has been added to pcregrep, to allow the size
1525 to be set at run time.
1526
1527 (d) Numerical values in pcregrep options can be followed by K or M, for
1528 example --buffer-size=50K.
1529
1530 (e) If a line being scanned overflows pcregrep's buffer, an error is now
1531 given and the return code is set to 2.
1532
1533 45. Add a pointer to the latest mark to the callout data block.
1534
1535 46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
1536 partial match of an empty string instead of no match. This was specific to
1537 the use of ".".
1538
1539 47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
1540 complete match instead of a partial match. This bug was dependent on both
1541 the PCRE_UTF8 and PCRE_DOTALL options being set.
1542
1543 48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
1544 starting byte set, because \b was not being ignored.
1545
1546
1547 Version 8.12 15-Jan-2011
1548 ------------------------
1549
1550 1. Fixed some typos in the markup of the man pages, and wrote a script that
1551 checks for such things as part of the documentation building process.
1552
1553 2. On a big-endian 64-bit system, pcregrep did not correctly process the
1554 --match-limit and --recursion-limit options (added for 8.11). In
1555 particular, this made one of the standard tests fail. (The integer value
1556 went into the wrong half of a long int.)
1557
1558 3. If the --colour option was given to pcregrep with -v (invert match), it
1559 did strange things, either producing crazy output, or crashing. It should,
1560 of course, ignore a request for colour when reporting lines that do not
1561 match.
1562
1563 4. Another pcregrep bug caused similar problems if --colour was specified with
1564 -M (multiline) and the pattern match finished with a line ending.
1565
1566 5. In pcregrep, when a pattern that ended with a literal newline sequence was
1567 matched in multiline mode, the following line was shown as part of the
1568 match. This seems wrong, so I have changed it.
1569
1570 6. Another pcregrep bug in multiline mode, when --colour was specified, caused
1571 the check for further matches in the same line (so they could be coloured)
1572 to overrun the end of the current line. If another match was found, it was
1573 incorrectly shown (and then shown again when found in the next line).
1574
1575 7. If pcregrep was compiled under Windows, there was a reference to the
1576 function pcregrep_exit() before it was defined. I am assuming this was
1577 the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
1578 reported by a user. I've moved the definition above the reference.
1579
1580
1581 Version 8.11 10-Dec-2010
1582 ------------------------
1583
1584 1. (*THEN) was not working properly if there were untried alternatives prior
1585 to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
1586 backtracked to try for "b" instead of moving to the next alternative branch
1587 at the same level (in this case, to look for "c"). The Perl documentation
1588 is clear that when (*THEN) is backtracked onto, it goes to the "next
1589 alternative in the innermost enclosing group".
1590
1591 2. (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
1592 such as (A(*COMMIT)B(*THEN)C|D) any failure after matching A should
1593 result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
1594 (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
1595 (*THEN).
1596
1597 3. If \s appeared in a character class, it removed the VT character from
1598 the class, even if it had been included by some previous item, for example
1599 in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
1600 of \s, but is part of the POSIX "space" class.)
1601
1602 4. A partial match never returns an empty string (because you can always
1603 match an empty string at the end of the subject); however the checking for
1604 an empty string was starting at the "start of match" point. This has been
1605 changed to the "earliest inspected character" point, because the returned
1606 data for a partial match starts at this character. This means that, for
1607 example, /(?<=abc)def/ gives a partial match for the subject "abc"
1608 (previously it gave "no match").
1609
1610 5. Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
1611 of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
1612 previously a full match would be given. However, setting PCRE_PARTIAL_HARD
1613 has an implication that the given string is incomplete (because a partial
1614 match is preferred over a full match). For this reason, these items now
1615 give a partial match in this situation. [Aside: previously, the one case
1616 /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
1617 match rather than a full match, which was wrong by the old rules, but is
1618 now correct.]
1619
1620 6. There was a bug in the handling of #-introduced comments, recognized when
1621 PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
1622 If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
1623 UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
1624 scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
1625 but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
1626 places in pcre_compile().
1627
1628 7. Related to (6) above, when pcre_compile() was skipping #-introduced
1629 comments when looking ahead for named forward references to subpatterns,
1630 the only newline sequence it recognized was NL. It now handles newlines
1631 according to the set newline convention.
1632
1633 8. SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
1634 former, but used strtoul(), whereas pcretest avoided strtoul() but did not
1635 cater for a lack of strerror(). These oversights have been fixed.
1636
1637 9. Added --match-limit and --recursion-limit to pcregrep.
1638
1639 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
1640
1641 11. When the -o option was used, pcregrep was setting a return code of 1, even
1642 when matches were found, and --line-buffered was not being honoured.
1643
1644 12. Added an optional parentheses number to the -o and --only-matching options
1645 of pcregrep.
1646
1647 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
1648 can match an empty string. The code to do it in pcretest and pcredemo
1649 needed fixing:
1650
1651 (a) When the newline convention was "crlf", pcretest got it wrong, skipping
1652 only one byte after an empty string match just before CRLF (this case
1653 just got forgotten; "any" and "anycrlf" were OK).
1654
1655 (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
1656 mode when an empty string match preceded an ASCII character followed by
1657 a non-ASCII character. (The code for advancing by one character rather
1658 than one byte was nonsense.)
1659
1660 (c) The pcredemo.c sample program did not have any code at all to handle
1661 the cases when CRLF is a valid newline sequence.
1662
1663 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
1664 as a starting offset was within the subject string. There is now a new
1665 error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
1666 negative or greater than the length of the string. In order to test this,
1667 pcretest is extended to allow the setting of negative starting offsets.
1668
1669 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
1670 starting offset points to the beginning of a UTF-8 character was
1671 unnecessarily clumsy. I tidied it up.
1672
1673 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
1674 bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
1675
1676 17. Nobody had reported that the --include_dir option, which was added in
1677 release 7.7 should have been called --include-dir (hyphen, not underscore)
1678 for compatibility with GNU grep. I have changed it to --include-dir, but
1679 left --include_dir as an undocumented synonym, and the same for
1680 --exclude-dir, though that is not available in GNU grep, at least as of
1681 release 2.5.4.
1682
1683 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
1684 characters from a string of bytes) have been redefined so as not to use
1685 loops, in order to improve performance in some environments. At the same
1686 time, I abstracted some of the common code into auxiliary macros to save
1687 repetition (this should not affect the compiled code).
1688
1689 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
1690 compile-time error is now given if \c is not followed by an ASCII
1691 character, that is, a byte less than 128. (In EBCDIC mode, the code is
1692 different, and any byte value is allowed.)
1693
1694 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
1695 START_OPTIMIZE option, which is now allowed at compile time - but just
1696 passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
1697 to pcregrep and other applications that have no direct access to PCRE
1698 options. The new /Y option in pcretest sets this option when calling
1699 pcre_compile().
1700
1701 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
1702 back references. Groups containing recursive back references were forced to
1703 be atomic by that change, but in the case of named groups, the amount of
1704 memory required was incorrectly computed, leading to "Failed: internal
1705 error: code overflow". This has been fixed.
1706
1707 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
1708 pcretest.c, to avoid build problems in some Borland environments.
1709
1710
1711 Version 8.10 25-Jun-2010
1712 ------------------------
1713
1714 1. Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
1715 THEN.
1716
1717 2. (*ACCEPT) was not working when inside an atomic group.
1718
1719 3. Inside a character class, \B is treated as a literal by default, but
1720 faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
1721 causes the error). The code is unchanged, but I tidied the documentation.
1722
1723 4. Inside a character class, PCRE always treated \R and \X as literals,
1724 whereas Perl faults them if its -w option is set. I have changed PCRE so
1725 that it faults them when PCRE_EXTRA is set.
1726
1727 5. Added support for \N, which always matches any character other than
1728 newline. (It is the same as "." when PCRE_DOTALL is not set.)
1729
1730 6. When compiling pcregrep with newer versions of gcc which may have
1731 FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
1732 declared with attribute warn_unused_result" were given. Just casting the
1733 result to (void) does not stop the warnings; a more elaborate fudge is
1734 needed. I've used a macro to implement this.
1735
1736 7. Minor change to pcretest.c to avoid a compiler warning.
1737
1738 8. Added four artifical Unicode properties to help with an option to make
1739 \s etc use properties (see next item). The new properties are: Xan
1740 (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
1741
1742 9. Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
1743 use Unicode properties. (*UCP) at the start of a pattern can be used to set
1744 this option. Modified pcretest to add /W to test this facility. Added
1745 REG_UCP to make it available via the POSIX interface.
1746
1747 10. Added --line-buffered to pcregrep.
1748
1749 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
1750 studied, and the match started with a letter with a code point greater than
1751 127 whose first byte was different to the first byte of the other case of
1752 the letter, the other case of this starting letter was not recognized
1753 (#976).
1754
1755 12. If a pattern that was studied started with a repeated Unicode property
1756 test, for example, \p{Nd}+, there was the theoretical possibility of
1757 setting up an incorrect bitmap of starting bytes, but fortunately it could
1758 not have actually happened in practice until change 8 above was made (it
1759 added property types that matched character-matching opcodes).
1760
1761 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
1762 possible starting bytes for non-anchored patterns.
1763
1764 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
1765 \R, and also a number of cases that involve Unicode properties, both
1766 explicit and implicit when PCRE_UCP is set.
1767
1768 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
1769 input, it could crash or give wrong results if characters with values
1770 greater than 0xc0 were present in the subject string. (Detail: it assumed
1771 UTF-8 input when processing these items.)
1772
1773 16. Added a lot of (int) casts to avoid compiler warnings in systems where
1774 size_t is 64-bit (#991).
1775
1776 17. Added a check for running out of memory when PCRE is compiled with
1777 --disable-stack-for-recursion (#990).
1778
1779 18. If the last data line in a file for pcretest does not have a newline on
1780 the end, a newline was missing in the output.
1781
1782 19. The default pcre_chartables.c file recognizes only ASCII characters (values
1783 less than 128) in its various bitmaps. However, there is a facility for
1784 generating tables according to the current locale when PCRE is compiled. It
1785 turns out that in some environments, 0x85 and 0xa0, which are Unicode space
1786 characters, are recognized by isspace() and therefore were getting set in
1787 these tables, and indeed these tables seem to approximate to ISO 8859. This
1788 caused a problem in UTF-8 mode when pcre_study() was used to create a list
1789 of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
1790 which of course cannot start UTF-8 characters. I have changed the code so
1791 that only real ASCII characters (less than 128) and the correct starting
1792 bytes for UTF-8 encodings are set for characters greater than 127 when in
1793 UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
1794 altogether.)
1795
1796 20. Added the /T option to pcretest so as to be able to run tests with non-
1797 standard character tables, thus making it possible to include the tests
1798 used for 19 above in the standard set of tests.
1799
1800 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
1801 reference to a subpattern the other side of a comment that contains an
1802 opening parenthesis caused either an internal compiling error, or a
1803 reference to the wrong subpattern.
1804
1805
1806 Version 8.02 19-Mar-2010
1807 ------------------------
1808
1809 1. The Unicode data tables have been updated to Unicode 5.2.0.
1810
1811 2. Added the option --libs-cpp to pcre-config, but only when C++ support is
1812 configured.
1813
1814 3. Updated the licensing terms in the pcregexp.pas file, as agreed with the
1815 original author of that file, following a query about its status.
1816
1817 4. On systems that do not have stdint.h (e.g. Solaris), check for and include
1818 inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
1819
1820 5. A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
1821 quantifier applied to a forward-referencing subroutine call, could compile
1822 incorrect code or give the error "internal error: previously-checked
1823 referenced subpattern not found".
1824
1825 6. Both MS Visual Studio and Symbian OS have problems with initializing
1826 variables to point to external functions. For these systems, therefore,
1827 pcre_malloc etc. are now initialized to local functions that call the
1828 relevant global functions.
1829
1830 7. There were two entries missing in the vectors called coptable and poptable
1831 in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
1832 I've fixed the data, and added a kludgy way of testing at compile time that
1833 the lengths are correct (equal to the number of opcodes).
1834
1835 8. Following on from 7, I added a similar kludge to check the length of the
1836 eint vector in pcreposix.c.
1837
1838 9. Error texts for pcre_compile() are held as one long string to avoid too
1839 much relocation at load time. To find a text, the string is searched,
1840 counting zeros. There was no check for running off the end of the string,
1841 which could happen if a new error number was added without updating the
1842 string.
1843
1844 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
1845
1846 11. \K was not working if it appeared in an atomic group or in a group that
1847 was called as a "subroutine", or in an assertion. Perl 5.11 documents that
1848 \K is "not well defined" if used in an assertion. PCRE now accepts it if
1849 the assertion is positive, but not if it is negative.
1850
1851 12. Change 11 fortuitously reduced the size of the stack frame used in the
1852 "match()" function of pcre_exec.c by one pointer. Forthcoming
1853 implementation of support for (*MARK) will need an extra pointer on the
1854 stack; I have reserved it now, so that the stack frame size does not
1855 decrease.
1856
1857 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
1858 item in branch that calls a recursion is a subroutine call - as in the
1859 second branch in the above example - was incorrectly given the compile-
1860 time error "recursive call could loop indefinitely" because pcre_compile()
1861 was not correctly checking the subroutine for matching a non-empty string.
1862
1863 14. The checks for overrunning compiling workspace could trigger after an
1864 overrun had occurred. This is a "should never occur" error, but it can be
1865 triggered by pathological patterns such as hundreds of nested parentheses.
1866 The checks now trigger 100 bytes before the end of the workspace.
1867
1868 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
1869
1870
1871 Version 8.01 19-Jan-2010
1872 ------------------------
1873
1874 1. If a pattern contained a conditional subpattern with only one branch (in
1875 particular, this includes all (*DEFINE) patterns), a call to pcre_study()
1876 computed the wrong minimum data length (which is of course zero for such
1877 subpatterns). This could cause incorrect "no match" results.
1878
1879 2. For patterns such as (?i)a(?-i)b|c where an option setting at the start of
1880 the pattern is reset in the first branch, pcre_compile() failed with
1881 "internal error: code overflow at offset...". This happened only when
1882 the reset was to the original external option setting. (An optimization
1883 abstracts leading options settings into an external setting, which was the
1884 cause of this.)
1885
1886 3. A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
1887 of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
1888 assertion pattern did not match (meaning that the assertion was true), it
1889 was incorrectly treated as false if the SKIP had been reached during the
1890 matching. This also applied to assertions used as conditions.
1891
1892 4. If an item that is not supported by pcre_dfa_exec() was encountered in an
1893 assertion subpattern, including such a pattern used as a condition,
1894 unpredictable results occurred, instead of the error return
1895 PCRE_ERROR_DFA_UITEM.
1896
1897 5. The C++ GlobalReplace function was not working like Perl for the special
1898 situation when an empty string is matched. It now does the fancy magic
1899 stuff that is necessary.
1900
1901 6. In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
1902 removed. (These were left over from very, very early versions of PCRE.)
1903
1904 7. Some cosmetic changes to the code to make life easier when compiling it
1905 as part of something else:
1906
1907 (a) Change DEBUG to PCRE_DEBUG.
1908
1909 (b) In pcre_compile(), rename the member of the "branch_chain" structure
1910 called "current" as "current_branch", to prevent a collision with the
1911 Linux macro when compiled as a kernel module.
1912
1913 (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
1914 prevent a collision with the Linux macro when compiled as a kernel
1915 module.
1916
1917 8. In pcre_compile() there are some checks for integer overflows that used to
1918 cast potentially large values to (double). This has been changed to that
1919 when building, a check for int64_t is made, and if it is found, it is used
1920 instead, thus avoiding the use of floating point arithmetic. (There is no
1921 other use of FP in PCRE.) If int64_t is not found, the fallback is to
1922 double.
1923
1924 9. Added two casts to avoid signed/unsigned warnings from VS Studio Express
1925 2005 (difference between two addresses compared to an unsigned value).
1926
1927 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
1928 custom one, because of the following reported problem in Windows:
1929
1930 - libbz2 uses the Pascal calling convention (WINAPI) for the functions
1931 under Win32.
1932 - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
1933 therefore missing the function definition.
1934 - The compiler thus generates a "C" signature for the test function.
1935 - The linker fails to find the "C" function.
1936 - PCRE fails to configure if asked to do so against libbz2.
1937
1938 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
1939 messages were output:
1940
1941 Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
1942 rerunning libtoolize, to keep the correct libtool macros in-tree.
1943 Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
1944
1945 I have done both of these things.
1946
1947 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
1948 most of the time, it *can* run out if it is given a pattern that contains a
1949 runaway infinite recursion. I updated the discussion in the pcrestack man
1950 page.
1951
1952 13. Now that we have gone to the x.xx style of version numbers, the minor
1953 version may start with zero. Using 08 or 09 is a bad idea because users
1954 might check the value of PCRE_MINOR in their code, and 08 or 09 may be
1955 interpreted as invalid octal numbers. I've updated the previous comment in
1956 configure.ac, and also added a check that gives an error if 08 or 09 are
1957 used.
1958
1959 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
1960 causing partial matching to fail when the end of the subject matched \W
1961 in a UTF-8 pattern where \W was quantified with a minimum of 3.
1962
1963 15. There were some discrepancies between the declarations in pcre_internal.h
1964 of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
1965 their definitions. The declarations used "const uschar *" and the
1966 definitions used USPTR. Even though USPTR is normally defined as "const
1967 unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
1968 reported that: "This difference in casting confuses some C++ compilers, for
1969 example, SunCC recognizes above declarations as different functions and
1970 generates broken code for hbpcre." I have changed the declarations to use
1971 USPTR.
1972
1973 16. GNU libtool is named differently on some systems. The autogen.sh script now
1974 tries several variants such as glibtoolize (MacOSX) and libtoolize1x
1975 (FreeBSD).
1976
1977 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
1978 (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
1979 comment: "Figure out how to create a longlong from a string: strtoll and
1980 equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
1981 instance, but it only takes 2 args instead of 3!"
1982
1983 18. A subtle bug concerned with back references has been fixed by a change of
1984 specification, with a corresponding code fix. A pattern such as
1985 ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
1986 refers, was giving matches when it shouldn't. For example, xa=xaaa would
1987 match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
1988 same bug. Such groups have to be quantified to be useful, or contained
1989 inside another quantified group. (If there's no repetition, the reference
1990 can never match.) The problem arises because, having left the group and
1991 moved on to the rest of the pattern, a later failure that backtracks into
1992 the group uses the captured value from the final iteration of the group
1993 rather than the correct earlier one. I have fixed this in PCRE by forcing
1994 any group that contains a reference to itself to be an atomic group; that
1995 is, there cannot be any backtracking into it once it has completed. This is
1996 similar to recursive and subroutine calls.
1997
1998
1999 Version 8.00 19-Oct-09
2000 ----------------------
2001
2002 1. The table for translating pcre_compile() error codes into POSIX error codes
2003 was out-of-date, and there was no check on the pcre_compile() error code
2004 being within the table. This could lead to an OK return being given in
2005 error.
2006
2007 2. Changed the call to open a subject file in pcregrep from fopen(pathname,
2008 "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
2009 in a Windows environment.
2010
2011 3. The pcregrep --count option prints the count for each file even when it is
2012 zero, as does GNU grep. However, pcregrep was also printing all files when
2013 --files-with-matches was added. Now, when both options are given, it prints
2014 counts only for those files that have at least one match. (GNU grep just
2015 prints the file name in this circumstance, but including the count seems
2016 more useful - otherwise, why use --count?) Also ensured that the
2017 combination -clh just lists non-zero counts, with no names.
2018
2019 4. The long form of the pcregrep -F option was incorrectly implemented as
2020 --fixed_strings instead of --fixed-strings. This is an incompatible change,
2021 but it seems right to fix it, and I didn't think it was worth preserving
2022 the old behaviour.
2023
2024 5. The command line items --regex=pattern and --regexp=pattern were not
2025 recognized by pcregrep, which required --regex pattern or --regexp pattern
2026 (with a space rather than an '='). The man page documented the '=' forms,
2027 which are compatible with GNU grep; these now work.
2028
2029 6. No libpcreposix.pc file was created for pkg-config; there was just
2030 libpcre.pc and libpcrecpp.pc. The omission has been rectified.
2031
2032 7. Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
2033 when UCP support is not needed, by modifying the Python script that
2034 generates it from Unicode data files. This should not matter if the module
2035 is correctly used as a library, but I received one complaint about 50K of
2036 unwanted data. My guess is that the person linked everything into his
2037 program rather than using a library. Anyway, it does no harm.
2038
2039 8. A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
2040 was a minimum greater than 1 for a wide character in a possessive
2041 repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
2042 which had an unlimited repeat of a nested, fixed maximum repeat of a wide
2043 character. Chaos in the form of incorrect output or a compiling loop could
2044 result.
2045
2046 9. The restrictions on what a pattern can contain when partial matching is
2047 requested for pcre_exec() have been removed. All patterns can now be
2048 partially matched by this function. In addition, if there are at least two
2049 slots in the offset vector, the offset of the earliest inspected character
2050 for the match and the offset of the end of the subject are set in them when
2051 PCRE_ERROR_PARTIAL is returned.
2052
2053 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
2054 synonymous with PCRE_PARTIAL, for backwards compatibility, and
2055 PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
2056 and may be more useful for multi-segment matching.
2057
2058 11. Partial matching with pcre_exec() is now more intuitive. A partial match
2059 used to be given if ever the end of the subject was reached; now it is
2060 given only if matching could not proceed because another character was
2061 needed. This makes a difference in some odd cases such as Z(*FAIL) with the
2062 string "Z", which now yields "no match" instead of "partial match". In the
2063 case of pcre_dfa_exec(), "no match" is given if every matching path for the
2064 final character ended with (*FAIL).
2065
2066 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
2067 if the pattern had a "must contain" character that was already found in the
2068 earlier partial match, unless partial matching was again requested. For
2069 example, with the pattern /dog.(body)?/, the "must contain" character is
2070 "g". If the first part-match was for the string "dog", restarting with
2071 "sbody" failed. This bug has been fixed.
2072
2073 13. The string returned by pcre_dfa_exec() after a partial match has been
2074 changed so that it starts at the first inspected character rather than the
2075 first character of the match. This makes a difference only if the pattern
2076 starts with a lookbehind assertion or \b or \B (\K is not supported by
2077 pcre_dfa_exec()). It's an incompatible change, but it makes the two
2078 matching functions compatible, and I think it's the right thing to do.
2079
2080 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
2081 so that the demonstration program is easily available in environments where
2082 PCRE has not been installed from source.
2083
2084 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
2085 libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
2086 library.
2087
2088 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
2089 It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
2090 is not the first non-POSIX option to be added. Clearly some people find
2091 these options useful.
2092
2093 17. If a caller to the POSIX matching function regexec() passes a non-zero
2094 value for nmatch with a NULL value for pmatch, the value of
2095 nmatch is forced to zero.
2096
2097 18. RunGrepTest did not have a test for the availability of the -u option of
2098 the diff command, as RunTest does. It now checks in the same way as
2099 RunTest, and also checks for the -b option.
2100
2101 19. If an odd number of negated classes containing just a single character
2102 interposed, within parentheses, between a forward reference to a named
2103 subpattern and the definition of the subpattern, compilation crashed with
2104 an internal error, complaining that it could not find the referenced
2105 subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
2106 [The bug was that it was starting one character too far in when skipping
2107 over the character class, thus treating the ] as data rather than
2108 terminating the class. This meant it could skip too much.]
2109
2110 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
2111 /g option in pcretest when the pattern contains \K, which makes it possible
2112 to have an empty string match not at the start, even when the pattern is
2113 anchored. Updated pcretest and pcredemo to use this option.
2114
2115 21. If the maximum number of capturing subpatterns in a recursion was greater
2116 than the maximum at the outer level, the higher number was returned, but
2117 with unset values at the outer level. The correct (outer level) value is
2118 now given.
2119
2120 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
2121 PCRE did not set those parentheses (unlike Perl). I have now found a way to
2122 make it do so. The string so far is captured, making this feature
2123 compatible with Perl.
2124
2125 23. The tests have been re-organized, adding tests 11 and 12, to make it
2126 possible to check the Perl 5.10 features against Perl 5.10.
2127
2128 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
2129 pattern matches a fixed length string. PCRE did not allow this; now it
2130 does. Neither allows recursion.
2131
2132 25. I finally figured out how to implement a request to provide the minimum
2133 length of subject string that was needed in order to match a given pattern.
2134 (It was back references and recursion that I had previously got hung up
2135 on.) This code has now been added to pcre_study(); it finds a lower bound
2136 to the length of subject needed. It is not necessarily the greatest lower
2137 bound, but using it to avoid searching strings that are too short does give
2138 some useful speed-ups. The value is available to calling programs via
2139 pcre_fullinfo().
2140
2141 26. While implementing 25, I discovered to my embarrassment that pcretest had
2142 not been passing the result of pcre_study() to pcre_dfa_exec(), so the
2143 study optimizations had never been tested with that matching function.
2144 Oops. What is worse, even when it was passed study data, there was a bug in
2145 pcre_dfa_exec() that meant it never actually used it. Double oops. There
2146 were also very few tests of studied patterns with pcre_dfa_exec().
2147
2148 27. If (?| is used to create subpatterns with duplicate numbers, they are now
2149 allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
2150 on the other side of the coin, they are no longer allowed to have different
2151 names, because these cannot be distinguished in PCRE, and this has caused
2152 confusion. (This is a difference from Perl.)
2153
2154 28. When duplicate subpattern names are present (necessarily with different
2155 numbers, as required by 27 above), and a test is made by name in a
2156 conditional pattern, either for a subpattern having been matched, or for
2157 recursion in such a pattern, all the associated numbered subpatterns are
2158 tested, and the overall condition is true if the condition is true for any
2159 one of them. This is the way Perl works, and is also more like the way
2160 testing by number works.
2161
2162
2163 Version 7.9 11-Apr-09
2164 ---------------------
2165
2166 1. When building with support for bzlib/zlib (pcregrep) and/or readline
2167 (pcretest), all targets were linked against these libraries. This included
2168 libpcre, libpcreposix, and libpcrecpp, even though they do not use these
2169 libraries. This caused unwanted dependencies to be created. This problem
2170 has been fixed, and now only pcregrep is linked with bzlib/zlib and only
2171 pcretest is linked with readline.
2172
2173 2. The "typedef int BOOL" in pcre_internal.h that was included inside the
2174 "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
2175 moved outside it again, because FALSE and TRUE are already defined in AIX,
2176 but BOOL is not.
2177
2178 3. The pcre_config() function was treating the PCRE_MATCH_LIMIT and
2179 PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
2180
2181 4. The pcregrep documentation said spaces were inserted as well as colons (or
2182 hyphens) following file names and line numbers when outputting matching
2183 lines. This is not true; no spaces are inserted. I have also clarified the
2184 wording for the --colour (or --color) option.
2185
2186 5. In pcregrep, when --colour was used with -o, the list of matching strings
2187 was not coloured; this is different to GNU grep, so I have changed it to be
2188 the same.
2189
2190 6. When --colo(u)r was used in pcregrep, only the first matching substring in
2191 each matching line was coloured. Now it goes on to look for further matches
2192 of any of the test patterns, which is the same behaviour as GNU grep.
2193
2194 7. A pattern that could match an empty string could cause pcregrep to loop; it
2195 doesn't make sense to accept an empty string match in pcregrep, so I have
2196 locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
2197 seems to be how GNU grep behaves. [But see later change 40 for release
2198 8.33.]
2199
2200 8. The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
2201 start or after a newline", because the conditional assertion was not being
2202 correctly handled. The rule now is that both the assertion and what follows
2203 in the first alternative must satisfy the test.
2204
2205 9. If auto-callout was enabled in a pattern with a conditional group whose
2206 condition was an assertion, PCRE could crash during matching, both with
2207 pcre_exec() and pcre_dfa_exec().
2208
2209 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
2210 used for matching.
2211
2212 11. Unicode property support in character classes was not working for
2213 characters (bytes) greater than 127 when not in UTF-8 mode.
2214
2215 12. Added the -M command line option to pcretest.
2216
2217 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
2218
2219 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
2220
2221 16. Added comments and documentation about mis-use of no_arg in the C++
2222 wrapper.
2223
2224 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
2225 from Martin Jerabek that uses macro names for all relevant character and
2226 string constants.
2227
2228 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
2229 SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
2230 SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
2231 these, but not everybody uses configure.
2232
2233 19. A conditional group that had only one branch was not being correctly
2234 recognized as an item that could match an empty string. This meant that an
2235 enclosing group might also not be so recognized, causing infinite looping
2236 (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
2237 with the subject "ab", where knowledge that the repeated group can match
2238 nothing is needed in order to break the loop.
2239
2240 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
2241 exec(), but without supplying a callout function, matching went wrong.
2242
2243 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
2244 leak if the size of the offset vector was greater than 30. When the vector
2245 is smaller, the saved offsets during recursion go onto a local stack
2246 vector, but for larger vectors malloc() is used. It was failing to free
2247 when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
2248 error, in fact).
2249
2250 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
2251 heapframe that is used only when UTF-8 support is enabled. This caused no
2252 problem, but was untidy.
2253
2254 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
2255 CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
2256 included within another project.
2257
2258 24. Steven Van Ingelgem's patches to add more options to the CMake support,
2259 slightly modified by me:
2260
2261 (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
2262 not building pcregrep.
2263
2264 (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
2265 if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
2266
2267 25. Forward references, both numeric and by name, in patterns that made use of
2268 duplicate group numbers, could behave incorrectly or give incorrect errors,
2269 because when scanning forward to find the reference group, PCRE was not
2270 taking into account the duplicate group numbers. A pattern such as
2271 ^X(?3)(a)(?|(b)|(q))(Y) is an example.
2272
2273 26. Changed a few more instances of "const unsigned char *" to USPTR, making
2274 the feature of a custom pointer more persuasive (as requested by a user).
2275
2276 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
2277 pcretest.c, inside #ifndefs, because it seems they are sometimes already
2278 pre-defined.
2279
2280 28. Added support for (*UTF8) at the start of a pattern.
2281
2282 29. Arrange for flags added by the "release type" setting in CMake to be shown
2283 in the configuration summary.
2284
2285
2286 Version 7.8 05-Sep-08
2287 ---------------------
2288
2289 1. Replaced UCP searching code with optimized version as implemented for Ad
2290 Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
2291 stage table and inline lookup instead of a function, giving speed ups of 2
2292 to 5 times on some simple patterns that I tested. Permission was given to
2293 distribute the MultiStage2.py script that generates the tables (it's not in
2294 the tarball, but is in the Subversion repository).
2295
2296 2. Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
2297 scripts.
2298
2299 3. Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
2300 a group with a zero qualifier. The result of the study could be incorrect,
2301 or the function might crash, depending on the pattern.
2302
2303 4. Caseless matching was not working for non-ASCII characters in back
2304 references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
2305 It now works when Unicode Property Support is available.
2306
2307 5. In pcretest, an escape such as \x{de} in the data was always generating
2308 a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
2309 non-UTF-8 mode. If the value is greater than 255, it gives a warning about
2310 truncation.
2311
2312 6. Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
2313
2314 7. Added two (int) casts to pcregrep when printing the difference of two
2315 pointers, in case they are 64-bit values.
2316
2317 8. Added comments about Mac OS X stack usage to the pcrestack man page and to
2318 test 2 if it fails.
2319
2320 9. Added PCRE_CALL_CONVENTION just before the names of all exported functions,
2321 and a #define of that name to empty if it is not externally set. This is to
2322 allow users of MSVC to set it if necessary.
2323
2324 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
2325 the convenience functions in the pcre_get.c source file.
2326
2327 11. An option change at the start of a pattern that had top-level alternatives
2328 could cause overwriting and/or a crash. This command provoked a crash in
2329 some environments:
2330
2331 printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
2332
2333 This potential security problem was recorded as CVE-2008-2371.
2334
2335 12. For a pattern where the match had to start at the beginning or immediately
2336 after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
2337 pcre_dfa_exec() could read past the end of the passed subject if there was
2338 no match. To help with detecting such bugs (e.g. with valgrind), I modified
2339 pcretest so that it places the subject at the end of its malloc-ed buffer.
2340
2341 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
2342 exec() might read past the end of the data buffer in UTF-8 mode.
2343
2344 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
2345 the data contained the byte 0x85 as part of a UTF-8 character within its
2346 first line. This applied both to normal and DFA matching.
2347
2348 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
2349 /^[^d]*?$/8 failed to match "abc".
2350
2351 16. Added a missing copyright notice to pcrecpp_internal.h.
2352
2353 17. Make it more clear in the documentation that values returned from
2354 pcre_exec() in ovector are byte offsets, not character counts.
2355
2356 18. Tidied a few places to stop certain compilers from issuing warnings.
2357
2358 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
2359 supplied by Stefan Weber. I made a further small update for 7.8 because
2360 there is a change of source arrangements: the pcre_searchfuncs.c module is
2361 replaced by pcre_ucd.c.
2362
2363
2364 Version 7.7 07-May-08
2365 ---------------------
2366
2367 1. Applied Craig's patch to sort out a long long problem: "If we can't convert
2368 a string to a long long, pretend we don't even have a long long." This is
2369 done by checking for the strtoq, strtoll, and _strtoi64 functions.
2370
2371 2. Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
2372 pre-7.6 versions, which defined a global no_arg variable instead of putting
2373 it in the RE class. (See also #8 below.)
2374
2375 3. Remove a line of dead code, identified by coverity and reported by Nuno
2376 Lopes.
2377
2378 4. Fixed two related pcregrep bugs involving -r with --include or --exclude:
2379
2380 (1) The include/exclude patterns were being applied to the whole pathnames
2381 of files, instead of just to the final components.
2382
2383 (2) If there was more than one level of directory, the subdirectories were
2384 skipped unless they satisfied the include/exclude conditions. This is
2385 inconsistent with GNU grep (and could even be seen as contrary to the
2386 pcregrep specification - which I improved to make it absolutely clear).
2387 The action now is always to scan all levels of directory, and just
2388 apply the include/exclude patterns to regular files.
2389
2390 5. Added the --include_dir and --exclude_dir patterns to pcregrep, and used
2391 --exclude_dir in the tests to avoid scanning .svn directories.
2392
2393 6. Applied Craig's patch to the QuoteMeta function so that it escapes the
2394 NUL character as backslash + 0 rather than backslash + NUL, because PCRE
2395 doesn't support NULs in patterns.
2396
2397 7. Added some missing "const"s to declarations of static tables in
2398 pcre_compile.c and pcre_dfa_exec.c.
2399
2400 8. Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
2401 caused by fix #2 above. (Subsequently also a second patch to fix the
2402 first patch. And a third patch - this was a messy problem.)
2403
2404 9. Applied Craig's patch to remove the use of push_back().
2405
2406 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
2407 matching function regexec().
2408
2409 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
2410 which, however, unlike Perl's \g{...}, are subroutine calls, not back
2411 references. PCRE supports relative numbers with this syntax (I don't think
2412 Oniguruma does).
2413
2414 12. Previously, a group with a zero repeat such as (...){0} was completely
2415 omitted from the compiled regex. However, this means that if the group
2416 was called as a subroutine from elsewhere in the pattern, things went wrong
2417 (an internal error was given). Such groups are now left in the compiled
2418 pattern, with a new opcode that causes them to be skipped at execution
2419 time.
2420
2421 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
2422 to the way PCRE behaves:
2423
2424 (a) A lone ] character is dis-allowed (Perl treats it as data).
2425
2426 (b) A back reference to an unmatched subpattern matches an empty string
2427 (Perl fails the current match path).
2428
2429 (c) A data ] in a character class must be notated as \] because if the
2430 first data character in a class is ], it defines an empty class. (In
2431 Perl it is not possible to have an empty class.) The empty class []
2432 never matches; it forces failure and is equivalent to (*FAIL) or (?!).
2433 The negative empty class [^] matches any one character, independently
2434 of the DOTALL setting.
2435
2436 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
2437 non-existent subpattern following a character class starting with ']' and
2438 containing () gave an internal compiling error instead of "reference to
2439 non-existent subpattern". Fortunately, when the pattern did exist, the
2440 compiled code was correct. (When scanning forwards to check for the
2441 existence of the subpattern, it was treating the data ']' as terminating
2442 the class, so got the count wrong. When actually compiling, the reference
2443 was subsequently set up correctly.)
2444
2445 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
2446 it was being rejected as not supported by pcre_dfa_exec(), even though
2447 other assertions are supported. I have made pcre_dfa_exec() support
2448 (*FAIL).
2449
2450 16. The implementation of 13c above involved the invention of a new opcode,
2451 OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
2452 cannot be changed at match time, I realized I could make a small
2453 improvement to matching performance by compiling OP_ALLANY instead of
2454 OP_ANY for "." when DOTALL was set, and then removing the runtime tests
2455 on the OP_ANY path.
2456
2457 17. Compiling pcretest on Windows with readline support failed without the
2458 following two fixes: (1) Make the unistd.h include conditional on
2459 HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
2460
2461 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
2462 ncurses library to be included for pcretest when ReadLine support is
2463 requested, but also to allow for it to be overridden. This patch came from
2464 Daniel Bergström.
2465
2466 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
2467 as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
2468 any errors with the current Unicode tables. Thanks to Peter Kankowski for
2469 spotting this.
2470
2471
2472 Version 7.6 28-Jan-08
2473 ---------------------
2474
2475 1. A character class containing a very large number of characters with
2476 codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
2477 overflow.
2478
2479 2. Patch to cut out the "long long" test in pcrecpp_unittest when
2480 HAVE_LONG_LONG is not defined.
2481
2482 3. Applied Christian Ehrlicher's patch to update the CMake build files to
2483 bring them up to date and include new features. This patch includes:
2484
2485 - Fixed PH's badly added libz and libbz2 support.
2486 - Fixed a problem with static linking.
2487 - Added pcredemo. [But later removed - see 7 below.]
2488 - Fixed dftables problem and added an option.
2489 - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
2490 HAVE_LONG_LONG.
2491 - Added readline support for pcretest.
2492 - Added an listing of the option settings after cmake has run.
2493
2494 4. A user submitted a patch to Makefile that makes it easy to create
2495 "pcre.dll" under mingw when using Configure/Make. I added stuff to
2496 Makefile.am that cause it to include this special target, without
2497 affecting anything else. Note that the same mingw target plus all
2498 the other distribution libraries and programs are now supported
2499 when configuring with CMake (see 6 below) instead of with
2500 Configure/Make.
2501
2502 5. Applied Craig's patch that moves no_arg into the RE class in the C++ code.
2503 This is an attempt to solve the reported problem "pcrecpp::no_arg is not
2504 exported in the Windows port". It has not yet been confirmed that the patch
2505 solves the problem, but it does no harm.
2506
2507 6. Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
2508 NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
2509 with CMake, and also correct the comment about stack recursion.
2510
2511 7. Remove the automatic building of pcredemo from the ./configure system and
2512 from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
2513 of a program that users should build themselves after PCRE is installed, so
2514 building it automatically is not really right. What is more, it gave
2515 trouble in some build environments.
2516
2517 8. Further tidies to CMakeLists.txt from Sheri and Christian.
2518
2519
2520 Version 7.5 10-Jan-08
2521 ---------------------
2522
2523 1. Applied a patch from Craig: "This patch makes it possible to 'ignore'
2524 values in parens when parsing an RE using the C++ wrapper."
2525
2526 2. Negative specials like \S did not work in character classes in UTF-8 mode.
2527 Characters greater than 255 were excluded from the class instead of being
2528 included.
2529
2530 3. The same bug as (2) above applied to negated POSIX classes such as
2531 [:^space:].
2532
2533 4. PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
2534 defined or documented. It seems to have been a typo for PCRE_STATIC, so
2535 I have changed it.
2536
2537 5. The construct (?&) was not diagnosed as a syntax error (it referenced the
2538 first named subpattern) and a construct such as (?&a) would reference the
2539 first named subpattern whose name started with "a" (in other words, the
2540 length check was missing). Both these problems are fixed. "Subpattern name
2541 expected" is now given for (?&) (a zero-length name), and this patch also
2542 makes it give the same error for \k'' (previously it complained that that
2543 was a reference to a non-existent subpattern).
2544
2545 6. The erroneous patterns (?+-a) and (?-+a) give different error messages;
2546 this is right because (?- can be followed by option settings as well as by
2547 digits. I have, however, made the messages clearer.
2548
2549 7. Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
2550 than the number used in the conditional) now cause a compile-time error.
2551 This is actually not compatible with Perl, which accepts such patterns, but
2552 treats the conditional as always being FALSE (as PCRE used to), but it
2553 seems to me that giving a diagnostic is better.
2554
2555 8. Change "alphameric" to the more common word "alphanumeric" in comments
2556 and messages.
2557
2558 9. Fix two occurrences of "backslash" in comments that should have been
2559 "backspace".
2560
2561 10. Remove two redundant lines of code that can never be obeyed (their function
2562 was moved elsewhere).
2563
2564 11. The program that makes PCRE's Unicode character property table had a bug
2565 which caused it to generate incorrect table entries for sequences of
2566 characters that have the same character type, but are in different scripts.
2567 It amalgamated them into a single range, with the script of the first of
2568 them. In other words, some characters were in the wrong script. There were
2569 thirteen such cases, affecting characters in the following ranges:
2570
2571 U+002b0 - U+002c1
2572 U+0060c - U+0060d
2573 U+0061e - U+00612
2574 U+0064b - U+0065e
2575 U+0074d - U+0076d
2576 U+01800 - U+01805
2577 U+01d00 - U+01d77
2578 U+01d9b - U+01dbf
2579 U+0200b - U+0200f
2580 U+030fc - U+030fe
2581 U+03260 - U+0327f
2582 U+0fb46 - U+0fbb1
2583 U+10450 - U+1049d
2584
2585 12. The -o option (show only the matching part of a line) for pcregrep was not
2586 compatible with GNU grep in that, if there was more than one match in a
2587 line, it showed only the first of them. It now behaves in the same way as
2588 GNU grep.
2589
2590 13. If the -o and -v options were combined for pcregrep, it printed a blank
2591 line for every non-matching line. GNU grep prints nothing, and pcregrep now
2592 does the same. The return code can be used to tell if there were any
2593 non-matching lines.
2594
2595 14. Added --file-offsets and --line-offsets to pcregrep.
2596
2597 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
2598 infinitely looping recursion. The bug was that positive lookaheads were not
2599 being skipped when checking for a possible empty match (negative lookaheads
2600 and both kinds of lookbehind were skipped).
2601
2602 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
2603 inclusion of <windows.h> to before rather than after the definition of
2604 INVALID_FILE_ATTRIBUTES (patch from David Byron).
2605
2606 17. Specifying a possessive quantifier with a specific limit for a Unicode
2607 character property caused pcre_compile() to compile bad code, which led at
2608 runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
2609 are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
2610 caused the error; without that there was no problem.
2611
2612 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
2613
2614 19. Added --enable-pcretest-libreadline.
2615
2616 20. In pcrecpp.cc, the variable 'count' was incremented twice in
2617 RE::GlobalReplace(). As a result, the number of replacements returned was
2618 double what it should be. I removed one of the increments, but Craig sent a
2619 later patch that removed the other one (the right fix) and added unit tests
2620 that check the return values (which was not done before).
2621
2622 21. Several CMake things:
2623
2624 (1) Arranged that, when cmake is used on Unix, the libraries end up with
2625 the names libpcre and libpcreposix, not just pcre and pcreposix.
2626
2627 (2) The above change means that pcretest and pcregrep are now correctly
2628 linked with the newly-built libraries, not previously installed ones.
2629
2630 (3) Added PCRE_SUPPORT_LIBREADLINE, PCRE_SUPPORT_LIBZ, PCRE_SUPPORT_LIBBZ2.
2631
2632 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
2633 crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
2634 UTF-8 newline character). The key issue is that the pattern starts .*;
2635 this means that the match must be either at the beginning, or after a
2636 newline. The bug was in the code for advancing after a failed match and
2637 checking that the new position followed a newline. It was not taking
2638 account of UTF-8 characters correctly.
2639
2640 23. PCRE was behaving differently from Perl in the way it recognized POSIX
2641 character classes. PCRE was not treating the sequence [:...:] as a
2642 character class unless the ... were all letters. Perl, however, seems to
2643 allow any characters between [: and :], though of course it rejects as
2644 unknown any "names" that contain non-letters, because all the known class
2645 names consist only of letters. Thus, Perl gives an error for [[:1234:]],
2646 for example, whereas PCRE did not - it did not recognize a POSIX character
2647 class. This seemed a bit dangerous, so the code has been changed to be
2648 closer to Perl. The behaviour is not identical to Perl, because PCRE will
2649 diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
2650 treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
2651 Perl does, and where it didn't before.
2652
2653 24. Rewrite so as to remove the single use of %n from pcregrep because in some
2654 Windows environments %n is disabled by default.
2655
2656
2657 Version 7.4 21-Sep-07
2658 ---------------------
2659
2660 1. Change 7.3/28 was implemented for classes by looking at the bitmap. This
2661 means that a class such as [\s] counted as "explicit reference to CR or
2662 LF". That isn't really right - the whole point of the change was to try to
2663 help when there was an actual mention of one of the two characters. So now
2664 the change happens only if \r or \n (or a literal CR or LF) character is
2665 encountered.
2666
2667 2. The 32-bit options word was also used for 6 internal flags, but the numbers
2668 of both had grown to the point where there were only 3 bits left.
2669 Fortunately, there was spare space in the data structure, and so I have
2670 moved the internal flags into a new 16-bit field to free up more option
2671 bits.
2672
2673 3. The appearance of (?J) at the start of a pattern set the DUPNAMES option,
2674 but did not set the internal JCHANGED flag - either of these is enough to
2675 control the way the "get" function works - but the PCRE_INFO_JCHANGED
2676 facility is supposed to tell if (?J) was ever used, so now (?J) at the
2677 start sets both bits.
2678
2679 4. Added options (at build time, compile time, exec time) to change \R from
2680 matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
2681
2682 5. doc/pcresyntax.html was missing from the distribution.
2683
2684 6. Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
2685 compatibility, even though it is no longer used.
2686
2687 7. Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
2688 strtoull to pcrecpp.cc to select the available functions in WIN32 when the
2689 windows.h file is present (where different names are used). [This was
2690 reversed later after testing - see 16 below.]
2691
2692 8. Changed all #include <config.h> to #include "config.h". There were also
2693 some further <pcre.h> cases that I changed to "pcre.h".
2694
2695 9. When pcregrep was used with the --colour option, it missed the line ending
2696 sequence off the lines that it output.
2697
2698 10. It was pointed out to me that arrays of string pointers cause lots of
2699 relocations when a shared library is dynamically loaded. A technique of
2700 using a single long string with a table of offsets can drastically reduce
2701 these. I have refactored PCRE in four places to do this. The result is
2702 dramatic:
2703
2704 Originally: 290
2705 After changing UCP table: 187
2706 After changing error message table: 43
2707 After changing table of "verbs" 36
2708 After changing table of Posix names 22
2709
2710 Thanks to the folks working on Gregex for glib for this insight.
2711
2712 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
2713 unicode-properties was also set.
2714
2715 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
2716
2717 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
2718 checked only for CRLF.
2719
2720 14. Added casts to pcretest.c to avoid compiler warnings.
2721
2722 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
2723
2724 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
2725 and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
2726 entirely. This removes changes made in 7 above.
2727
2728 17. The CMake files have been updated, and there is now more information about
2729 building with CMake in the NON-UNIX-USE document.
2730
2731
2732 Version 7.3 28-Aug-07
2733 ---------------------
2734
2735 1. In the rejigging of the build system that eventually resulted in 7.1, the
2736 line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
2737 brackets there is not right, since it causes compilers to look for an
2738 installed pcre.h, not the version that is in the source that is being
2739 compiled (which of course may be different). I have changed it back to:
2740
2741 #include "pcre.h"
2742
2743 I have a vague recollection that the change was concerned with compiling in
2744 different directories, but in the new build system, that is taken care of
2745 by the VPATH setting the Makefile.
2746
2747 2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
2748 when the subject happened to end in the byte 0x85 (e.g. if the last
2749 character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
2750 characters but of course it shouldn't be taken as a newline when it is part
2751 of another character. The bug was that, for an unlimited repeat of . in
2752 not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
2753 characters when looking for a newline.
2754
2755 3. A small performance improvement in the DOTALL UTF-8 mode .* case.
2756
2757 4. Debugging: adjusted the names of opcodes for different kinds of parentheses
2758 in debug output.
2759
2760 5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
2761 long printing in the pcrecpp unittest when running under MinGW.
2762
2763 6. ESC_K was left out of the EBCDIC table.
2764
2765 7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
2766 parentheses; I made it 1000, which seemed large enough. Unfortunately, the
2767 limit also applies to "virtual nesting" when a pattern is recursive, and in
2768 this case 1000 isn't so big. I have been able to remove this limit at the
2769 expense of backing off one optimization in certain circumstances. Normally,
2770 when pcre_exec() would call its internal match() function recursively and
2771 immediately return the result unconditionally, it uses a "tail recursion"
2772 feature to save stack. However, when a subpattern that can match an empty
2773 string has an unlimited repetition quantifier, it no longer makes this
2774 optimization. That gives it a stack frame in which to save the data for
2775 checking that an empty string has been matched. Previously this was taken
2776 from the 1000-entry workspace that had been reserved. So now there is no
2777 explicit limit, but more stack is used.
2778
2779 8. Applied Daniel's patches to solve problems with the import/export magic
2780 syntax that is required for Windows, and which was going wrong for the
2781 pcreposix and pcrecpp parts of the library. These were overlooked when this
2782 problem was solved for the main library.
2783
2784 9. There were some crude static tests to avoid integer overflow when computing
2785 the size of patterns that contain repeated groups with explicit upper
2786 limits. As the maximum quantifier is 65535, the maximum group length was
2787 set at 30,000 so that the product of these two numbers did not overflow a
2788 32-bit integer. However, it turns out that people want to use groups that
2789 are longer than 30,000 bytes (though not repeat them that many times).
2790 Change 7.0/17 (the refactoring of the way the pattern size is computed) has
2791 made it possible to implement the integer overflow checks in a much more
2792 dynamic way, which I have now done. The artificial limitation on group
2793 length has been removed - we now have only the limit on the total length of
2794 the compiled pattern, which depends on the LINK_SIZE setting.
2795
2796 10. Fixed a bug in the documentation for get/copy named substring when
2797 duplicate names are permitted. If none of the named substrings are set, the
2798 functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
2799 empty string.
2800
2801 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
2802 instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
2803 because the ] is interpreted as the first data character and the
2804 terminating ] is not found. PCRE has been made compatible with Perl in this
2805 regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
2806 cause memory overwriting.
2807
2808 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
2809 string has been matched (to stop an infinite loop). It was not recognizing
2810 a conditional subpattern that could match an empty string if that
2811 subpattern was within another subpattern. For example, it looped when
2812 trying to match (((?(1)X|))*) but it was OK with ((?(1)X|)*) where the
2813 condition was not nested. This bug has been fixed.
2814
2815 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
2816 past the start of the subject in the presence of bytes with the top bit
2817 set, for example "\x8aBCD".
2818
2819 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
2820 (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
2821
2822 14. Optimized (?!) to (*FAIL).
2823
2824 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
2825 This restricts code points to be within the range 0 to 0x10FFFF, excluding
2826 the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
2827 full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
2828 does: it's just the validity check that is more restrictive.
2829
2830 16. Inserted checks for integer overflows during escape sequence (backslash)
2831 processing, and also fixed erroneous offset values for syntax errors during
2832 backslash processing.
2833
2834 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
2835 for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
2836
2837 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
2838 caused an overrun.
2839
2840 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
2841 something other than just ASCII characters) inside a group that had an
2842 unlimited repeat caused a loop at compile time (while checking to see
2843 whether the group could match an empty string).
2844
2845 20. Debugging a pattern containing \p or \P could cause a crash. For example,
2846 [\P{Any}] did so. (Error in the code for printing property names.)
2847
2848 21. An orphan \E inside a character class could cause a crash.
2849
2850 22. A repeated capturing bracket such as (A)? could cause a wild memory
2851 reference during compilation.
2852
2853 23. There are several functions in pcre_compile() that scan along a compiled
2854 expression for various reasons (e.g. to see if it's fixed length for look
2855 behind). There were bugs in these functions when a repeated \p or \P was
2856 present in the pattern. These operators have additional parameters compared
2857 with \d, etc, and these were not being taken into account when moving along
2858 the compiled data. Specifically:
2859
2860 (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
2861 length.
2862
2863 (b) An item such as \pL+ within a repeated group could cause crashes or
2864 loops.
2865
2866 (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
2867 "reference to non-existent subpattern" error.
2868
2869 (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
2870
2871 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
2872 characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
2873
2874 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
2875
2876 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
2877 character were causing crashes (broken optimization).
2878
2879 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
2880 \p or \P) caused a compile-time loop.
2881
2882 28. More problems have arisen in unanchored patterns when CRLF is a valid line
2883 break. For example, the unstudied pattern [\r\n]A does not match the string
2884 "\r\nA" because change 7.0/46 below moves the current point on by two
2885 characters after failing to match at the start. However, the pattern \nA
2886 *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
2887 the same is true. There doesn't seem any very clean way out of this, but
2888 what I have chosen to do makes the common cases work: PCRE now takes note
2889 of whether there can be an explicit match for \r or \n anywhere in the
2890 pattern, and if so, 7.0/46 no longer applies. As part of this change,
2891 there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
2892 pattern has explicit CR or LF references.
2893
2894 29. Added (*CR) etc for changing newline setting at start of pattern.
2895
2896
2897 Version 7.2 19-Jun-07
2898 ---------------------
2899
2900 1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
2901 which is apparently normally available under Windows.
2902
2903 2. Re-jig the pcregrep tests with different newline settings in an attempt
2904 to make them independent of the local environment's newline setting.
2905
2906 3. Add code to configure.ac to remove -g from the CFLAGS default settings.
2907
2908 4. Some of the "internals" tests were previously cut out when the link size
2909 was not 2, because the output contained actual offsets. The recent new
2910 "Z" feature of pcretest means that these can be cut out, making the tests
2911 usable with all link sizes.
2912
2913 5. Implemented Stan Switzer's goto replacement for longjmp() when not using
2914 stack recursion. This gives a massive performance boost under BSD, but just
2915 a small improvement under Linux. However, it saves one field in the frame
2916 in all cases.
2917
2918 6. Added more features from the forthcoming Perl 5.10:
2919
2920 (a) (?-n) (where n is a string of digits) is a relative subroutine or
2921 recursion call. It refers to the nth most recently opened parentheses.
2922
2923 (b) (?+n) is also a relative subroutine call; it refers to the nth next
2924 to be opened parentheses.
2925
2926 (c) Conditions that refer to capturing parentheses can be specified
2927 relatively, for example, (?(-2)... or (?(+3)...
2928
2929 (d) \K resets the start of the current match so that everything before
2930 is not part of it.
2931
2932 (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
2933
2934 (f) \g{name} is another synonym - part of Perl 5.10's unification of
2935 reference syntax.
2936
2937 (g) (?| introduces a group in which the numbering of parentheses in each
2938 alternative starts with the same number.
2939
2940 (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
2941
2942 7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
2943 PCRE_INFO_JCHANGED.
2944
2945 8. A pattern such as (.*(.)?)* caused pcre_exec() to fail by either not
2946 terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
2947 for detecting groups that can match an empty string.
2948
2949 9. A pattern with a very large number of alternatives (more than several
2950 hundred) was running out of internal workspace during the pre-compile
2951 phase, where pcre_compile() figures out how much memory will be needed. A
2952 bit of new cunning has reduced the workspace needed for groups with
2953 alternatives. The 1000-alternative test pattern now uses 12 bytes of
2954 workspace instead of running out of the 4096 that are available.
2955
2956 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
2957
2958 11. Applied patch from Google to remove an optimization that didn't quite work.
2959 The report of the bug said:
2960
2961 pcrecpp::RE("a*").FullMatch("aaa") matches, while
2962 pcrecpp::RE("a*?").FullMatch("aaa") does not, and
2963 pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
2964
2965 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
2966 it matched the wrong number of bytes.
2967
2968
2969 Version 7.1 24-Apr-07
2970 ---------------------
2971
2972 1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
2973 that is more "standard", making use of automake and other Autotools. There
2974 is some re-arrangement of the files and adjustment of comments consequent
2975 on this.
2976
2977 2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
2978 for recursive directory scanning broke on some systems because the files
2979 are not scanned in any specific order and on different systems the order
2980 was different. A call to "sort" has been inserted into RunGrepTest for the
2981 approprate test as a short-term fix. In the longer term there may be an
2982 alternative.
2983
2984 3. I had an email from Eric Raymond about problems translating some of PCRE's
2985 man pages to HTML (despite the fact that I distribute HTML pages, some
2986 people do their own conversions for various reasons). The problems
2987 concerned the use of low-level troff macros .br and .in. I have therefore
2988 removed all such uses from the man pages (some were redundant, some could
2989 be replaced by .nf/.fi pairs). The 132html script that I use to generate
2990 HTML has been updated to handle .nf/.fi and to complain if it encounters
2991 .br or .in.
2992
2993 4. Updated comments in configure.ac that get placed in config.h.in and also
2994 arranged for config.h to be included in the distribution, with the name
2995 config.h.generic, for the benefit of those who have to compile without
2996 Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
2997
2998 5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
2999 Weber: (1) pcre_internal.h was missing some function renames; (2) updated
3000 makevp.bat for the current PCRE, using the additional files
3001 makevp_c.txt, makevp_l.txt, and pcregexp.pas.
3002
3003 6. A Windows user reported a minor discrepancy with test 2, which turned out
3004 to be caused by a trailing space on an input line that had got lost in his
3005 copy. The trailing space was an accident, so I've just removed it.
3006
3007 7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
3008 that is needed.
3009
3010 8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
3011 as "const" (a) because they are and (b) because it helps the PHP
3012 maintainers who have recently made a script to detect big data structures
3013 in the php code that should be moved to the .rodata section. I remembered
3014 to update Builducptable as well, so it won't revert if ucptable.h is ever
3015 re-created.
3016
3017 9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
3018 pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
3019 order to be able to cut out the UTF-8 tables in the latter when UTF-8
3020 support is not required. This saves 1.5-2K of code, which is important in
3021 some applications.
3022
3023 Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
3024 so as not to refer to the tables, even though these functions will never be
3025 called when UTF-8 support is disabled. Otherwise there are problems with a
3026 shared library.
3027
3028 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
3029
3030 (a) It was defining its arguments as char * instead of void *.
3031
3032 (b) It was assuming that all moves were upwards in memory; this was true
3033 a long time ago when I wrote it, but is no longer the case.
3034
3035 The emulated memove() is provided for those environments that have neither
3036 memmove() nor bcopy(). I didn't think anyone used it these days, but that
3037 is clearly not the case, as these two bugs were recently reported.
3038
3039 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
3040 and Detrail to create the HTML documentation, the .txt form of the man
3041 pages, and it removes trailing spaces from listed files. It also creates
3042 pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
3043 case, it wraps all the #defines with #ifndefs. This script should be run
3044 before "make dist".
3045
3046 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
3047 with Unicode property support.
3048
3049 (a) For a maximizing quantifier, if the two different cases of the
3050 character were of different lengths in their UTF-8 codings (there are
3051 some cases like this - I found 11), and the matching function had to
3052 back up over a mixture of the two cases, it incorrectly assumed they
3053 were both the same length.
3054
3055 (b) When PCRE was configured to use the heap rather than the stack for
3056 recursion during matching, it was not correctly preserving the data for
3057 the other case of a UTF-8 character when checking ahead for a match
3058 while processing a minimizing repeat. If the check also involved
3059 matching a wide character, but failed, corruption could cause an
3060 erroneous result when trying to check for a repeat of the original
3061 character.
3062
3063 13. Some tidying changes to the testing mechanism:
3064
3065 (a) The RunTest script now detects the internal link size and whether there
3066 is UTF-8 and UCP support by running ./pcretest -C instead of relying on
3067 values substituted by "configure". (The RunGrepTest script already did
3068 this for UTF-8.) The configure.ac script no longer substitutes the
3069 relevant variables.
3070
3071 (b) The debugging options /B and /D in pcretest show the compiled bytecode
3072 with length and offset values. This means that the output is different
3073 for different internal link sizes. Test 2 is skipped for link sizes
3074 other than 2 because of this, bypassing the problem. Unfortunately,
3075 there was also a test in test 3 (the locale tests) that used /B and
3076 failed for link sizes other than 2. Rather than cut the whole test out,
3077 I have added a new /Z option to pcretest that replaces the length and
3078 offset values with spaces. This is now used to make test 3 independent
3079 of link size. (Test 2 will be tidied up later.)
3080
3081 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
3082 segmentation fault instead of returning the appropriate error message.
3083
3084 15. In multiline mode when the newline sequence was set to "any", the pattern
3085 ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
3086 This doesn't seem right; it now treats the CRLF combination as the line
3087 ending, and so does not match in that case. It's only a pattern such as ^$
3088 that would hit this one: something like ^ABC$ would have failed after \r
3089 and then tried again after \r\n.
3090
3091 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
3092 in an attempt to make files that differ only in their line terminators
3093 compare equal. This works on Linux.
3094
3095 17. Under certain error circumstances pcregrep might try to free random memory
3096 as it exited. This is now fixed, thanks to valgrind.
3097
3098 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
3099 "abc\r\n\r\n", it found an unwanted second match after the second \r. This
3100 was because its rules for how to advance for /g after matching an empty
3101 string at the end of a line did not allow for this case. They now check for
3102 it specially.
3103
3104 20. pcretest is supposed to handle patterns and data of any length, by
3105 extending its buffers when necessary. It was getting this wrong when the
3106 buffer for a data line had to be extended.
3107
3108 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
3109 CRLF as a newline sequence.
3110
3111 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
3112 out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
3113 I have nevertheless tidied it up.
3114
3115 23. Added some casts to kill warnings from HP-UX ia64 compiler.
3116
3117 24. Added a man page for pcre-config.
3118
3119
3120 Version 7.0 19-Dec-06
3121 ---------------------
3122
3123 1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
3124 moving to gcc 4.1.1.
3125
3126 2. The -S option for pcretest uses setrlimit(); I had omitted to #include
3127 sys/time.h, which is documented as needed for this function. It doesn't
3128 seem to matter on Linux, but it showed up on some releases of OS X.
3129
3130 3. It seems that there are systems where bytes whose values are greater than
3131 127 match isprint() in the "C" locale. The "C" locale should be the
3132 default when a C program starts up. In most systems, only ASCII printing
3133 characters match isprint(). This difference caused the output from pcretest
3134 to vary, making some of the tests fail. I have changed pcretest so that:
3135
3136 (a) When it is outputting text in the compiled version of a pattern, bytes
3137 other than 32-126 are always shown as hex escapes.
3138
3139 (b) When it is outputting text that is a matched part of a subject string,
3140 it does the same, unless a different locale has been set for the match
3141 (using the /L modifier). In this case, it uses isprint() to decide.
3142
3143 4. Fixed a major bug that caused incorrect computation of the amount of memory
3144 required for a compiled pattern when options that changed within the
3145 pattern affected the logic of the preliminary scan that determines the
3146 length. The relevant options are -x, and -i in UTF-8 mode. The result was
3147 that the computed length was too small. The symptoms of this bug were
3148 either the PCRE error "internal error: code overflow" from pcre_compile(),
3149 or a glibc crash with a message such as "pcretest: free(): invalid next
3150 size (fast)". Examples of patterns that provoked this bug (shown in
3151 pcretest format) are:
3152
3153 /(?-x: )/x
3154 /(?x)(?-x: \s*#\s*)/
3155 /((?i)[\x{c0}])/8
3156 /(?i:[\x{c0}])/8
3157
3158 HOWEVER: Change 17 below makes this fix obsolete as the memory computation
3159 is now done differently.
3160
3161 5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
3162 wrapper classes; (b) implement a new function in the C++ scanner that is
3163 more efficient than the old way of doing things because it avoids levels of
3164 recursion in the regex matching; (c) add a paragraph to the documentation
3165 for the FullMatch() function.
3166
3167 6. The escape sequence \n was being treated as whatever was defined as
3168 "newline". Not only was this contrary to the documentation, which states
3169 that \n is character 10 (hex 0A), but it also went horribly wrong when
3170 "newline" was defined as CRLF. This has been fixed.
3171
3172 7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
3173 was being set to -1 for the "end of line" case (supposedly a value that no
3174 character can have). Though this value is never used (the check for end of
3175 line is "zero bytes in current character"), it caused compiler complaints.
3176 I've changed it to 0xffffffff.
3177
3178 8. In pcre_version.c, the version string was being built by a sequence of
3179 C macros that, in the event of PCRE_PRERELEASE being defined as an empty
3180 string (as it is for production releases) called a macro with an empty
3181 argument. The C standard says the result of this is undefined. The gcc
3182 compiler treats it as an empty string (which was what was wanted) but it is
3183 reported that Visual C gives an error. The source has been hacked around to
3184 avoid this problem.
3185
3186 9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
3187 builds of pcretest, and changed the call to _setmode() to use _O_BINARY
3188 instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
3189 of them did).
3190
3191 10. Originally, pcretest opened its input and output without "b"; then I was
3192 told that "b" was needed in some environments, so it was added for release
3193 5.0 to both the input and output. (It makes no difference on Unix-like
3194 systems.) Later I was told that it is wrong for the input on Windows. I've
3195 now abstracted the modes into two macros, to make it easier to fiddle with
3196 them, and removed "b" from the input mode under Windows.
3197
3198 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
3199
3200 12. Added -help and --help to pcretest as an official way of being reminded
3201 of the options.
3202
3203 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
3204 and pcrecpp.cc because they annoy compilers at high warning levels.
3205
3206 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
3207
3208 15. Fixed an occurrence of == in configure.ac that should have been = (shell
3209 scripts are not C programs :-) and which was not noticed because it works
3210 on Linux.
3211
3212 16. pcretest is supposed to handle any length of pattern and data line (as one
3213 line or as a continued sequence of lines) by extending its input buffer if
3214 necessary. This feature was broken for very long pattern lines, leading to
3215 a string of junk being passed to pcre_compile() if the pattern was longer
3216 than about 50K.
3217
3218 17. I have done a major re-factoring of the way pcre_compile() computes the
3219 amount of memory needed for a compiled pattern. Previously, there was code
3220 that made a preliminary scan of the pattern in order to do this. That was
3221 OK when PCRE was new, but as the facilities have expanded, it has become
3222 harder and harder to keep it in step with the real compile phase, and there
3223 have been a number of bugs (see for example, 4 above). I have now found a
3224 cunning way of running the real compile function in a "fake" mode that
3225 enables it to compute how much memory it would need, while actually only
3226 ever using a few hundred bytes of working memory and without too many
3227 tests of the mode. This should make future maintenance and development
3228 easier. A side effect of this work is that the limit of 200 on the nesting
3229 depth of parentheses has been removed (though this was never a serious
3230 limitation, I suspect). However, there is a downside: pcre_compile() now
3231 runs more slowly than before (30% or more, depending on the pattern). I
3232 hope this isn't a big issue. There is no effect on runtime performance.
3233
3234 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
3235 newline (only possible for the last line of a file) and it was a
3236 pattern that set a locale (followed by /Lsomething), pcretest crashed.
3237
3238 19. Added additional timing features to pcretest. (1) The -tm option now times
3239 matching only, not compiling. (2) Both -t and -tm can be followed, as a
3240 separate command line item, by a number that specifies the number of
3241 repeats to use when timing. The default is 50000; this gives better
3242 precision, but takes uncomfortably long for very large patterns.
3243
3244 20. Extended pcre_study() to be more clever in cases where a branch of a
3245 subpattern has no definite first character. For example, (a*|b*)[cd] would
3246 previously give no result from pcre_study(). Now it recognizes that the
3247 first character must be a, b, c, or d.
3248
3249 21. There was an incorrect error "recursive call could loop indefinitely" if
3250 a subpattern (or the entire pattern) that was being tested for matching an
3251 empty string contained only one non-empty item after a nested subpattern.
3252 For example, the pattern (?>\x{100}*)\d(?R) provoked this error
3253 incorrectly, because the \d was being skipped in the check.
3254
3255 22. The pcretest program now has a new pattern option /B and a command line
3256 option -b, which is equivalent to adding /B to every pattern. This causes
3257 it to show the compiled bytecode, without the additional information that
3258 -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
3259 is the same as /B/I).
3260
3261 23. A new optimization is now able automatically to treat some sequences such
3262 as a*b as a*+b. More specifically, if something simple (such as a character
3263 or a simple class like \d) has an unlimited quantifier, and is followed by
3264 something that cannot possibly match the quantified thing, the quantifier
3265 is automatically "possessified".
3266
3267 24. A recursive reference to a subpattern whose number was greater than 39
3268 went wrong under certain circumstances in UTF-8 mode. This bug could also
3269 have affected the operation of pcre_study().
3270
3271 25. Realized that a little bit of performance could be had by replacing
3272 (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
3273
3274 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
3275
3276 27. Possessive quantifiers such as a++ were previously implemented by turning
3277 them into atomic groups such as ($>a+). Now they have their own opcodes,
3278 which improves performance. This includes the automatically created ones
3279 from 23 above.
3280
3281 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
3282 lookahead was broken if it was not anchored. PCRE was mistakenly expecting
3283 the first matched character to be a colon. This applied both to named and
3284 numbered groups.
3285
3286 29. The ucpinternal.h header file was missing its idempotency #ifdef.
3287
3288 30. I was sent a "project" file called libpcre.a.dev which I understand makes
3289 building PCRE on Windows easier, so I have included it in the distribution.
3290
3291 31. There is now a check in pcretest against a ridiculously large number being
3292 returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
3293 loop, the loop is abandoned.
3294
3295 32. Forward references to subpatterns in conditions such as (?(2)...) where
3296 subpattern 2 is defined later cause pcre_compile() to search forwards in
3297 the pattern for the relevant set of parentheses. This search went wrong
3298 when there were unescaped parentheses in a character class, parentheses
3299 escaped with \Q...\E, or parentheses in a #-comment in /x mode.
3300
3301 33. "Subroutine" calls and backreferences were previously restricted to
3302 referencing subpatterns earlier in the regex. This restriction has now
3303 been removed.
3304
3305 34. Added a number of extra features that are going to be in Perl 5.10. On the
3306 whole, these are just syntactic alternatives for features that PCRE had
3307 previously implemented using the Python syntax or my own invention. The
3308 other formats are all retained for compatibility.
3309
3310 (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
3311 as (?P<name>...). The new forms, as well as being in Perl 5.10, are
3312 also .NET compatible.
3313
3314 (b) A recursion or subroutine call to a named group can now be defined as
3315 (?&name) as well as (?P>name).
3316
3317 (c) A backreference to a named group can now be defined as \k<name> or
3318 \k'name' as well as (?P=name). The new forms, as well as being in Perl
3319 5.10, are also .NET compatible.
3320
3321 (d) A conditional reference to a named group can now use the syntax
3322 (?(<name>) or (?('name') as well as (?(name).
3323
3324 (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
3325 groups (named and numbered) that are never evaluated inline, but can be
3326 called as "subroutines" from elsewhere. In effect, the DEFINE condition
3327 is always false. There may be only one alternative in such a group.
3328
3329 (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
3330 as the simple (?(R). The condition is true only if the most recent
3331 recursion is that of the given number or name. It does not search out
3332 through the entire recursion stack.
3333
3334 (g) The escape \gN or \g{N} has been added, where N is a positive or
3335 negative number, specifying an absolute or relative reference.
3336
3337 35. Tidied to get rid of some further signed/unsigned compiler warnings and
3338 some "unreachable code" warnings.
3339
3340 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
3341 things, this adds five new scripts.
3342
3343 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
3344 There were also incompatibilities regarding the handling of \Q..\E inside
3345 character classes, for example with patterns like [\Qa\E-\Qz\E] where the
3346 hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
3347
3348 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
3349 matches an empty string, and forcibly breaks the loop. There were bugs in
3350 this code in non-simple cases. For a pattern such as ^(a()*)* matched
3351 against aaaa the result was just "a" rather than "aaaa", for example. Two
3352 separate and independent bugs (that affected different cases) have been
3353 fixed.
3354
3355 39. Refactored the code to abolish the use of different opcodes for small
3356 capturing bracket numbers. This is a tidy that I avoided doing when I
3357 removed the limit on the number of capturing brackets for 3.5 back in 2001.
3358 The new approach is not only tidier, it makes it possible to reduce the
3359 memory needed to fix the previous bug (38).
3360
3361 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
3362 sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
3363 processing dot, circumflex, or dollar metacharacters, or #-comments in /x
3364 mode.
3365
3366 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
3367 report.
3368
3369 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
3370 copy construction and assignment in the C++ wrapper.
3371
3372 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
3373 couple of bugs that could have given wrong results in the "--newline=crlf"
3374 case.
3375
3376 44. Added a number of casts and did some reorganization of signed/unsigned int
3377 variables following suggestions from Dair Grant. Also renamed the variable
3378 "this" as "item" because it is a C++ keyword.
3379
3380 45. Arranged for dftables to add
3381
3382 #include "pcre_internal.h"
3383
3384 to pcre_chartables.c because without it, gcc 4.x may remove the array
3385 definition from the final binary if PCRE is built into a static library and
3386 dead code stripping is activated.
3387
3388 46. For an unanchored pattern, if a match attempt fails at the start of a
3389 newline sequence, and the newline setting is CRLF or ANY, and the next two
3390 characters are CRLF, advance by two characters instead of one.
3391
3392
3393 Version 6.7 04-Jul-06
3394 ---------------------
3395
3396 1. In order to handle tests when input lines are enormously long, pcretest has
3397 been re-factored so that it automatically extends its buffers when
3398 necessary. The code is crude, but this _is_ just a test program. The
3399 default size has been increased from 32K to 50K.
3400
3401 2. The code in pcre_study() was using the value of the re argument before
3402 testing it for NULL. (Of course, in any sensible call of the function, it
3403 won't be NULL.)
3404
3405 3. The memmove() emulation function in pcre_internal.h, which is used on
3406 systems that lack both memmove() and bcopy() - that is, hardly ever -
3407 was missing a "static" storage class specifier.
3408
3409 4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
3410 containing an extended class (one that cannot be represented by a bitmap
3411 because it contains high-valued characters or Unicode property items, e.g.
3412 [\pZ]). Almost always one would set UTF-8 mode when processing such a
3413 pattern, but PCRE should not loop if you do not (it no longer does).
3414 [Detail: two cases were found: (a) a repeated subpattern containing an
3415 extended class; (b) a recursive reference to a subpattern that followed a
3416 previous extended class. It wasn't skipping over the extended class
3417 correctly when UTF-8 mode was not set.]
3418
3419 5. A negated single-character class was not being recognized as fixed-length
3420 in lookbehind assertions such as (?<=[^f]), leading to an incorrect
3421 compile error "lookbehind assertion is not fixed length".
3422
3423 6. The RunPerlTest auxiliary script was showing an unexpected difference
3424 between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
3425 write a Perl script that can interpret lines of an input file either as
3426 byte characters or as UTF-8, which is what "perltest" was being required to
3427 do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
3428 can't do is switch easily at run time between having the "use utf8;" pragma
3429 or not. In the end, I fudged it by using the RunPerlTest script to insert
3430 "use utf8;" explicitly for the UTF-8 tests.
3431
3432 7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
3433 the end of the subject string, contrary to the documentation and to what
3434 Perl does. This was true of both matching functions. Now it matches only at
3435 the start of the subject and immediately after *internal* newlines.
3436
3437 8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
3438 a pointer to an int instead of a pointer to an unsigned long int. This
3439 caused problems on 64-bit systems.
3440
3441 9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
3442 instance of the 'standard' template library not being so standard".
3443
3444 10. There was no check on the number of named subpatterns nor the maximum
3445 length of a subpattern name. The product of these values is used to compute
3446 the size of the memory block for a compiled pattern. By supplying a very
3447 long subpattern name and a large number of named subpatterns, the size
3448 computation could be caused to overflow. This is now prevented by limiting
3449 the length of names to 32 characters, and the number of named subpatterns
3450 to 10,000.
3451
3452 11. Subpatterns that are repeated with specific counts have to be replicated in
3453 the compiled pattern. The size of memory for this was computed from the
3454 length of the subpattern and the repeat count. The latter is limited to
3455 65535, but there was no limit on the former, meaning that integer overflow
3456 could in principle occur. The compiled length of a repeated subpattern is
3457 now limited to 30,000 bytes in order to prevent this.
3458
3459 12. Added the optional facility to have named substrings with the same name.
3460
3461 13. Added the ability to use a named substring as a condition, using the
3462 Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
3463 are numbers (not recommended). Forward references are permitted.
3464
3465 14. Added forward references in named backreferences (if you see what I mean).
3466
3467 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
3468 pattern could run off the end of the subject. For example, the pattern
3469 "(?s)(.{1,5})"8 did this with the subject "ab".
3470
3471 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
3472 PCRE_CASELESS was set when matching characters that were quantified with ?
3473 or *.
3474
3475 17. A character class other than a single negated character that had a minimum
3476 but no maximum quantifier - for example [ab]{6,} - was not handled
3477 correctly by pce_dfa_exec(). It would match only one character.
3478
3479 18. A valid (though odd) pattern that looked like a POSIX character
3480 class but used an invalid character after [ (for example [[,abc,]]) caused
3481 pcre_compile() to give the error "Failed: internal error: code overflow" or
3482 in some cases to crash with a glibc free() error. This could even happen if
3483 the pattern terminated after [[ but there just happened to be a sequence of
3484 letters, a binary zero, and a closing ] in the memory that followed.
3485
3486 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
3487 over the years. Originally (before any Unicode support), just the bottom 8
3488 bits were taken. Thus, for example, \500 really meant \100. Nowadays the
3489 output from "man perlunicode" includes this:
3490
3491 The regular expression compiler produces polymorphic opcodes. That
3492 is, the pattern adapts to the data and automatically switches to
3493 the Unicode character scheme when presented with Unicode data--or
3494 instead uses a traditional byte scheme when presented with byte
3495 data.
3496
3497 Sadly, a wide octal escape does not cause a switch, and in a string with
3498 no other multibyte characters, these octal escapes are treated as before.
3499 Thus, in Perl, the pattern /\500/ actually matches \100 but the pattern
3500 /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
3501 Unicode string.
3502
3503 I have not perpetrated such confusion in PCRE. Up till now, it took just
3504 the bottom 8 bits, as in old Perl. I have now made octal escapes with
3505 values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
3506 translate to the appropriate multibyte character.
3507
3508 29. Applied some refactoring to reduce the number of warnings from Microsoft
3509 and Borland compilers. This has included removing the fudge introduced
3510 seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
3511 a warning about an unused variable.
3512
3513 21. PCRE has not included VT (character 0x0b) in the set of whitespace
3514 characters since release 4.0, because Perl (from release 5.004) does not.
3515 [Or at least, is documented not to: some releases seem to be in conflict
3516 with the documentation.] However, when a pattern was studied with
3517 pcre_study() and all its branches started with \s, PCRE still included VT
3518 as a possible starting character. Of course, this did no harm; it just
3519 caused an unnecessary match attempt.
3520
3521 22. Removed a now-redundant internal flag bit that recorded the fact that case
3522 dependency changed within the pattern. This was once needed for "required
3523 byte" processing, but is no longer used. This recovers a now-scarce options
3524 bit. Also moved the least significant internal flag bit to the most-
3525 significant bit of the word, which was not previously used (hangover from
3526 the days when it was an int rather than a uint) to free up another bit for
3527 the future.
3528
3529 23. Added support for CRLF line endings as well as CR and LF. As well as the
3530 default being selectable at build time, it can now be changed at runtime
3531 via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
3532 specify that it is scanning data with non-default line endings.
3533
3534 24. Changed the definition of CXXLINK to make it agree with the definition of
3535 LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
3536
3537 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
3538 recursions. This makes a big different to stack usage for some patterns.
3539
3540 26. If a subpattern containing a named recursion or subroutine reference such
3541 as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
3542 the space required for the compiled pattern went wrong and gave too small a
3543 value. Depending on the environment, this could lead to "Failed: internal
3544 error: code overflow at offset 49" or "glibc detected double free or
3545 corruption" errors.
3546
3547 27. Applied patches from Google (a) to support the new newline modes and (b) to
3548 advance over multibyte UTF-8 characters in GlobalReplace.
3549
3550 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
3551 difference for some implementation of PCRE in some Windows version.
3552
3553 29. Added some extra testing facilities to pcretest:
3554
3555 \q<number> in a data line sets the "match limit" value
3556 \Q<number> in a data line sets the "match recursion limt" value
3557 -S <number> sets the stack size, where <number> is in megabytes
3558
3559 The -S option isn't available for Windows.
3560
3561
3562 Version 6.6 06-Feb-06
3563 ---------------------
3564
3565 1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
3566 in pcreposix.h. I have copied the definition from pcre.h.
3567
3568 2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
3569 because pcre.h is no longer a built file.
3570
3571 3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
3572 not normally included in the compiled code.
3573
3574
3575 Version 6.5 01-Feb-06
3576 ---------------------
3577
3578 1. When using the partial match feature with pcre_dfa_exec(), it was not
3579 anchoring the second and subsequent partial matches at the new starting
3580 point. This could lead to incorrect results. For example, with the pattern
3581 /1234/, partially matching against "123" and then "a4" gave a match.
3582
3583 2. Changes to pcregrep:
3584
3585 (a) All non-match returns from pcre_exec() were being treated as failures
3586 to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
3587 error message is output. Some extra information is given for the
3588 PCRE_ERROR_MATCHLIMIT and PCRE_ERROR_RECURSIONLIMIT errors, which are
3589 probably the only errors that are likely to be caused by users (by
3590 specifying a regex that has nested indefinite repeats, for instance).
3591 If there are more than 20 of these errors, pcregrep is abandoned.
3592
3593 (b) A binary zero was treated as data while matching, but terminated the
3594 output line if it was written out. This has been fixed: binary zeroes
3595 are now no different to any other data bytes.
3596
3597 (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
3598 used to set a locale for matching. The --locale=xxxx long option has
3599 been added (no short equivalent) to specify a locale explicitly on the
3600 pcregrep command, overriding the environment variables.
3601
3602 (d) When -B was used with -n, some line numbers in the output were one less
3603 than they should have been.
3604
3605 (e) Added the -o (--only-matching) option.
3606
3607 (f) If -A or -C was used with -c (count only), some lines of context were
3608 accidentally printed for the final match.
3609
3610 (g) Added the -H (--with-filename) option.
3611
3612 (h) The combination of options -rh failed to suppress file names for files
3613 that were found from directory arguments.
3614
3615 (i) Added the -D (--devices) and -d (--directories) options.
3616
3617 (j) Added the -F (--fixed-strings) option.
3618
3619 (k) Allow "-" to be used as a file name for -f as well as for a data file.
3620
3621 (l) Added the --colo(u)r option.
3622
3623 (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
3624 is not present by default.
3625
3626 3. A nasty bug was discovered in the handling of recursive patterns, that is,
3627 items such as (?R) or (?1), when the recursion could match a number of
3628 alternatives. If it matched one of the alternatives, but subsequently,
3629 outside the recursion, there was a failure, the code tried to back up into
3630 the recursion. However, because of the way PCRE is implemented, this is not
3631 possible, and the result was an incorrect result from the match.
3632
3633 In order to prevent this happening, the specification of recursion has
3634 been changed so that all such subpatterns are automatically treated as
3635 atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
3636
3637 4. I had overlooked the fact that, in some locales, there are characters for
3638 which isalpha() is true but neither isupper() nor islower() are true. In
3639 the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
3640 and ordfeminine) are like this. This affected the treatment of \w and \W
3641 when they appeared in character classes, but not when they appeared outside
3642 a character class. The bit map for "word" characters is now created
3643 separately from the results of isalnum() instead of just taking it from the
3644 upper, lower, and digit maps. (Plus the underscore character, of course.)
3645
3646 5. The above bug also affected the handling of POSIX character classes such as
3647 [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
3648 permanent tables. Instead, the bit maps for such a class were previously
3649 created as the appropriate unions of the upper, lower, and digit bitmaps.
3650 Now they are created by subtraction from the [[:word:]] class, which has
3651 its own bitmap.
3652
3653 6. The [[:blank:]] character class matches horizontal, but not vertical space.
3654 It is created by subtracting the vertical space characters (\x09, \x0a,
3655 \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
3656 subtraction was done in the overall bitmap for a character class, meaning
3657 that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
3658 be recognized. This bug has been fixed.
3659
3660 7. Patches from the folks at Google:
3661
3662 (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
3663 real life, but is still worth protecting against".
3664
3665 (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
3666 regular expressions".
3667
3668 (c) pcre_scanner.cc: avoid use of std::count() because not all systems
3669 have it.
3670
3671 (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
3672 "configure" and the latter not, in order to fix a problem somebody had
3673 with compiling the Arg class on HP-UX.
3674
3675 (e) Improve the error-handling of the C++ wrapper a little bit.
3676
3677 (f) New tests for checking recursion limiting.
3678
3679 8. The pcre_memmove() function, which is used only if the environment does not
3680 have a standard memmove() function (and is therefore rarely compiled),
3681 contained two bugs: (a) use of int instead of size_t, and (b) it was not
3682 returning a result (though PCRE never actually uses the result).
3683
3684 9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
3685 large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
3686 returned instead of calling malloc() with an overflowing number that would
3687 most likely cause subsequent chaos.
3688
3689 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
3690
3691 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
3692 with this option is matched, the nmatch and pmatch options of regexec() are
3693 ignored.
3694
3695 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
3696 provided in case anyone wants to the the POSIX interface with UTF-8
3697 strings.
3698
3699 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
3700 C++ linking (needed for some HP-UX environments).
3701
3702 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
3703 (unused parameter) and in the pcre_printint() function (omitted "default"
3704 switch label when the default is to do nothing).
3705
3706 15. Added some code to make it possible, when PCRE is compiled as a C++
3707 library, to replace subject pointers for pcre_exec() with a smart pointer
3708 class, thus making it possible to process discontinuous strings.
3709
3710 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
3711 much the same function. They were added by different people who were trying
3712 to make PCRE easy to compile on non-Unix systems. It has been suggested
3713 that PCRE_EXPORT be abolished now that there is more automatic apparatus
3714 for compiling on Windows systems. I have therefore replaced it with
3715 PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
3716 defaults to "extern" for C or "extern C" for C++, which works fine on
3717 Unix-like systems. It is now possible to override the value of PCRE_DATA_
3718 SCOPE with something explicit in config.h. In addition:
3719
3720 (a) pcreposix.h still had just "extern" instead of either of these macros;
3721 I have replaced it with PCRE_DATA_SCOPE.
3722
3723 (b) Functions such as _pcre_xclass(), which are internal to the library,
3724 but external in the C sense, all had PCRE_EXPORT in their definitions.
3725 This is apparently wrong for the Windows case, so I have removed it.
3726 (It makes no difference on Unix-like systems.)
3727
3728 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
3729 of recursive calls to match(). This is different to MATCH_LIMIT because
3730 that limits the total number of calls to match(), not all of which increase
3731 the depth of recursion. Limiting the recursion depth limits the amount of
3732 stack (or heap if NO_RECURSE is set) that is used. The default can be set
3733 when PCRE is compiled, and changed at run time. A patch from Google adds
3734 this functionality to the C++ interface.
3735
3736 18. Changes to the handling of Unicode character properties:
3737
3738 (a) Updated the table to Unicode 4.1.0.
3739
3740 (b) Recognize characters that are not in the table as "Cn" (undefined).
3741
3742 (c) I revised the way the table is implemented to a much improved format
3743 which includes recognition of ranges. It now supports the ranges that
3744 are defined in UnicodeData.txt, and it also amalgamates other
3745 characters into ranges. This has reduced the number of entries in the
3746 table from around 16,000 to around 3,000, thus reducing its size
3747 considerably. I realized I did not need to use a tree structure after
3748 all - a binary chop search is just as efficient. Having reduced the
3749 number of entries, I extended their size from 6 bytes to 8 bytes to
3750 allow for more data.
3751
3752 (d) Added support for Unicode script names via properties such as \p{Han}.
3753
3754 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
3755 matching that character.
3756
3757 20. When matching a repeated Unicode property with a minimum greater than zero,
3758 (for example \pL{2,}), PCRE could look past the end of the subject if it
3759 reached it while seeking the minimum number of characters. This could
3760 happen only if some of the characters were more than one byte long, because
3761 there is a check for at least the minimum number of bytes.
3762
3763 21. Refactored the implementation of \p and \P so as to be more general, to
3764 allow for more different types of property in future. This has changed the
3765 compiled form incompatibly. Anybody with saved compiled patterns that use
3766 \p or \P will have to recompile them.
3767
3768 22. Added "Any" and "L&" to the supported property types.
3769
3770 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
3771 but give a compile time error if the value is greater than 0xff.
3772
3773 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
3774 accidentally not being installed or uninstalled.
3775
3776 25. The pcre.h file was built from pcre.h.in, but the only changes that were
3777 made were to insert the current release number. This seemed silly, because
3778 it made things harder for people building PCRE on systems that don't run
3779 "configure". I have turned pcre.h into a distributed file, no longer built
3780 by "configure", with the version identification directly included. There is
3781 no longer a pcre.h.in file.
3782
3783 However, this change necessitated a change to the pcre-config script as
3784 well. It is built from pcre-config.in, and one of the substitutions was the
3785 release number. I have updated configure.ac so that ./configure now finds
3786 the release number by grepping pcre.h.
3787
3788 26. Added the ability to run the tests under valgrind.
3789
3790
3791 Version 6.4 05-Sep-05
3792 ---------------------
3793
3794 1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
3795 "--" to be printed when multiple files were scanned, even when none of the
3796 -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
3797 consider it to be a bug, and have restored the previous behaviour.
3798
3799 2. A couple of code tidies to get rid of compiler warnings.
3800
3801 3. The pcretest program used to cheat by referring to symbols in the library
3802 whose names begin with _pcre_. These are internal symbols that are not
3803 really supposed to be visible externally, and in some environments it is
3804 possible to suppress them. The cheating is now confined to including
3805 certain files from the library's source, which is a bit cleaner.
3806
3807 4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
3808 file's purpose clearer.
3809
3810 5. Reorganized pcre_ucp_findchar().
3811
3812
3813 Version 6.3 15-Aug-05
3814 ---------------------
3815
3816 1. The file libpcre.pc.in did not have general read permission in the tarball.
3817
3818 2. There were some problems when building without C++ support:
3819
3820 (a) If C++ support was not built, "make install" and "make test" still
3821 tried to test it.
3822
3823 (b) There were problems when the value of CXX was explicitly set. Some
3824 changes have been made to try to fix these, and ...
3825
3826 (c) --disable-cpp can now be used to explicitly disable C++ support.
3827
3828 (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
3829 backslash in a target when C++ was disabled. This confuses some
3830 versions of "make", apparently. Using an intermediate variable solves
3831 this. (Same for CPP_LOBJ.)
3832
3833 3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
3834 (non-Windows) now includes $(CFLAGS) because these flags are sometimes
3835 necessary on certain architectures.
3836
3837 4. Added a setting of -export-symbols-regex to the link command to remove
3838 those symbols that are exported in the C sense, but actually are local
3839 within the library, and not documented. Their names all begin with
3840 "_pcre_". This is not a perfect job, because (a) we have to except some
3841 symbols that pcretest ("illegally") uses, and (b) the facility isn't always
3842 available (and never for static libraries). I have made a note to try to
3843 find a way round (a) in the future.
3844
3845
3846 Version 6.2 01-Aug-05
3847 ---------------------
3848
3849 1. There was no test for integer overflow of quantifier values. A construction
3850 such as {1111111111111111} would give undefined results. What is worse, if
3851 a minimum quantifier for a parenthesized subpattern overflowed and became
3852 negative, the calculation of the memory size went wrong. This could have
3853 led to memory overwriting.
3854
3855 2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
3856
3857 3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
3858 operating environments where this matters.
3859
3860 4. Applied Giuseppe Maxia's patch to add additional features for controlling
3861 PCRE options from within the C++ wrapper.
3862
3863 5. Named capturing subpatterns were not being correctly counted when a pattern
3864 was compiled. This caused two problems: (a) If there were more than 100
3865 such subpatterns, the calculation of the memory needed for the whole
3866 compiled pattern went wrong, leading to an overflow error. (b) Numerical
3867 back references of the form \12, where the number was greater than 9, were
3868 not recognized as back references, even though there were sufficient
3869 previous subpatterns.
3870
3871 6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
3872 versions of gcc, e.g. 2.95.4.
3873
3874
3875 Version 6.1 21-Jun-05
3876 ---------------------
3877
3878 1. There was one reference to the variable "posix" in pcretest.c that was not
3879 surrounded by "#if !defined NOPOSIX".
3880
3881 2. Make it possible to compile pcretest without DFA support, UTF8 support, or
3882 the cross-check on the old pcre_info() function, for the benefit of the
3883 cut-down version of PCRE that is currently imported into Exim.
3884
3885 3. A (silly) pattern starting with (?i)(?-i) caused an internal space
3886 allocation error. I've done the easy fix, which wastes 2 bytes for sensible
3887 patterns that start (?i) but I don't think that matters. The use of (?i) is
3888 just an example; this all applies to the other options as well.
3889
3890 4. Since libtool seems to echo the compile commands it is issuing, the output
3891 from "make" can be reduced a bit by putting "@" in front of each libtool
3892 compile command.
3893
3894 5. Patch from the folks at Google for configure.in to be a bit more thorough
3895 in checking for a suitable C++ installation before trying to compile the
3896 C++ stuff. This should fix a reported problem when a compiler was present,
3897 but no suitable headers.
3898
3899 6. The man pages all had just "PCRE" as their title. I have changed them to
3900 be the relevant file name. I have also arranged that these names are
3901 retained in the file doc/pcre.txt, which is a concatenation in text format
3902 of all the man pages except the little individual ones for each function.
3903
3904 7. The NON-UNIX-USE file had not been updated for the different set of source
3905 files that come with release 6. I also added a few comments about the C++
3906 wrapper.
3907
3908
3909 Version 6.0 07-Jun-05
3910 ---------------------
3911
3912 1. Some minor internal re-organization to help with my DFA experiments.
3913
3914 2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
3915 didn't matter for the library itself when fully configured, but did matter
3916 when compiling without UCP support, or within Exim, where the ucp files are
3917 not imported.
3918
3919 3. Refactoring of the library code to split up the various functions into
3920 different source modules. The addition of the new DFA matching code (see
3921 below) to a single monolithic source would have made it really too
3922 unwieldy, quite apart from causing all the code to be include in a
3923 statically linked application, when only some functions are used. This is
3924 relevant even without the DFA addition now that patterns can be compiled in
3925 one application and matched in another.
3926
3927 The downside of splitting up is that there have to be some external
3928 functions and data tables that are used internally in different modules of
3929 the library but which are not part of the API. These have all had their
3930 names changed to start with "_pcre_" so that they are unlikely to clash
3931 with other external names.
3932
3933 4. Added an alternate matching function, pcre_dfa_exec(), which matches using
3934 a different (DFA) algorithm. Although it is slower than the original
3935 function, it does have some advantages for certain types of matching
3936 problem.
3937
3938 5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
3939 including restarting after a partial match.
3940
3941 6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
3942 defined when compiling for Windows was sent to me. I have put it into the
3943 code, though I have no means of testing or verifying it.
3944
3945 7. Added the pcre_refcount() auxiliary function.
3946
3947 8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
3948 match before or at the first newline in the subject string. In pcretest,
3949 the /f option on a pattern can be used to set this.
3950
3951 9. A repeated \w when used in UTF-8 mode with characters greater than 256
3952 would behave wrongly. This has been present in PCRE since release 4.0.
3953
3954 10. A number of changes to the pcregrep command:
3955
3956 (a) Refactored how -x works; insert ^(...)$ instead of setting
3957 PCRE_ANCHORED and checking the length, in preparation for adding
3958 something similar for -w.
3959
3960 (b) Added the -w (match as a word) option.
3961
3962 (c) Refactored the way lines are read and buffered so as to have more
3963 than one at a time available.
3964
3965 (d) Implemented a pcregrep test script.
3966
3967 (e) Added the -M (multiline match) option. This allows patterns to match
3968 over several lines of the subject. The buffering ensures that at least
3969 8K, or the rest of the document (whichever is the shorter) is available
3970 for matching (and similarly the previous 8K for lookbehind assertions).
3971
3972 (f) Changed the --help output so that it now says
3973
3974 -w, --word-regex(p)
3975
3976 instead of two lines, one with "regex" and the other with "regexp"
3977 because that confused at least one person since the short forms are the
3978 same. (This required a bit of code, as the output is generated
3979 automatically from a table. It wasn't just a text change.)
3980
3981 (g) -- can be used to terminate pcregrep options if the next thing isn't an
3982 option but starts with a hyphen. Could be a pattern or a path name
3983 starting with a hyphen, for instance.
3984
3985 (h) "-" can be given as a file name to represent stdin.
3986
3987 (i) When file names are being printed, "(standard input)" is used for
3988 the standard input, for compatibility with GNU grep. Previously
3989 "<stdin>" was used.
3990
3991 (j) The option --label=xxx can be used to supply a name to be used for
3992 stdin when file names are being printed. There is no short form.
3993
3994 (k) Re-factored the options decoding logic because we are going to add
3995 two more options that take data. Such options can now be given in four
3996 different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
3997
3998 (l) Added the -A, -B, and -C options for requesting that lines of context
3999 around matches be printed.
4000
4001 (m) Added the -L option to print the names of files that do not contain
4002 any matching lines, that is, the complement of -l.
4003
4004 (n) The return code is 2 if any file cannot be opened, but pcregrep does
4005 continue to scan other files.
4006
4007 (o) The -s option was incorrectly implemented. For compatibility with other
4008 greps, it now suppresses the error message for a non-existent or non-
4009 accessible file (but not the return code). There is a new option called
4010 -q that suppresses the output of matching lines, which was what -s was
4011 previously doing.
4012
4013 (p) Added --include and --exclude options to specify files for inclusion
4014 and exclusion when recursing.
4015
4016 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
4017 Hopefully, it now does.
4018
4019 12. Missing cast in pcre_study().
4020
4021 13. Added an "uninstall" target to the makefile.
4022
4023 14. Replaced "extern" in the function prototypes in Makefile.in with
4024 "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
4025 world, but is set differently for Windows.
4026
4027 15. Added a second compiling function called pcre_compile2(). The only
4028 difference is that it has an extra argument, which is a pointer to an
4029 integer error code. When there is a compile-time failure, this is set
4030 non-zero, in addition to the error test pointer being set to point to an
4031 error message. The new argument may be NULL if no error number is required
4032 (but then you may as well call pcre_compile(), which is now just a
4033 wrapper). This facility is provided because some applications need a
4034 numeric error indication, but it has also enabled me to tidy up the way
4035 compile-time errors are handled in the POSIX wrapper.
4036
4037 16. Added VPATH=.libs to the makefile; this should help when building with one
4038 prefix path and installing with another. (Or so I'm told by someone who
4039 knows more about this stuff than I do.)
4040
4041 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
4042 passes PCRE_DOTALL to the pcre_compile() function, making the "." character
4043 match everything, including newlines. This is not POSIX-compatible, but
4044 somebody wanted the feature. From pcretest it can be activated by using
4045 both the P and the s flags.
4046
4047 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
4048
4049 19. libpcre.pc was being incorrectly installed as executable.
4050
4051 20. A couple of places in pcretest check for end-of-line by looking for '\n';
4052 it now also looks for '\r' so that it will work unmodified on Windows.
4053
4054 21. Added Google's contributed C++ wrapper to the distribution.
4055
4056 22. Added some untidy missing memory free() calls in pcretest, to keep
4057 Electric Fence happy when testing.
4058
4059
4060
4061 Version 5.0 13-Sep-04
4062 ---------------------
4063
4064 1. Internal change: literal characters are no longer packed up into items
4065 containing multiple characters in a single byte-string. Each character
4066 is now matched using a separate opcode. However, there may be more than one
4067 byte in the character in UTF-8 mode.
4068
4069 2. The pcre_callout_block structure has two new fields: pattern_position and
4070 next_item_length. These contain the offset in the pattern to the next match
4071 item, and its length, respectively.
4072
4073 3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
4074 insertion of callouts before each pattern item. Added the /C option to
4075 pcretest to make use of this.
4076
4077 4. On the advice of a Windows user, the lines
4078
4079 #if defined(_WIN32) || defined(WIN32)
4080 _setmode( _fileno( stdout ), 0x8000 );
4081 #endif /* defined(_WIN32) || defined(WIN32) */
4082
4083 have been added to the source of pcretest. This apparently does useful
4084 magic in relation to line terminators.
4085
4086 5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
4087 for the benefit of those environments where the "b" makes a difference.
4088
4089 6. The icc compiler has the same options as gcc, but "configure" doesn't seem
4090 to know about it. I have put a hack into configure.in that adds in code
4091 to set GCC=yes if CC=icc. This seems to end up at a point in the
4092 generated configure script that is early enough to affect the setting of
4093 compiler options, which is what is needed, but I have no means of testing
4094 whether it really works. (The user who reported this had patched the
4095 generated configure script, which of course I cannot do.)
4096
4097 LATER: After change 22 below (new libtool files), the configure script
4098 seems to know about icc (and also ecc). Therefore, I have commented out
4099 this hack in configure.in.
4100
4101 7. Added support for pkg-config (2 patches were sent in).
4102
4103 8. Negated POSIX character classes that used a combination of internal tables
4104 were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
4105 [[:^ascii]]. Typically, they would match almost any characters. The other
4106 POSIX classes were not broken in this way.
4107
4108 9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
4109 to find the match, as PCRE was deluded into thinking that the match had to
4110 start at the start point or following a newline. The same bug applied to
4111 patterns with negative forward assertions or any backward assertions
4112 preceding ".*" at the start, unless the pattern required a fixed first
4113 character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
4114
4115 10. In UTF-8 mode, when moving forwards in the subject after a failed match
4116 starting at the last subject character, bytes beyond the end of the subject
4117 string were read.
4118
4119 11. Renamed the variable "class" as "classbits" to make life easier for C++
4120 users. (Previously there was a macro definition, but it apparently wasn't
4121 enough.)
4122
4123 12. Added the new field "tables" to the extra data so that tables can be passed
4124 in at exec time, or the internal tables can be re-selected. This allows
4125 a compiled regex to be saved and re-used at a later time by a different
4126 program that might have everything at different addresses.
4127
4128 13. Modified the pcre-config script so that, when run on Solaris, it shows a
4129 -R library as well as a -L library.
4130
4131 14. The debugging options of pcretest (-d on the command line or D on a
4132 pattern) showed incorrect output for anything following an extended class
4133 that contained multibyte characters and which was followed by a quantifier.
4134
4135 15. Added optional support for general category Unicode character properties
4136 via the \p, \P, and \X escapes. Unicode property support implies UTF-8
4137 support. It adds about 90K to the size of the library. The meanings of the
4138 inbuilt class escapes such as \d and \s have NOT been changed.
4139
4140 16. Updated pcredemo.c to include calls to free() to release the memory for the
4141 compiled pattern.
4142
4143 17. The generated file chartables.c was being created in the source directory
4144 instead of in the building directory. This caused the build to fail if the
4145 source directory was different from the building directory, and was
4146 read-only.
4147
4148 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
4149 file. No doubt somebody will tell me if they don't make sense... Also added
4150 Dan Mooney's comments about building on OpenVMS.
4151
4152 19. Added support for partial matching via the PCRE_PARTIAL option for
4153 pcre_exec() and the \P data escape in pcretest.
4154
4155 20. Extended pcretest with 3 new pattern features:
4156
4157 (i) A pattern option of the form ">rest-of-line" causes pcretest to
4158 write the compiled pattern to the file whose name is "rest-of-line".
4159 This is a straight binary dump of the data, with the saved pointer to
4160 the character tables forced to be NULL. The study data, if any, is
4161 written too. After writing, pcretest reads a new pattern.
4162
4163 (ii) If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
4164 compiled pattern from the given file. There must not be any
4165 occurrences of "<" in the file name (pretty unlikely); if there are,
4166 pcretest will instead treat the initial "<" as a pattern delimiter.
4167 After reading in the pattern, pcretest goes on to read data lines as
4168 usual.
4169
4170 (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
4171 and 16-bit fields in a compiled pattern, to simulate a pattern that
4172 was compiled on a host of opposite endianness.
4173
4174 21. The pcre-exec() function can now cope with patterns that were compiled on
4175 hosts of opposite endianness, with this restriction:
4176
4177 As for any compiled expression that is saved and used later, the tables
4178 pointer field cannot be preserved; the extra_data field in the arguments
4179 to pcre_exec() should be used to pass in a tables address if a value
4180 other than the default internal tables were used at compile time.
4181
4182 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
4183 now diagnosed as an error. Previously, most of the time, a negative number
4184 would have been treated as zero, but if in addition "ovector" was passed as
4185 NULL, a crash could occur.
4186
4187 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
4188 new versions from the libtool 1.5 distribution (the last one is a copy of
4189 a file called libtool.m4). This seems to have fixed the need to patch
4190 "configure" to support Darwin 1.3 (which I used to do). However, I still
4191 had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
4192 workstation).
4193
4194 24. Changed the PCRE licence to be the more standard "BSD" licence.
4195
4196
4197 Version 4.5 01-Dec-03
4198 ---------------------
4199
4200 1. There has been some re-arrangement of the code for the match() function so
4201 that it can be compiled in a version that does not call itself recursively.
4202 Instead, it keeps those local variables that need separate instances for
4203 each "recursion" in a frame on the heap, and gets/frees frames whenever it
4204 needs to "recurse". Keeping track of where control must go is done by means
4205 of setjmp/longjmp. The whole thing is implemented by a set of macros that
4206 hide most of the details from the main code, and operates only if
4207 NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
4208 "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
4209 operating.
4210
4211 To make it easier for callers to provide specially tailored get/free
4212 functions for this usage, two new functions, pcre_stack_malloc, and
4213 pcre_stack_free, are used. They are always called in strict stacking order,
4214 and the size of block requested is always the same.
4215
4216 The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
4217 PCRE has been compiled to use the stack or the heap for recursion. The
4218 -C option of pcretest uses this to show which version is compiled.
4219
4220 A new data escape \S, is added to pcretest; it causes the amounts of store
4221 obtained and freed by both kinds of malloc/free at match time to be added
4222 to the output.
4223
4224 2. Changed the locale test to use "fr_FR" instead of "fr" because that's
4225 what's available on my current Linux desktop machine.
4226
4227 3. When matching a UTF-8 string, the test for a valid string at the start has
4228 been extended. If start_offset is not zero, PCRE now checks that it points
4229 to a byte that is the start of a UTF-8 character. If not, it returns
4230 PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
4231 this is necessary because there may be backward assertions in the pattern.
4232 When matching the same subject several times, it may save resources to use
4233 PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
4234
4235 4. The code for checking the validity of UTF-8 strings has been tightened so
4236 that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
4237 containing "overlong sequences".
4238
4239 5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
4240 I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
4241 should have been "&", but it just so happened that all the cases this let
4242 through by mistake were picked up later in the function.
4243
4244 6. I had used a variable called "isblank" - this is a C99 function, causing
4245 some compilers to warn. To avoid this, I renamed it (as "blankclass").
4246
4247 7. Cosmetic: (a) only output another newline at the end of pcretest if it is
4248 prompting; (b) run "./pcretest /dev/null" at the start of the test script
4249 so the version is shown; (c) stop "make test" echoing "./RunTest".
4250
4251 8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
4252
4253 9. The prototype for memmove() for systems that don't have it was using
4254 size_t, but the inclusion of the header that defines size_t was later. I've
4255 moved the #includes for the C headers earlier to avoid this.
4256
4257 10. Added some adjustments to the code to make it easier to compiler on certain
4258 special systems:
4259
4260 (a) Some "const" qualifiers were missing.
4261 (b) Added the macro EXPORT before all exported functions; by default this
4262 is defined to be empty.
4263 (c) Changed the dftables auxiliary program (that builds chartables.c) so
4264 that it reads its output file name as an argument instead of writing
4265 to the standard output and assuming this can be redirected.
4266
4267 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
4268 class containing characters with values greater than 255, PCRE compilation
4269 went into a loop.
4270
4271 12. A recursive reference to a subpattern that was within another subpattern
4272 that had a minimum quantifier of zero caused PCRE to crash. For example,
4273 (x(y(?2))z)? provoked this bug with a subject that got as far as the
4274 recursion. If the recursively-called subpattern itself had a zero repeat,
4275 that was OK.
4276
4277 13. In pcretest, the buffer for reading a data line was set at 30K, but the
4278 buffer into which it was copied (for escape processing) was still set at
4279 1024, so long lines caused crashes.
4280
4281 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
4282 "internal error: code overflow...". This applied to any character class
4283 that was followed by a possessive quantifier.
4284
4285 15. Modified the Makefile to add libpcre.la as a prerequisite for
4286 libpcreposix.la because I was told this is needed for a parallel build to
4287 work.
4288
4289 16. If a pattern that contained .* following optional items at the start was
4290 studied, the wrong optimizing data was generated, leading to matching
4291 errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
4292 matching string must start with a or b or c. The correct conclusion for
4293 this pattern is that a match can start with any character.
4294
4295
4296 Version 4.4 13-Aug-03
4297 ---------------------
4298
4299 1. In UTF-8 mode, a character class containing characters with values between
4300 127 and 255 was not handled correctly if the compiled pattern was studied.
4301 In fixing this, I have also improved the studying algorithm for such
4302 classes (slightly).
4303
4304 2. Three internal functions had redundant arguments passed to them. Removal
4305 might give a very teeny performance improvement.
4306
4307 3. Documentation bug: the value of the capture_top field in a callout is *one
4308 more than* the number of the hightest numbered captured substring.
4309
4310 4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
4311 in incorrectly linking with a previously installed version. They now link
4312 explicitly with libpcre.la.
4313
4314 5. configure.in no longer needs to recognize Cygwin specially.
4315
4316 6. A problem in pcre.in for Windows platforms is fixed.
4317
4318 7. If a pattern was successfully studied, and the -d (or /D) flag was given to
4319 pcretest, it used to include the size of the study block as part of its
4320 output. Unfortunately, the structure contains a field that has a different
4321 size on different hardware architectures. This meant that the tests that
4322 showed this size failed. As the block is currently always of a fixed size,
4323 this information isn't actually particularly useful in pcretest output, so
4324 I have just removed it.
4325
4326 8. Three pre-processor statements accidentally did not start in column 1.
4327 Sadly, there are *still* compilers around that complain, even though
4328 standard C has not required this for well over a decade. Sigh.
4329
4330 9. In pcretest, the code for checking callouts passed small integers in the
4331 callout_data field, which is a void * field. However, some picky compilers
4332 complained about the casts involved for this on 64-bit systems. Now
4333 pcretest passes the address of the small integer instead, which should get
4334 rid of the warnings.
4335
4336 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
4337 both compile and run time, and gives an error if an invalid UTF-8 sequence
4338 is found. There is a option for disabling this check in cases where the
4339 string is known to be correct and/or the maximum performance is wanted.
4340
4341 11. In response to a bug report, I changed one line in Makefile.in from
4342
4343 -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
4344 to
4345 -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
4346
4347 to look similar to other lines, but I have no way of telling whether this
4348 is the right thing to do, as I do not use Windows. No doubt I'll get told
4349 if it's wrong...
4350
4351
4352 Version 4.3 21-May-03
4353 ---------------------
4354
4355 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
4356 Makefile.
4357
4358 2. Some refactoring to improve the quality of the code:
4359
4360 (i) The utf8_table... variables are now declared "const".
4361
4362 (ii) The code for \cx, which used the "case flipping" table to upper case
4363 lower case letters, now just substracts 32. This is ASCII-specific,
4364 but the whole concept of \cx is ASCII-specific, so it seems
4365 reasonable.
4366
4367 (iii) PCRE was using its character types table to recognize decimal and
4368 hexadecimal digits in the pattern. This is silly, because it handles
4369 only 0-9, a-f, and A-F, but the character types table is locale-
4370 specific, which means strange things might happen. A private
4371 table is now used for this - though it costs 256 bytes, a table is
4372 much faster than multiple explicit tests. Of course, the standard
4373 character types table is still used for matching digits in subject
4374 strings against \d.
4375
4376 (iv) Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
4377 ending in _t are). So I've renamed it as ESC_tee.
4378
4379 3. The first argument for regexec() in the POSIX wrapper should have been
4380 defined as "const".
4381
4382 4. Changed pcretest to use malloc() for its buffers so that they can be
4383 Electric Fenced for debugging.
4384
4385 5. There were several places in the code where, in UTF-8 mode, PCRE would try
4386 to read one or more bytes before the start of the subject string. Often this
4387 had no effect on PCRE's behaviour, but in some circumstances it could
4388 provoke a segmentation fault.
4389
4390 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
4391 to try to read one or more bytes before the start of the subject string.
4392
4393 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
4394 UTF-8 support could misbehave in various ways if the subject string
4395 contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
4396 area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
4397 back over UTF-8 characters.)
4398
4399
4400 Version 4.2 14-Apr-03
4401 ---------------------
4402
4403 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
4404
4405 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
4406 [ON_WINDOWS]: new variable, "#" on non-Windows platforms
4407 [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
4408 [WIN_PREFIX]: new variable, "cyg" for Cygwin
4409 * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
4410 and BUILD_EXEEXT
4411 Note: automatic setting of the BUILD variables is not yet working
4412 set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
4413 compile-time but not at link-time
4414 [LINK]: use for linking executables only
4415 make different versions for Windows and non-Windows
4416 [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
4417 libraries
4418 [LINK_FOR_BUILD]: new variable
4419 [OBJEXT]: use throughout
4420 [EXEEXT]: use throughout
4421 <winshared>: new target
4422 <wininstall>: new target
4423 <dftables.o>: use native compiler
4424 <dftables>: use native linker
4425 <install>: handle Windows platform correctly
4426 <clean>: ditto
4427 <check>: ditto
4428 copy DLL to top builddir before testing
4429
4430 As part of these changes, -no-undefined was removed again. This was reported
4431 to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
4432 in any case.
4433
4434 3. Some tidies to get rid of compiler warnings:
4435
4436 . In the match_data structure, match_limit was an unsigned long int, whereas
4437 match_call_count was an int. I've made them both unsigned long ints.
4438
4439 . In pcretest the fact that a const uschar * doesn't automatically cast to
4440 a void * provoked a warning.
4441
4442 . Turning on some more compiler warnings threw up some "shadow" variables
4443 and a few more missing casts.
4444
4445 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
4446 option, a class that contained a single character with a value between 128
4447 and 255 (e.g. /[\xFF]/) caused PCRE to crash.
4448
4449 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
4450 option, a class that contained several characters, but with at least one
4451 whose value was between 128 and 255 caused PCRE to crash.
4452
4453
4454 Version 4.1 12-Mar-03
4455 ---------------------
4456
4457 1. Compiling with gcc -pedantic found a couple of places where casts were
4458 needed, and a string in dftables.c that was longer than standard compilers are
4459 required to support.
4460
4461 2. Compiling with Sun's compiler found a few more places where the code could
4462 be tidied up in order to avoid warnings.
4463
4464 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
4465 first of these names is deprecated in the latest Autoconf in favour of the name
4466 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
4467 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
4468 analogy I have changed it to CFLAGS_FOR_BUILD.
4469
4470 4. Added -no-undefined to the linking command in the Makefile, because this is
4471 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
4472 linking step for the pcreposix library.
4473
4474 5. PCRE was failing to diagnose the case of two named groups with the same
4475 name.
4476
4477 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
4478 literal character that is needed in the subject for a match, and scans along to
4479 ensure that it is present before embarking on the full matching process. This
4480 saves time in cases of nested unlimited repeats that are never going to match.
4481 Problem: the scan can take a lot of time if the subject is very long (e.g.
4482 megabytes), thus penalizing straightforward matches. It is now done only if the
4483 amount of subject to be scanned is less than 1000 bytes.
4484
4485 7. A lesser problem with the same optimization is that it was recording the
4486 first character of an anchored pattern as "needed", thus provoking a search
4487 right along the subject, even when the first match of the pattern was going to
4488 fail. The "needed" character is now not set for anchored patterns, unless it
4489 follows something in the pattern that is of non-fixed length. Thus, it still
4490 fulfils its original purpose of finding quick non-matches in cases of nested
4491 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
4492
4493
4494 Version 4.0 17-Feb-03
4495 ---------------------
4496
4497 1. If a comment in an extended regex that started immediately after a meta-item
4498 extended to the end of string, PCRE compiled incorrect data. This could lead to
4499 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
4500
4501 2. Moved to autoconf 2.53 and libtool 1.4.2.
4502
4503 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
4504 the special perltest8 script is no longer needed - all the tests can be run
4505 from a single perltest script.
4506
4507 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
4508 by \s. It has now been removed in PCRE. This means it isn't recognized as
4509 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
4510 class [:space:] *does* include VT, thereby creating a mess.
4511
4512 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
4513 space and tab.
4514
4515 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
4516 its new features into the main test script, reducing the number of scripts.
4517
4518 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
4519 were backward compatible, and made the (?i) apply to the whole pattern, as if
4520 /i were given. Now it behaves more logically, and applies the option setting
4521 only to what follows. PCRE has been changed to follow suit. However, if it
4522 finds options settings right at the start of the pattern, it extracts them into
4523 the global options, as before. Thus, they show up in the info data.
4524
4525 8. Added support for the \Q...\E escape sequence. Characters in between are
4526 treated as literals. This is slightly different from Perl in that $ and @ are
4527 also handled as literals inside the quotes. In Perl, they will cause variable
4528 interpolation. Note the following examples:
4529
4530 Pattern PCRE matches Perl matches
4531
4532 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
4533 \Qabc\$xyz\E abc\$xyz abc\$xyz
4534 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
4535
4536 For compatibility with Perl, \Q...\E sequences are recognized inside character
4537 classes as well as outside them.
4538
4539 9. Re-organized 3 code statements in pcretest to avoid "overflow in
4540 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
4541 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
4542 signed/unsigned warnings.
4543
4544 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
4545 option for pcretest, so I've replaced it by a simple function that does just
4546 that job.
4547
4548 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
4549 "pcregrep -".
4550
4551 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
4552 Java package. This provides some syntactic sugar for simple cases of what my
4553 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
4554 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
4555 item, you can use this simplified notation. Note that only makes sense with
4556 greedy quantifiers. Consequently, the use of the possessive quantifier forces
4557 greediness, whatever the setting of the PCRE_UNGREEDY option.
4558
4559 13. A change of greediness default within a pattern was not taking effect at
4560 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
4561 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
4562 was abstracted outside.
4563
4564 14. PCRE now supports the \G assertion. It is true when the current matching
4565 position is at the start point of the match. This differs from \A when the
4566 starting offset is non-zero. Used with the /g option of pcretest (or similar
4567 code), it works in the same way as it does for Perl's /g option. If all
4568 alternatives of a regex begin with \G, the expression is anchored to the start
4569 match position, and the "anchored" flag is set in the compiled expression.
4570
4571 15. Some bugs concerning the handling of certain option changes within patterns
4572 have been fixed. These applied to options other than (?ims). For example,
4573 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
4574 been the other way round. Some of this was related to change 7 above.
4575
4576 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
4577 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
4578 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
4579 POSIX classes only within a class (e.g. /[[:alpha:]]/).
4580
4581 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
4582 mode. Unlike ".", it always matches newline, whatever the setting of
4583 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
4584 assertions. Perl allows it, but it doesn't (in general) work because it can't
4585 calculate the length of the lookbehind. At least, that's the case for Perl
4586 5.8.0 - I've been told they are going to document that it doesn't work in
4587 future.
4588
4589 18. Added an error diagnosis for escapes that PCRE does not support: these are
4590 \L, \l, \N, \P, \p, \U, \u, and \X.
4591
4592 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
4593 reading past the end of the pattern in cases such as /[abcd/.
4594
4595 20. PCRE was getting more memory than necessary for patterns with classes that
4596 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
4597
4598 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
4599 compiling PCRE for use with Virtual Pascal.
4600
4601 22. Small fix to the Makefile to make it work properly if the build is done
4602 outside the source tree.
4603
4604 23. Added a new extension: a condition to go with recursion. If a conditional
4605 subpattern starts with (?(R) the "true" branch is used if recursion has
4606 happened, whereas the "false" branch is used only at the top level.
4607
4608 24. When there was a very long string of literal characters (over 255 bytes
4609 without UTF support, over 250 bytes with UTF support), the computation of how
4610 much memory was required could be incorrect, leading to segfaults or other
4611 strange effects.
4612
4613 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
4614 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
4615 there was a subsequent back reference to those brackets. This meant that, for
4616 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
4617 possible to check for precisely this case. All we can do is abandon the
4618 optimization if .* occurs inside capturing brackets when there are any back
4619 references whatsoever. (See below for a better fix that came later.)
4620
4621 26. The handling of the optimization for finding the first character of a
4622 non-anchored pattern, and for finding a character that is required later in the
4623 match were failing in some cases. This didn't break the matching; it just
4624 failed to optimize when it could. The way this is done has been re-implemented.
4625
4626 27. Fixed typo in error message for invalid (?R item (it said "(?p").
4627
4628 28. Added a new feature that provides some of the functionality that Perl
4629 provides with (?{...}). The facility is termed a "callout". The way it is done
4630 in PCRE is for the caller to provide an optional function, by setting
4631 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
4632 global variable. By default it is unset, which disables all calling out. To get
4633 the function called, the regex must include (?C) at appropriate points. This
4634 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
4635 This provides a means of identifying different callout points. When PCRE
4636 reaches such a point in the regex, if pcre_callout has been set, the external
4637 function is called. It is provided with data in a structure called
4638 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
4639 matching continues; if it returns a non-zero value, the match at the current
4640 point fails. However, backtracking will occur if possible. [This was changed
4641 later and other features added - see item 49 below.]
4642
4643 29. pcretest is upgraded to test the callout functionality. It provides a
4644 callout function that displays information. By default, it shows the start of
4645 the match and the current position in the text. There are some new data escapes
4646 to vary what happens:
4647
4648 \C+ in addition, show current contents of captured substrings
4649 \C- do not supply a callout function
4650 \C!n return 1 when callout number n is reached
4651 \C!n!m return 1 when callout number n is reached for the mth time
4652
4653 30. If pcregrep was called with the -l option and just a single file name, it
4654 output "<stdin>" if a match was found, instead of the file name.
4655
4656 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
4657 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
4658 pcre_exec(). This saves a malloc/free per call. The default value of
4659 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
4660 when configuring.
4661
4662 32. The default maximum size of a compiled pattern is 64K. There have been a
4663 few cases of people hitting this limit. The code now uses macros to handle the
4664 storing of links as offsets within the compiled pattern. It defaults to 2-byte
4665 links, but this can be changed to 3 or 4 bytes by --with-link-size when
4666 configuring. Tests 2 and 5 work only with 2-byte links because they output
4667 debugging information about compiled patterns.
4668
4669 33. Internal code re-arrangements:
4670
4671 (a) Moved the debugging function for printing out a compiled regex into
4672 its own source file (printint.c) and used #include to pull it into
4673 pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
4674 separate copies.
4675
4676 (b) Defined the list of op-code names for debugging as a macro in
4677 internal.h so that it is next to the definition of the opcodes.
4678
4679 (c) Defined a table of op-code lengths for simpler skipping along compiled
4680 code. This is again a macro in internal.h so that it is next to the
4681 definition of the opcodes.
4682
4683 34. Added support for recursive calls to individual subpatterns, along the
4684 lines of Robin Houston's patch (but implemented somewhat differently).
4685
4686 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
4687 allow it to read and process whole directories in Win32. This code was
4688 contributed by Lionel Fourquaux; it has not been tested by me.
4689
4690 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
4691 used to name a group. Names consist of alphanumerics and underscores, and must
4692 be unique. Back references use the syntax (?P=name) and recursive calls use
4693 (?P>name) which is a PCRE extension to the Python extension. Groups still have
4694 numbers. The function pcre_fullinfo() can be used after compilation to extract
4695 a name/number map. There are three relevant calls:
4696
4697 PCRE_INFO_NAMEENTRYSIZE yields the size of each entry in the map
4698 PCRE_INFO_NAMECOUNT yields the number of entries
4699 PCRE_INFO_NAMETABLE yields a pointer to the map.
4700
4701 The map is a vector of fixed-size entries. The size of each entry depends on
4702 the length of the longest name used. The first two bytes of each entry are the
4703 group number, most significant byte first. There follows the corresponding
4704 name, zero terminated. The names are in alphabetical order.
4705
4706 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
4707 case instead of 255. Making it the same both with and without UTF-8 support
4708 means that the same test output works with both.
4709
4710 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
4711 calling malloc() with a zero argument.
4712
4713 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
4714 optimization. I've improved things by keeping a bitmap of backreferences with
4715 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
4716 fact referenced, the optimization can be applied. It is unlikely that a
4717 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
4718 the match to follow \n) will appear inside brackets with a number greater than
4719 31, but if it does, any back reference > 31 suppresses the optimization.
4720
4721 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
4722 of disabling numbered capturing parentheses. Any opening parenthesis that is
4723 not followed by ? behaves as if it were followed by ?: but named parentheses
4724 can still be used for capturing (and they will acquire numbers in the usual
4725 way).
4726
4727 41. Redesigned the return codes from the match() function into yes/no/error so
4728 that errors can be passed back from deep inside the nested calls. A malloc
4729 failure while inside a recursive subpattern call now causes the
4730 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
4731
4732 42. It is now possible to set a limit on the number of times the match()
4733 function is called in a call to pcre_exec(). This facility makes it possible to
4734 limit the amount of recursion and backtracking, though not in a directly
4735 obvious way, because the match() function is used in a number of different
4736 circumstances. The count starts from zero for each position in the subject
4737 string (for non-anchored patterns). The default limit is, for compatibility, a
4738 large number, namely 10 000 000. You can change this in two ways:
4739
4740 (a) When configuring PCRE before making, you can use --with-match-limit=n
4741 to set a default value for the compiled library.
4742
4743 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
4744 a different value is set. See 45 below.
4745
4746 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
4747
4748 43. Added a new function pcre_config(int, void *) to enable run-time extraction
4749 of things that can be changed at compile time. The first argument specifies
4750 what is wanted and the second points to where the information is to be placed.
4751 The current list of available information is:
4752
4753 PCRE_CONFIG_UTF8
4754
4755 The output is an integer that is set to one if UTF-8 support is available;
4756 otherwise it is set to zero.
4757
4758 PCRE_CONFIG_NEWLINE
4759
4760 The output is an integer that it set to the value of the code that is used for
4761 newline. It is either LF (10) or CR (13).
4762
4763 PCRE_CONFIG_LINK_SIZE
4764
4765 The output is an integer that contains the number of bytes used for internal
4766 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
4767
4768 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
4769
4770 The output is an integer that contains the threshold above which the POSIX
4771 interface uses malloc() for output vectors. See item 31 above.
4772
4773 PCRE_CONFIG_MATCH_LIMIT
4774
4775 The output is an unsigned integer that contains the default limit of the number
4776 of match() calls in a pcre_exec() execution. See 42 above.
4777
4778 44. pcretest has been upgraded by the addition of the -C option. This causes it
4779 to extract all the available output from the new pcre_config() function, and to
4780 output it. The program then exits immediately.
4781
4782 45. A need has arisen to pass over additional data with calls to pcre_exec() in
4783 order to support additional features. One way would have been to define
4784 pcre_exec2() (for example) with extra arguments, but this would not have been
4785 extensible, and would also have required all calls to the original function to
4786 be mapped to the new one. Instead, I have chosen to extend the mechanism that
4787 is used for passing in "extra" data from pcre_study().
4788
4789 The pcre_extra structure is now exposed and defined in pcre.h. It currently
4790 contains the following fields:
4791
4792 flags a bitmap indicating which of the following fields are set
4793 study_data opaque data from pcre_study()
4794 match_limit a way of specifying a limit on match() calls for a specific
4795 call to pcre_exec()
4796 callout_data data for callouts (see 49 below)
4797
4798 The flag bits are also defined in pcre.h, and are
4799
4800 PCRE_EXTRA_STUDY_DATA
4801 PCRE_EXTRA_MATCH_LIMIT
4802 PCRE_EXTRA_CALLOUT_DATA
4803
4804 The pcre_study() function now returns one of these new pcre_extra blocks, with
4805 the actual study data pointed to by the study_data field, and the
4806 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
4807 before. That is, this change is entirely upwards-compatible and requires no
4808 change to existing code.
4809
4810 If you want to pass in additional data to pcre_exec(), you can either place it
4811 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
4812 block.
4813
4814 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
4815 data string contains the escape sequence \M, pcretest calls pcre_exec() several
4816 times with different match limits, until it finds the minimum value needed for
4817 pcre_exec() to complete. The value is then output. This can be instructive; for
4818 most simple matches the number is quite small, but for pathological cases it
4819 gets very large very quickly.
4820
4821 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
4822 returns the size of the data block pointed to by the study_data field in a
4823 pcre_extra block, that is, the value that was passed as the argument to
4824 pcre_malloc() when PCRE was getting memory in which to place the information
4825 created by pcre_study(). The fourth argument should point to a size_t variable.
4826 pcretest has been extended so that this information is shown after a successful
4827 pcre_study() call when information about the compiled regex is being displayed.
4828
4829 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
4830 because what follows is always an absolute path. (Later: it turns out that this
4831 is more than cosmetic for MinGW, because it doesn't like empty path
4832 components.)
4833
4834 49. Some changes have been made to the callout feature (see 28 above):
4835
4836 (i) A callout function now has three choices for what it returns:
4837
4838 0 => success, carry on matching
4839 > 0 => failure at this point, but backtrack if possible
4840 < 0 => serious error, return this value from pcre_exec()
4841
4842 Negative values should normally be chosen from the set of PCRE_ERROR_xxx
4843 values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
4844 "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
4845 use by callout functions. It will never be used by PCRE itself.
4846
4847 (ii) The pcre_extra structure (see 45 above) has a void * field called
4848 callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
4849 pcre_callout_block structure has a field of the same name. The contents of
4850 the field passed in the pcre_extra structure are passed to the callout
4851 function in the corresponding field in the callout block. This makes it
4852 easier to use the same callout-containing regex from multiple threads. For
4853 testing, the pcretest program has a new data escape
4854
4855 \C*n pass the number n (may be negative) as callout_data
4856
4857 If the callout function in pcretest receives a non-zero value as
4858 callout_data, it returns that value.
4859
4860 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
4861 there were some redundant $(CFLAGS) in commands that are now specified as
4862 $(LINK), which already includes $(CFLAGS).
4863
4864 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
4865 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
4866 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
4867 one-byte characters throughout. Note that case-insensitive matching applies
4868 only to characters whose values are less than 256. PCRE doesn't support the
4869 notion of cases for higher-valued characters.
4870
4871 (i) A character class whose characters are all within 0-255 is handled as
4872 a bit map, and the map is inverted for negative classes. Previously, a
4873 character > 255 always failed to match such a class; however it should
4874 match if the class was a negative one (e.g. [^ab]). This has been fixed.
4875
4876 (ii) A negated character class with a single character < 255 is coded as
4877 "not this character" (OP_NOT). This wasn't working properly when the test
4878 character was multibyte, either singly or repeated.
4879
4880 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
4881 mode, for example: \x{100}{2,3}.
4882
4883 (iv) The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
4884 singly or repeated) now correctly test multibyte characters. However,
4885 PCRE doesn't recognize any characters with values greater than 255 as
4886 digits, spaces, or word characters. Such characters always match \D, \S,
4887 and \W, and never match \d, \s, or \w.
4888
4889 (v) Classes may now contain characters and character ranges with values
4890 greater than 255. For example: [ab\x{100}-\x{400}].
4891
4892 (vi) pcregrep now has a --utf-8 option (synonym -u) which makes it call
4893 PCRE in UTF-8 mode.
4894
4895 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
4896 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
4897 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
4898 value.)
4899
4900 53. The single man page has become too large. I have therefore split it up into
4901 a number of separate man pages. These also give rise to individual HTML pages;
4902 these are now put in a separate directory, and there is an index.html page that
4903 lists them all. Some hyperlinking between the pages has been installed.
4904
4905 54. Added convenience functions for handling named capturing parentheses.
4906
4907 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
4908 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
4909 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
4910 are faulted.
4911
4912 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
4913 calling configure. These values are used when compiling the dftables.c program
4914 which is run to generate the source of the default character tables. They
4915 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
4916 you will need to set these values.
4917
4918 57. Updated the building process for Windows DLL, as provided by Fred Cox.
4919
4920
4921 Version 3.9 02-Jan-02
4922 ---------------------
4923
4924 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
4925
4926 2. If --disable-static was given, the building process failed when trying to
4927 build pcretest and pcregrep. (For some reason it was using libtool to compile
4928 them, which is not right, as they aren't part of the library.)
4929
4930
4931 Version 3.8 18-Dec-01
4932 ---------------------
4933
4934 1. The experimental UTF-8 code was completely screwed up. It was packing the
4935 bytes in the wrong order. How dumb can you get?
4936
4937
4938 Version 3.7 29-Oct-01
4939 ---------------------
4940
4941 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
4942 This caused pcretest, when used on the test data, to segfault. Unfortunately,
4943 this didn't happen under Solaris 8, where I normally test things.
4944
4945 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
4946 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
4947 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
4948 this fix an hour or so after the initial 3.7 release.)
4949
4950
4951 Version 3.6 23-Oct-01
4952 ---------------------
4953
4954 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
4955 offsets passed as NULL with zero offset count.
4956
4957 2. The config.guess and config.sub files had not been updated when I moved to
4958 the latest autoconf.
4959
4960
4961 Version 3.5 15-Aug-01
4962 ---------------------
4963
4964 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
4965 had been forgotten.
4966
4967 2. By using declared but undefined structures, we can avoid using "void"
4968 definitions in pcre.h while keeping the internal definitions of the structures
4969 private.
4970
4971 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
4972 user point of view, this means that both static and shared libraries are built
4973 by default, but this can be individually controlled. More of the work of
4974 handling this static/shared cases is now inside libtool instead of PCRE's make
4975 file.
4976
4977 4. The pcretest utility is now installed along with pcregrep because it is
4978 useful for users (to test regexs) and by doing this, it automatically gets
4979 relinked by libtool. The documentation has been turned into a man page, so
4980 there are now .1, .txt, and .html versions in /doc.
4981
4982 5. Upgrades to pcregrep:
4983 (i) Added long-form option names like gnu grep.
4984 (ii) Added --help to list all options with an explanatory phrase.
4985 (iii) Added -r, --recursive to recurse into sub-directories.
4986 (iv) Added -f, --file to read patterns from a file.
4987
4988 6. pcre_exec() was referring to its "code" argument before testing that
4989 argument for NULL (and giving an error if it was NULL).
4990
4991 7. Upgraded Makefile.in to allow for compiling in a different directory from
4992 the source directory.
4993
4994 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
4995 options bits, the pointer it was passed was to an int instead of to an unsigned
4996 long int. This mattered only on 64-bit systems.
4997
4998 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
4999 generated) instead of pcre.in, which it its source. Also made the same change
5000 in several of the .c files.
5001
5002 10. A new release of gcc defines printf() as a macro, which broke pcretest
5003 because it had an ifdef in the middle of a string argument for printf(). Fixed
5004 by using separate calls to printf().
5005
5006 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
5007 script, to force use of CR or LF instead of \n in the source. On non-Unix
5008 systems, the value can be set in config.h.
5009
5010 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
5011 absolute limit. Changed the text of the error message to make this clear, and
5012 likewise updated the man page.
5013
5014 13. The limit of 99 on the number of capturing subpatterns has been removed.
5015 The new limit is 65535, which I hope will not be a "real" limit.
5016
5017
5018 Version 3.4 22-Aug-00
5019 ---------------------
5020
5021 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
5022
5023 2. Diagnose condition (?(0) as an error instead of crashing on matching.
5024
5025
5026 Version 3.3 01-Aug-00
5027 ---------------------
5028
5029 1. If an octal character was given, but the value was greater than \377, it
5030 was not getting masked to the least significant bits, as documented. This could
5031 lead to crashes in some systems.
5032
5033 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
5034 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
5035
5036 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
5037 These just pass their arguments on to (pcre_free)(), but they are provided
5038 because some uses of PCRE bind it to non-C systems that can call its functions,
5039 but cannot call free() or pcre_free() directly.
5040
5041 4. Add "make test" as a synonym for "make check". Corrected some comments in
5042 the Makefile.
5043
5044 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
5045 Makefile.
5046
5047 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
5048 command called pgrep for grepping around the active processes.
5049
5050 7. Added the beginnings of support for UTF-8 character strings.
5051
5052 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
5053 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
5054 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
5055 out for the ar command.)
5056
5057
5058 Version 3.2 12-May-00
5059 ---------------------
5060
5061 This is purely a bug fixing release.
5062
5063 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
5064 of ZA. This was just one example of several cases that could provoke this bug,
5065 which was introduced by change 9 of version 2.00. The code for breaking
5066 infinite loops after an iteration that matches an empty string was't working
5067 correctly.
5068
5069 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
5070 when matched against abbab (for example). After matching an empty string, it
5071 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
5072 caused it to match further down the string than it should.
5073
5074 3. The code contained an inclusion of sys/types.h. It isn't clear why this
5075 was there because it doesn't seem to be needed, and it causes trouble on some
5076 systems, as it is not a Standard C header. It has been removed.
5077
5078 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
5079 were reported on the Macintosh. The changes were from
5080
5081 while ((c = *(++ptr)) != 0 && c != '\n');
5082 to
5083 while ((c = *(++ptr)) != 0 && c != '\n') ;
5084
5085 Totally extraordinary, but if that's what it takes...
5086
5087 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
5088 available. Added HAVE_BCOPY and an autoconf test for it; if neither
5089 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
5090 assumes the way PCRE uses memmove() (always moving upwards).
5091
5092 6. PCRE is being used in one environment where strchr() is not available. There
5093 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
5094 faster code anyway.
5095
5096
5097 Version 3.1 09-Feb-00
5098 ---------------------
5099
5100 The only change in this release is the fixing of some bugs in Makefile.in for
5101 the "install" target:
5102
5103 (1) It was failing to install pcreposix.h.
5104
5105 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
5106
5107
5108 Version 3.0 01-Feb-00
5109 ---------------------
5110
5111 1. Add support for the /+ modifier to perltest (to output $` like it does in
5112 pcretest).
5113
5114 2. Add support for the /g modifier to perltest.
5115
5116 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
5117 matches null strings.
5118
5119 4. Fix perltest so that it doesn't do unwanted things when fed an empty
5120 pattern. Perl treats empty patterns specially - it reuses the most recent
5121 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
5122 effect.
5123
5124 5. The POSIX interface was broken in that it was just handing over the POSIX
5125 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
5126 required a bigger vector, with some working space on the end. This means that
5127 the POSIX wrapper now has to get and free some memory, and copy the results.
5128
5129 6. Added some simple autoconf support, placing the test data and the
5130 documentation in separate directories, re-organizing some of the
5131 information files, and making it build pcre-config (a GNU standard). Also added
5132 libtool support for building PCRE as a shared library, which is now the
5133 default.
5134
5135 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
5136 09 are not valid octal constants. Single digits will be used for minor values
5137 less than 10.
5138
5139 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
5140 existing programs that set these in the POSIX interface can use PCRE without
5141 modification.
5142
5143 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
5144 return all that pcre_info() returns, plus additional data. The pcre_info()
5145 function is retained for compatibility, but is considered to be obsolete.
5146
5147 10. Added experimental recursion feature (?R) to handle one common case that
5148 Perl 5.6 will be able to do with (?p{...}).
5149
5150 11. Added support for POSIX character classes like [:alpha:], which Perl is
5151 adopting.
5152
5153
5154 Version 2.08 31-Aug-99
5155 ----------------------
5156
5157 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
5158 trying to match at the startoffset position, but instead was moving forward to
5159 the next newline as if a previous match had failed.
5160
5161 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
5162 and could get into a loop if a null string was matched other than at the start
5163 of the subject.
5164
5165 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
5166 be distinguished at compile time, and for completeness also added PCRE_DATE.
5167
5168 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
5169 in GnuWin32 environments.
5170
5171
5172 Version 2.07 29-Jul-99
5173 ----------------------
5174
5175 1. The documentation is now supplied in plain text form and HTML as well as in
5176 the form of man page sources.
5177
5178 2. C++ compilers don't like assigning (void *) values to other pointer types.
5179 In particular this affects malloc(). Although there is no problem in Standard
5180 C, I've put in casts to keep C++ compilers happy.
5181
5182 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
5183 should be (const char *).
5184
5185 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
5186 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
5187 However, I haven't made this a standard facility. The documentation doesn't
5188 mention it, and the Makefile doesn't support it.
5189
5190 5. The Makefile now contains an "install" target, with editable destinations at
5191 the top of the file. The pcretest program is not installed.
5192
5193 6. pgrep -V now gives the PCRE version number and date.
5194
5195 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
5196 causing the entire string to be ignored, instead of just the last character.
5197
5198 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
5199 non-matching string, it can take a very, very long time, even for strings of
5200 quite modest length, because of the nested recursion. PCRE now does better in
5201 some of these cases. It does this by remembering the last required literal
5202 character in the pattern, and pre-searching the subject to ensure it is present
5203 before running the real match. In other words, it applies a heuristic to detect
5204 some types of certain failure quickly, and in the above example, if presented
5205 with a string that has no trailing " it gives "no match" very quickly.
5206
5207 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
5208 other alternatives are tried instead.
5209
5210
5211 Version 2.06 09-Jun-99
5212 ----------------------
5213
5214 1. Change pcretest's output for amount of store used to show just the code
5215 space, because the remainder (the data block) varies in size between 32-bit and
5216 64-bit systems.
5217
5218 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
5219 start matching at. This allows lookbehinds to work when searching for multiple
5220 occurrences in a string.
5221
5222 3. Added additional options to pcretest for testing multiple occurrences:
5223
5224 /+ outputs the rest of the string that follows a match
5225 /g loops for multiple occurrences, using the new startoffset argument
5226 /G loops for multiple occurrences by passing an incremented pointer
5227
5228 4. PCRE wasn't doing the "first character" optimization for patterns starting
5229 with \b or \B, though it was doing it for other lookbehind assertions. That is,
5230 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
5231 the letter 'x'. On long subject strings, this gives a significant speed-up.
5232
5233
5234 Version 2.05 21-Apr-99
5235 ----------------------
5236
5237 1. Changed the type of magic_number from int to long int so that it works
5238 properly on 16-bit systems.
5239
5240 2. Fixed a bug which caused patterns starting with .* not to work correctly
5241 when the subject string contained newline characters. PCRE was assuming
5242 anchoring for such patterns in all cases, which is not correct because .* will
5243 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
5244 DOTALL is set at top level; otherwise it knows that patterns starting with .*
5245 must be retried after every newline in the subject.
5246
5247
5248 Version 2.04 18-Feb-99
5249 ----------------------
5250
5251 1. For parenthesized subpatterns with repeats whose minimum was zero, the
5252 computation of the store needed to hold the pattern was incorrect (too large).
5253 If such patterns were nested a few deep, this could multiply and become a real
5254 problem.
5255
5256 2. Added /M option to pcretest to show the memory requirement of a specific
5257 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
5258
5259 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
5260 compiled in such a way that the backtracking after subsequent failure was
5261 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
5262 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
5263
5264
5265 Version 2.03 02-Feb-99
5266 ----------------------
5267
5268 1. Fixed typo and small mistake in man page.
5269
5270 2. Added 4th condition (GPL supersedes if conflict) and created separate
5271 LICENCE file containing the conditions.
5272
5273 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
5274 Perl, that is the internal \ allows the delimiter to be included in the
5275 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
5276 the final delimiter, add \ to the end of the pattern (to test the error).
5277
5278 4. Added the convenience functions for extracting substrings after a successful
5279 match. Updated pcretest to make it able to test these functions.
5280
5281
5282 Version 2.02 14-Jan-99
5283 ----------------------
5284
5285 1. Initialized the working variables associated with each extraction so that
5286 their saving and restoring doesn't refer to uninitialized store.
5287
5288 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
5289 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
5290 fix the problem.
5291
5292 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
5293 calls, and wasn't printing the correct value for compiling calls. Increased the
5294 default value of LOOPREPEAT, and the number of significant figures in the
5295 times.
5296
5297 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
5298
5299 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
5300 a building problem on Windows NT with a FAT file system.
5301
5302
5303 Version 2.01 21-Oct-98
5304 ----------------------
5305
5306 1. Changed the API for pcre_compile() to allow for the provision of a pointer
5307 to character tables built by pcre_maketables() in the current locale. If NULL
5308 is passed, the default tables are used.
5309
5310
5311 Version 2.00 24-Sep-98
5312 ----------------------
5313
5314 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
5315 it any more.
5316
5317 2. Allow quantification of (?>) groups, and make it work correctly.
5318
5319 3. The first character computation wasn't working for (?>) groups.
5320
5321 4. Correct the implementation of \Z (it is permitted to match on the \n at the
5322 end of the subject) and add 5.005's \z, which really does match only at the
5323 very end of the subject.
5324
5325 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
5326
5327 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
5328 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
5329 localized options. All options to pcre_study() were also removed.
5330
5331 7. Add other new features from 5.005:
5332
5333 $(?<= positive lookbehind
5334 $(?<! negative lookbehind
5335 (?imsx-imsx) added the unsetting capability
5336 such a setting is global if at outer level; local otherwise
5337 (?imsx-imsx:) non-capturing groups with option setting
5338 (?(cond)re|re) conditional pattern matching
5339
5340 A backreference to itself in a repeated group matches the previous
5341 captured string.
5342
5343 8. General tidying up of studying (both automatic and via "study")
5344 consequential on the addition of new assertions.
5345
5346 9. As in 5.005, unlimited repeated groups that could match an empty substring
5347 are no longer faulted at compile time. Instead, the loop is forcibly broken at
5348 runtime if any iteration does actually match an empty substring.
5349
5350 10. Include the RunTest script in the distribution.
5351
5352 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
5353 discrepancies, some of which were old and were also with respect to 5.004. They
5354 have now been fixed.
5355
5356
5357 Version 1.09 28-Apr-98
5358 ----------------------
5359
5360 1. A negated single character class followed by a quantifier with a minimum
5361 value of one (e.g. [^x]{1,6} ) was not compiled correctly. This could lead to
5362 program crashes, or just wrong answers. This did not apply to negated classes
5363 containing more than one character, or to minima other than one.
5364
5365
5366 Version 1.08 27-Mar-98
5367 ----------------------
5368
5369 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
5370
5371 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
5372 latter must appear before anything that relies on it in the pattern.
5373
5374
5375 Version 1.07 16-Feb-98
5376 ----------------------
5377
5378 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
5379 repeat of a potentially empty string).
5380
5381
5382 Version 1.06 23-Jan-98
5383 ----------------------
5384
5385 1. Added Markus Oberhumer's little patches for C++.
5386
5387 2. Literal strings longer than 255 characters were broken.
5388
5389
5390 Version 1.05 23-Dec-97
5391 ----------------------
5392
5393 1. Negated character classes containing more than one character were failing if
5394 PCRE_CASELESS was set at run time.
5395
5396
5397 Version 1.04 19-Dec-97
5398 ----------------------
5399
5400 1. Corrected the man page, where some "const" qualifiers had been omitted.
5401
5402 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
5403 input syntax.
5404
5405 3. Fixed memory leak which occurred when a regex with back references was
5406 matched with an offsets vector that wasn't big enough. The temporary memory
5407 that is used in this case wasn't being freed if the match failed.
5408
5409 4. Tidied pcretest to ensure it frees memory that it gets.
5410
5411 5. Temporary memory was being obtained in the case where the passed offsets
5412 vector was exactly big enough.
5413
5414 6. Corrected definition of offsetof() from change 5 below.
5415
5416 7. I had screwed up change 6 below and broken the rules for the use of
5417 setjmp(). Now fixed.
5418
5419
5420 Version 1.03 18-Dec-97
5421 ----------------------
5422
5423 1. A erroneous regex with a missing opening parenthesis was correctly
5424 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
5425 on some systems.
5426
5427 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
5428 it was reported that one broken compiler failed on the former because "code" is
5429 also an independent variable.
5430
5431 3. The erroneous regex a[]b caused an array overrun reference.
5432
5433 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
5434 fail on data ending with that character. (It was going on too far, and checking
5435 the next character, typically a binary zero.) This was specific to the
5436 optimized code for single-character negative classes.
5437
5438 5. Added a contributed patch from the TIN world which does the following:
5439
5440 + Add an undef for memmove, in case the the system defines a macro for it.
5441
5442 + Add a definition of offsetof(), in case there isn't one. (I don't know
5443 the reason behind this - offsetof() is part of the ANSI standard - but
5444 it does no harm).
5445
5446 + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
5447 most of the places where whitespace preceded '#'. I have given up and
5448 allowed the remaining 2 cases to be at the margin.
5449
5450 + Rename some variables in pcre to eliminate shadowing. This seems very
5451 pedantic, but does no harm, of course.
5452
5453 6. Moved the call to setjmp() into its own function, to get rid of warnings
5454 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
5455
5456 7. Constructs such as \d{8,} were compiling into the equivalent of
5457 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
5458 outcome, but in this particular case used more store than had been allocated,
5459 which caused the bug to be discovered because it threw up an internal error.
5460
5461 8. The debugging code in both pcre and pcretest for outputting the compiled
5462 form of a regex was going wrong in the case of back references followed by
5463 curly-bracketed repeats.
5464
5465
5466 Version 1.02 12-Dec-97
5467 ----------------------
5468
5469 1. Typos in pcre.3 and comments in the source fixed.
5470
5471 2. Applied a contributed patch to get rid of places where it used to remove
5472 'const' from variables, and fixed some signed/unsigned and uninitialized
5473 variable warnings.
5474
5475 3. Added the "runtest" target to Makefile.
5476
5477 4. Set default compiler flag to -O2 rather than just -O.
5478
5479
5480 Version 1.01 19-Nov-97
5481 ----------------------
5482
5483 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
5484 like /([ab]*)*/, that is, for classes with more than one character in them.
5485
5486 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
5487 as /((?>a*))*/ (a PCRE_EXTRA facility).
5488
5489
5490 Version 1.00 18-Nov-97
5491 ----------------------
5492
5493 1. Added compile-time macros to support systems such as SunOS4 which don't have
5494 memmove() or strerror() but have other things that can be used instead.
5495
5496 2. Arranged that "make clean" removes the executables.
5497
5498
5499 Version 0.99 27-Oct-97
5500 ----------------------
5501
5502 1. Fixed bug in code for optimizing classes with only one character. It was
5503 initializing a 32-byte map regardless, which could cause it to run off the end
5504 of the memory it had got.
5505
5506 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
5507
5508
5509 Version 0.98 22-Oct-97
5510 ----------------------
5511
5512 1. Fixed bug in code for handling temporary memory usage when there are more
5513 back references than supplied space in the ovector. This could cause segfaults.
5514
5515
5516 Version 0.97 21-Oct-97
5517 ----------------------
5518
5519 1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
5520
5521 2. Optimized negated single characters not to use a bit map.
5522
5523 3. Brought error texts together as macro definitions; clarified some of them;
5524 fixed one that was wrong - it said "range out of order" when it meant "invalid
5525 escape sequence".
5526
5527 4. Changed some char * arguments to const char *.
5528
5529 5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
5530
5531 6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
5532 pcretest.
5533
5534
5535 Version 0.96 16-Oct-97
5536 ----------------------
5537
5538 1. Added a simple "pgrep" utility to the distribution.
5539
5540 2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
5541 unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
5542 where "ddd" means "one or more decimal digits".
5543
5544 3. Fixed serious bug. If a pattern had a back reference, but the call to
5545 pcre_exec() didn't supply a large enough ovector to record the related
5546 identifying subpattern, the match always failed. PCRE now remembers the number
5547 of the largest back reference, and gets some temporary memory in which to save
5548 the offsets during matching if necessary, in order to ensure that
5549 backreferences always work.
5550
5551 4. Increased the compatibility with Perl in a number of ways:
5552
5553 (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
5554 to request this handling. The option can be set at compile or exec time.
5555
5556 (b) $ matches before a terminating newline by default; an option
5557 PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
5558 mode). The option can be set at compile or exec time.
5559
5560 (c) The handling of \ followed by a digit other than 0 is now supposed to be
5561 the same as Perl's. If the decimal number it represents is less than 10
5562 or there aren't that many previous left capturing parentheses, an octal
5563 escape is read. Inside a character class, it's always an octal escape,
5564 even if it is a single digit.
5565
5566 (d) An escaped but undefined alphabetic character is taken as a literal,
5567 unless PCRE_EXTRA is set. Currently this just reserves the remaining
5568 escapes.
5569
5570 (e) {0} is now permitted. (The previous item is removed from the compiled
5571 pattern).
5572
5573 5. Changed all the names of code files so that the basic parts are no longer
5574 than 10 characters, and abolished the teeny "globals.c" file.
5575
5576 6. Changed the handling of character classes; they are now done with a 32-byte
5577 bit map always.
5578
5579 7. Added the -d and /D options to pcretest to make it possible to look at the
5580 internals of compilation without having to recompile pcre.
5581
5582
5583 Version 0.95 23-Sep-97
5584 ----------------------
5585
5586 1. Fixed bug in pre-pass concerning escaped "normal" characters such as \x5c or
5587 \x20 at the start of a run of normal characters. These were being treated as
5588 real characters, instead of the source characters being re-checked.
5589
5590
5591 Version 0.94 18-Sep-97
5592 ----------------------
5593
5594 1. The functions are now thread-safe, with the caveat that the global variables
5595 containing pointers to malloc() and free() or alternative functions are the
5596 same for all threads.
5597
5598 2. Get pcre_study() to generate a bitmap of initial characters for non-
5599 anchored patterns when this is possible, and use it if passed to pcre_exec().
5600
5601
5602 Version 0.93 15-Sep-97
5603 ----------------------
5604
5605 1. /(b)|(:+)/ was computing an incorrect first character.
5606
5607 2. Add pcre_study() to the API and the passing of pcre_extra to pcre_exec(),
5608 but not actually doing anything yet.
5609
5610 3. Treat "-" characters in classes that cannot be part of ranges as literals,
5611 as Perl does (e.g. [-az] or [az-]).
5612
5613 4. Set the anchored flag if a branch starts with .* or .*? because that tests
5614 all possible positions.
5615
5616 5. Split up into different modules to avoid including unneeded functions in a
5617 compiled binary. However, compile and exec are still in one module. The "study"
5618 function is split off.
5619
5620 6. The character tables are now in a separate module whose source is generated
5621 by an auxiliary program - but can then be edited by hand if required. There are
5622 now no calls to isalnum(), isspace(), isdigit(), isxdigit(), tolower() or
5623 toupper() in the code.
5624
5625 7. Turn the malloc/free funtions variables into pcre_malloc and pcre_free and
5626 make them global. Abolish the function for setting them, as the caller can now
5627 set them directly.
5628
5629
5630 Version 0.92 11-Sep-97
5631 ----------------------
5632
5633 1. A repeat with a fixed maximum and a minimum of 1 for an ordinary character
5634 (e.g. /a{1,3}/) was broken (I mis-optimized it).
5635
5636 2. Caseless matching was not working in character classes if the characters in
5637 the pattern were in upper case.
5638
5639 3. Make ranges like [W-c] work in the same way as Perl for caseless matching.
5640
5641 4. Make PCRE_ANCHORED public and accept as a compile option.
5642
5643 5. Add an options word to pcre_exec() and accept PCRE_ANCHORED and
5644 PCRE_CASELESS at run time. Add escapes \A and \I to pcretest to cause it to
5645 pass them.
5646
5647 6. Give an error if bad option bits passed at compile or run time.
5648
5649 7. Add PCRE_MULTILINE at compile and exec time, and (?m) as well. Add \M to
5650 pcretest to cause it to pass that flag.
5651
5652 8. Add pcre_info(), to get the number of identifying subpatterns, the stored
5653 options, and the first character, if set.
5654
5655 9. Recognize C+ or C{n,m} where n >= 1 as providing a fixed starting character.
5656
5657
5658 Version 0.91 10-Sep-97
5659 ----------------------
5660
5661 1. PCRE was failing to diagnose unlimited repeats of subpatterns that could
5662 match the empty string as in /(a*)*/. It was looping and ultimately crashing.
5663
5664 2. PCRE was looping on encountering an indefinitely repeated back reference to
5665 a subpattern that had matched an empty string, e.g. /(a|)\1*/. It now does what
5666 Perl does - treats the match as successful.
5667
5668 ****

Properties

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

  ViewVC Help
Powered by ViewVC 1.1.5