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

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1537 - (show annotations)
Fri Mar 27 17:48:28 2015 UTC (4 years, 6 months ago) by ph10
File size: 275184 byte(s)
Fix mutual recursion inside other groups stack overflow bug.
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 and given the CVE number
85 CVE-2015-2325.
86
87 23. A pattern such as "((?+1)(\1))/" containing a forward reference subroutine
88 call within a group that also contained a recursive back reference caused
89 incorrect code to be compiled. This bug was reported as "heap overflow",
90 discovered by Kai Lu of Fortinet's FortiGuard Labs, and given the CVE
91 number CVE-2015-2326.
92
93 24. Computing the size of the JIT read-only data in advance has been a source
94 of various issues, and new ones are still appear unfortunately. To fix
95 existing and future issues, size computation is eliminated from the code,
96 and replaced by on-demand memory allocation.
97
98 25. A pattern such as /(?i)[A-`]/, where characters in the other case are
99 adjacent to the end of the range, and the range contained characters with
100 more than one other case, caused incorrect behaviour when compiled in UTF
101 mode. In that example, the range a-j was left out of the class.
102
103 26. Fix JIT compilation of conditional blocks, which assertion
104 is converted to (*FAIL). E.g: /(?(?!))/.
105
106 27. The pattern /(?(?!)^)/ caused references to random memory. This bug was
107 discovered by the LLVM fuzzer.
108
109 28. The assertion (?!) is optimized to (*FAIL). This was not handled correctly
110 when this assertion was used as a condition, for example (?(?!)a|b). In
111 pcre2_match() it worked by luck; in pcre2_dfa_match() it gave an incorrect
112 error about an unsupported item.
113
114 29. For some types of pattern, for example /Z*(|d*){216}/, the auto-
115 possessification code could take exponential time to complete. A recursion
116 depth limit of 1000 has been imposed to limit the resources used by this
117 optimization.
118
119 30. A pattern such as /(*UTF)[\S\V\H]/, which contains a negated special class
120 such as \S in non-UCP mode, explicit wide characters (> 255) can be ignored
121 because \S ensures they are all in the class. The code for doing this was
122 interacting badly with the code for computing the amount of space needed to
123 compile the pattern, leading to a buffer overflow. This bug was discovered
124 by the LLVM fuzzer.
125
126 31. A pattern such as /((?2)+)((?1))/ which has mutual recursion nested inside
127 other kinds of group caused stack overflow at compile time. This bug was
128 discovered by the LLVM fuzzer.
129
130
131 Version 8.36 26-September-2014
132 ------------------------------
133
134 1. Got rid of some compiler warnings in the C++ modules that were shown up by
135 -Wmissing-field-initializers and -Wunused-parameter.
136
137 2. The tests for quantifiers being too big (greater than 65535) were being
138 applied after reading the number, and stupidly assuming that integer
139 overflow would give a negative number. The tests are now applied as the
140 numbers are read.
141
142 3. Tidy code in pcre_exec.c where two branches that used to be different are
143 now the same.
144
145 4. The JIT compiler did not generate match limit checks for certain
146 bracketed expressions with quantifiers. This may lead to exponential
147 backtracking, instead of returning with PCRE_ERROR_MATCHLIMIT. This
148 issue should be resolved now.
149
150 5. Fixed an issue, which occures when nested alternatives are optimized
151 with table jumps.
152
153 6. Inserted two casts and changed some ints to size_t in the light of some
154 reported 64-bit compiler warnings (Bugzilla 1477).
155
156 7. Fixed a bug concerned with zero-minimum possessive groups that could match
157 an empty string, which sometimes were behaving incorrectly in the
158 interpreter (though correctly in the JIT matcher). This pcretest input is
159 an example:
160
161 '\A(?:[^"]++|"(?:[^"]*+|"")*+")++'
162 NON QUOTED "QUOT""ED" AFTER "NOT MATCHED
163
164 the interpreter was reporting a match of 'NON QUOTED ' only, whereas the
165 JIT matcher and Perl both matched 'NON QUOTED "QUOT""ED" AFTER '. The test
166 for an empty string was breaking the inner loop and carrying on at a lower
167 level, when possessive repeated groups should always return to a higher
168 level as they have no backtrack points in them. The empty string test now
169 occurs at the outer level.
170
171 8. Fixed a bug that was incorrectly auto-possessifying \w+ in the pattern
172 ^\w+(?>\s*)(?<=\w) which caused it not to match "test test".
173
174 9. Give a compile-time error for \o{} (as Perl does) and for \x{} (which Perl
175 doesn't).
176
177 10. Change 8.34/15 introduced a bug that caused the amount of memory needed
178 to hold a pattern to be incorrectly computed (too small) when there were
179 named back references to duplicated names. This could cause "internal
180 error: code overflow" or "double free or corruption" or other memory
181 handling errors.
182
183 11. When named subpatterns had the same prefixes, back references could be
184 confused. For example, in this pattern:
185
186 /(?P<Name>a)?(?P<Name2>b)?(?(<Name>)c|d)*l/
187
188 the reference to 'Name' was incorrectly treated as a reference to a
189 duplicate name.
190
191 12. A pattern such as /^s?c/mi8 where the optional character has more than
192 one "other case" was incorrectly compiled such that it would only try to
193 match starting at "c".
194
195 13. When a pattern starting with \s was studied, VT was not included in the
196 list of possible starting characters; this should have been part of the
197 8.34/18 patch.
198
199 14. If a character class started [\Qx]... where x is any character, the class
200 was incorrectly terminated at the ].
201
202 15. If a pattern that started with a caseless match for a character with more
203 than one "other case" was studied, PCRE did not set up the starting code
204 unit bit map for the list of possible characters. Now it does. This is an
205 optimization improvement, not a bug fix.
206
207 16. The Unicode data tables have been updated to Unicode 7.0.0.
208
209 17. Fixed a number of memory leaks in pcregrep.
210
211 18. Avoid a compiler warning (from some compilers) for a function call with
212 a cast that removes "const" from an lvalue by using an intermediate
213 variable (to which the compiler does not object).
214
215 19. Incorrect code was compiled if a group that contained an internal recursive
216 back reference was optional (had quantifier with a minimum of zero). This
217 example compiled incorrect code: /(((a\2)|(a*)\g<-1>))*/ and other examples
218 caused segmentation faults because of stack overflows at compile time.
219
220 20. A pattern such as /((?(R)a|(?1)))+/, which contains a recursion within a
221 group that is quantified with an indefinite repeat, caused a compile-time
222 loop which used up all the system stack and provoked a segmentation fault.
223 This was not the same bug as 19 above.
224
225 21. Add PCRECPP_EXP_DECL declaration to operator<< in pcre_stringpiece.h.
226 Patch by Mike Frysinger.
227
228
229 Version 8.35 04-April-2014
230 --------------------------
231
232 1. A new flag is set, when property checks are present in an XCLASS.
233 When this flag is not set, PCRE can perform certain optimizations
234 such as studying these XCLASS-es.
235
236 2. The auto-possessification of character sets were improved: a normal
237 and an extended character set can be compared now. Furthermore
238 the JIT compiler optimizes more character set checks.
239
240 3. Got rid of some compiler warnings for potentially uninitialized variables
241 that show up only when compiled with -O2.
242
243 4. A pattern such as (?=ab\K) that uses \K in an assertion can set the start
244 of a match later then the end of the match. The pcretest program was not
245 handling the case sensibly - it was outputting from the start to the next
246 binary zero. It now reports this situation in a message, and outputs the
247 text from the end to the start.
248
249 5. Fast forward search is improved in JIT. Instead of the first three
250 characters, any three characters with fixed position can be searched.
251 Search order: first, last, middle.
252
253 6. Improve character range checks in JIT. Characters are read by an inprecise
254 function now, which returns with an unknown value if the character code is
255 above a certain threshold (e.g: 256). The only limitation is that the value
256 must be bigger than the threshold as well. This function is useful when
257 the characters above the threshold are handled in the same way.
258
259 7. The macros whose names start with RAWUCHAR are placeholders for a future
260 mode in which only the bottom 21 bits of 32-bit data items are used. To
261 make this more memorable for those maintaining the code, the names have
262 been changed to start with UCHAR21, and an extensive comment has been added
263 to their definition.
264
265 8. Add missing (new) files sljitNativeTILEGX.c and sljitNativeTILEGX-encoder.c
266 to the export list in Makefile.am (they were accidentally omitted from the
267 8.34 tarball).
268
269 9. The informational output from pcretest used the phrase "starting byte set"
270 which is inappropriate for the 16-bit and 32-bit libraries. As the output
271 for "first char" and "need char" really means "non-UTF-char", I've changed
272 "byte" to "char", and slightly reworded the output. The documentation about
273 these values has also been (I hope) clarified.
274
275 10. Another JIT related optimization: use table jumps for selecting the correct
276 backtracking path, when more than four alternatives are present inside a
277 bracket.
278
279 11. Empty match is not possible, when the minimum length is greater than zero,
280 and there is no \K in the pattern. JIT should avoid empty match checks in
281 such cases.
282
283 12. In a caseless character class with UCP support, when a character with more
284 than one alternative case was not the first character of a range, not all
285 the alternative cases were added to the class. For example, s and \x{17f}
286 are both alternative cases for S: the class [RST] was handled correctly,
287 but [R-T] was not.
288
289 13. The configure.ac file always checked for pthread support when JIT was
290 enabled. This is not used in Windows, so I have put this test inside a
291 check for the presence of windows.h (which was already tested for).
292
293 14. Improve pattern prefix search by a simplified Boyer-Moore algorithm in JIT.
294 The algorithm provides a way to skip certain starting offsets, and usually
295 faster than linear prefix searches.
296
297 15. Change 13 for 8.20 updated RunTest to check for the 'fr' locale as well
298 as for 'fr_FR' and 'french'. For some reason, however, it then used the
299 Windows-specific input and output files, which have 'french' screwed in.
300 So this could never have worked. One of the problems with locales is that
301 they aren't always the same. I have now updated RunTest so that it checks
302 the output of the locale test (test 3) against three different output
303 files, and it allows the test to pass if any one of them matches. With luck
304 this should make the test pass on some versions of Solaris where it was
305 failing. Because of the uncertainty, the script did not used to stop if
306 test 3 failed; it now does. If further versions of a French locale ever
307 come to light, they can now easily be added.
308
309 16. If --with-pcregrep-bufsize was given a non-integer value such as "50K",
310 there was a message during ./configure, but it did not stop. This now
311 provokes an error. The invalid example in README has been corrected.
312 If a value less than the minimum is given, the minimum value has always
313 been used, but now a warning is given.
314
315 17. If --enable-bsr-anycrlf was set, the special 16/32-bit test failed. This
316 was a bug in the test system, which is now fixed. Also, the list of various
317 configurations that are tested for each release did not have one with both
318 16/32 bits and --enable-bar-anycrlf. It now does.
319
320 18. pcretest was missing "-C bsr" for displaying the \R default setting.
321
322 19. Little endian PowerPC systems are supported now by the JIT compiler.
323
324 20. The fast forward newline mechanism could enter to an infinite loop on
325 certain invalid UTF-8 input. Although we don't support these cases
326 this issue can be fixed by a performance optimization.
327
328 21. Change 33 of 8.34 is not sufficient to ensure stack safety because it does
329 not take account if existing stack usage. There is now a new global
330 variable called pcre_stack_guard that can be set to point to an external
331 function to check stack availability. It is called at the start of
332 processing every parenthesized group.
333
334 22. A typo in the code meant that in ungreedy mode the max/min qualifier
335 behaved like a min-possessive qualifier, and, for example, /a{1,3}b/U did
336 not match "ab".
337
338 23. When UTF was disabled, the JIT program reported some incorrect compile
339 errors. These messages are silenced now.
340
341 24. Experimental support for ARM-64 and MIPS-64 has been added to the JIT
342 compiler.
343
344 25. Change all the temporary files used in RunGrepTest to be different to those
345 used by RunTest so that the tests can be run simultaneously, for example by
346 "make -j check".
347
348
349 Version 8.34 15-December-2013
350 -----------------------------
351
352 1. Add pcre[16|32]_jit_free_unused_memory to forcibly free unused JIT
353 executable memory. Patch inspired by Carsten Klein.
354
355 2. ./configure --enable-coverage defined SUPPORT_GCOV in config.h, although
356 this macro is never tested and has no effect, because the work to support
357 coverage involves only compiling and linking options and special targets in
358 the Makefile. The comment in config.h implied that defining the macro would
359 enable coverage support, which is totally false. There was also support for
360 setting this macro in the CMake files (my fault, I just copied it from
361 configure). SUPPORT_GCOV has now been removed.
362
363 3. Make a small performance improvement in strlen16() and strlen32() in
364 pcretest.
365
366 4. Change 36 for 8.33 left some unreachable statements in pcre_exec.c,
367 detected by the Solaris compiler (gcc doesn't seem to be able to diagnose
368 these cases). There was also one in pcretest.c.
369
370 5. Cleaned up a "may be uninitialized" compiler warning in pcre_exec.c.
371
372 6. In UTF mode, the code for checking whether a group could match an empty
373 string (which is used for indefinitely repeated groups to allow for
374 breaking an infinite loop) was broken when the group contained a repeated
375 negated single-character class with a character that occupied more than one
376 data item and had a minimum repetition of zero (for example, [^\x{100}]* in
377 UTF-8 mode). The effect was undefined: the group might or might not be
378 deemed as matching an empty string, or the program might have crashed.
379
380 7. The code for checking whether a group could match an empty string was not
381 recognizing that \h, \H, \v, \V, and \R must match a character.
382
383 8. Implemented PCRE_INFO_MATCH_EMPTY, which yields 1 if the pattern can match
384 an empty string. If it can, pcretest shows this in its information output.
385
386 9. Fixed two related bugs that applied to Unicode extended grapheme clusters
387 that were repeated with a maximizing qualifier (e.g. \X* or \X{2,5}) when
388 matched by pcre_exec() without using JIT:
389
390 (a) If the rest of the pattern did not match after a maximal run of
391 grapheme clusters, the code for backing up to try with fewer of them
392 did not always back up over a full grapheme when characters that do not
393 have the modifier quality were involved, e.g. Hangul syllables.
394
395 (b) If the match point in a subject started with modifier character, and
396 there was no match, the code could incorrectly back up beyond the match
397 point, and potentially beyond the first character in the subject,
398 leading to a segfault or an incorrect match result.
399
400 10. A conditional group with an assertion condition could lead to PCRE
401 recording an incorrect first data item for a match if no other first data
402 item was recorded. For example, the pattern (?(?=ab)ab) recorded "a" as a
403 first data item, and therefore matched "ca" after "c" instead of at the
404 start.
405
406 11. Change 40 for 8.33 (allowing pcregrep to find empty strings) showed up a
407 bug that caused the command "echo a | ./pcregrep -M '|a'" to loop.
408
409 12. The source of pcregrep now includes z/OS-specific code so that it can be
410 compiled for z/OS as part of the special z/OS distribution.
411
412 13. Added the -T and -TM options to pcretest.
413
414 14. The code in pcre_compile.c for creating the table of named capturing groups
415 has been refactored. Instead of creating the table dynamically during the
416 actual compiling pass, the information is remembered during the pre-compile
417 pass (on the stack unless there are more than 20 named groups, in which
418 case malloc() is used) and the whole table is created before the actual
419 compile happens. This has simplified the code (it is now nearly 150 lines
420 shorter) and prepared the way for better handling of references to groups
421 with duplicate names.
422
423 15. A back reference to a named subpattern when there is more than one of the
424 same name now checks them in the order in which they appear in the pattern.
425 The first one that is set is used for the reference. Previously only the
426 first one was inspected. This change makes PCRE more compatible with Perl.
427
428 16. Unicode character properties were updated from Unicode 6.3.0.
429
430 17. The compile-time code for auto-possessification has been refactored, based
431 on a patch by Zoltan Herczeg. It now happens after instead of during
432 compilation. The code is cleaner, and more cases are handled. The option
433 PCRE_NO_AUTO_POSSESS is added for testing purposes, and the -O and /O
434 options in pcretest are provided to set it. It can also be set by
435 (*NO_AUTO_POSSESS) at the start of a pattern.
436
437 18. The character VT has been added to the default ("C" locale) set of
438 characters that match \s and are generally treated as white space,
439 following this same change in Perl 5.18. There is now no difference between
440 "Perl space" and "POSIX space". Whether VT is treated as white space in
441 other locales depends on the locale.
442
443 19. The code for checking named groups as conditions, either for being set or
444 for being recursed, has been refactored (this is related to 14 and 15
445 above). Processing unduplicated named groups should now be as fast at
446 numerical groups, and processing duplicated groups should be faster than
447 before.
448
449 20. Two patches to the CMake build system, by Alexander Barkov:
450
451 (1) Replace the "source" command by "." in CMakeLists.txt because
452 "source" is a bash-ism.
453
454 (2) Add missing HAVE_STDINT_H and HAVE_INTTYPES_H to config-cmake.h.in;
455 without these the CMake build does not work on Solaris.
456
457 21. Perl has changed its handling of \8 and \9. If there is no previously
458 encountered capturing group of those numbers, they are treated as the
459 literal characters 8 and 9 instead of a binary zero followed by the
460 literals. PCRE now does the same.
461
462 22. Following Perl, added \o{} to specify codepoints in octal, making it
463 possible to specify values greater than 0777 and also making them
464 unambiguous.
465
466 23. Perl now gives an error for missing closing braces after \x{... instead of
467 treating the string as literal. PCRE now does the same.
468
469 24. RunTest used to grumble if an inappropriate test was selected explicitly,
470 but just skip it when running all tests. This make it awkward to run ranges
471 of tests when one of them was inappropriate. Now it just skips any
472 inappropriate tests, as it always did when running all tests.
473
474 25. If PCRE_AUTO_CALLOUT and PCRE_UCP were set for a pattern that contained
475 character types such as \d or \w, too many callouts were inserted, and the
476 data that they returned was rubbish.
477
478 26. In UCP mode, \s was not matching two of the characters that Perl matches,
479 namely NEL (U+0085) and MONGOLIAN VOWEL SEPARATOR (U+180E), though they
480 were matched by \h. The code has now been refactored so that the lists of
481 the horizontal and vertical whitespace characters used for \h and \v (which
482 are defined only in one place) are now also used for \s.
483
484 27. Add JIT support for the 64 bit TileGX architecture.
485 Patch by Jiong Wang (Tilera Corporation).
486
487 28. Possessive quantifiers for classes (both explicit and automatically
488 generated) now use special opcodes instead of wrapping in ONCE brackets.
489
490 29. Whereas an item such as A{4}+ ignored the possessivenes of the quantifier
491 (because it's meaningless), this was not happening when PCRE_CASELESS was
492 set. Not wrong, but inefficient.
493
494 30. Updated perltest.pl to add /u (force Unicode mode) when /W (use Unicode
495 properties for \w, \d, etc) is present in a test regex. Otherwise if the
496 test contains no characters greater than 255, Perl doesn't realise it
497 should be using Unicode semantics.
498
499 31. Upgraded the handling of the POSIX classes [:graph:], [:print:], and
500 [:punct:] when PCRE_UCP is set so as to include the same characters as Perl
501 does in Unicode mode.
502
503 32. Added the "forbid" facility to pcretest so that putting tests into the
504 wrong test files can sometimes be quickly detected.
505
506 33. There is now a limit (default 250) on the depth of nesting of parentheses.
507 This limit is imposed to control the amount of system stack used at compile
508 time. It can be changed at build time by --with-parens-nest-limit=xxx or
509 the equivalent in CMake.
510
511 34. Character classes such as [A-\d] or [a-[:digit:]] now cause compile-time
512 errors. Perl warns for these when in warning mode, but PCRE has no facility
513 for giving warnings.
514
515 35. Change 34 for 8.13 allowed quantifiers on assertions, because Perl does.
516 However, this was not working for (?!) because it is optimized to (*FAIL),
517 for which PCRE does not allow quantifiers. The optimization is now disabled
518 when a quantifier follows (?!). I can't see any use for this, but it makes
519 things uniform.
520
521 36. Perl no longer allows group names to start with digits, so I have made this
522 change also in PCRE. It simplifies the code a bit.
523
524 37. In extended mode, Perl ignores spaces before a + that indicates a
525 possessive quantifier. PCRE allowed a space before the quantifier, but not
526 before the possessive +. It now does.
527
528 38. The use of \K (reset reported match start) within a repeated possessive
529 group such as (a\Kb)*+ was not working.
530
531 40. Document that the same character tables must be used at compile time and
532 run time, and that the facility to pass tables to pcre_exec() and
533 pcre_dfa_exec() is for use only with saved/restored patterns.
534
535 41. Applied Jeff Trawick's patch CMakeLists.txt, which "provides two new
536 features for Builds with MSVC:
537
538 1. Support pcre.rc and/or pcreposix.rc (as is already done for MinGW
539 builds). The .rc files can be used to set FileDescription and many other
540 attributes.
541
542 2. Add an option (-DINSTALL_MSVC_PDB) to enable installation of .pdb files.
543 This allows higher-level build scripts which want .pdb files to avoid
544 hard-coding the exact files needed."
545
546 42. Added support for [[:<:]] and [[:>:]] as used in the BSD POSIX library to
547 mean "start of word" and "end of word", respectively, as a transition aid.
548
549 43. A minimizing repeat of a class containing codepoints greater than 255 in
550 non-UTF 16-bit or 32-bit modes caused an internal error when PCRE was
551 compiled to use the heap for recursion.
552
553 44. Got rid of some compiler warnings for unused variables when UTF but not UCP
554 is configured.
555
556
557 Version 8.33 28-May-2013
558 ------------------------
559
560 1. Added 'U' to some constants that are compared to unsigned integers, to
561 avoid compiler signed/unsigned warnings. Added (int) casts to unsigned
562 variables that are added to signed variables, to ensure the result is
563 signed and can be negated.
564
565 2. Applied patch by Daniel Richard G for quashing MSVC warnings to the
566 CMake config files.
567
568 3. Revise the creation of config.h.generic so that all boolean macros are
569 #undefined, whereas non-boolean macros are #ifndef/#endif-ed. This makes
570 overriding via -D on the command line possible.
571
572 4. Changing the definition of the variable "op" in pcre_exec.c from pcre_uchar
573 to unsigned int is reported to make a quite noticeable speed difference in
574 a specific Windows environment. Testing on Linux did also appear to show
575 some benefit (and it is clearly not harmful). Also fixed the definition of
576 Xop which should be unsigned.
577
578 5. Related to (4), changing the definition of the intermediate variable cc
579 in repeated character loops from pcre_uchar to pcre_uint32 also gave speed
580 improvements.
581
582 6. Fix forward search in JIT when link size is 3 or greater. Also removed some
583 unnecessary spaces.
584
585 7. Adjust autogen.sh and configure.ac to lose warnings given by automake 1.12
586 and later.
587
588 8. Fix two buffer over read issues in 16 and 32 bit modes. Affects JIT only.
589
590 9. Optimizing fast_forward_start_bits in JIT.
591
592 10. Adding support for callouts in JIT, and fixing some issues revealed
593 during this work. Namely:
594
595 (a) Unoptimized capturing brackets incorrectly reset on backtrack.
596
597 (b) Minimum length was not checked before the matching is started.
598
599 11. The value of capture_last that is passed to callouts was incorrect in some
600 cases when there was a capture on one path that was subsequently abandoned
601 after a backtrack. Also, the capture_last value is now reset after a
602 recursion, since all captures are also reset in this case.
603
604 12. The interpreter no longer returns the "too many substrings" error in the
605 case when an overflowing capture is in a branch that is subsequently
606 abandoned after a backtrack.
607
608 13. In the pathological case when an offset vector of size 2 is used, pcretest
609 now prints out the matched string after a yield of 0 or 1.
610
611 14. Inlining subpatterns in recursions, when certain conditions are fulfilled.
612 Only supported by the JIT compiler at the moment.
613
614 15. JIT compiler now supports 32 bit Macs thanks to Lawrence Velazquez.
615
616 16. Partial matches now set offsets[2] to the "bumpalong" value, that is, the
617 offset of the starting point of the matching process, provided the offsets
618 vector is large enough.
619
620 17. The \A escape now records a lookbehind value of 1, though its execution
621 does not actually inspect the previous character. This is to ensure that,
622 in partial multi-segment matching, at least one character from the old
623 segment is retained when a new segment is processed. Otherwise, if there
624 are no lookbehinds in the pattern, \A might match incorrectly at the start
625 of a new segment.
626
627 18. Added some #ifdef __VMS code into pcretest.c to help VMS implementations.
628
629 19. Redefined some pcre_uchar variables in pcre_exec.c as pcre_uint32; this
630 gives some modest performance improvement in 8-bit mode.
631
632 20. Added the PCRE-specific property \p{Xuc} for matching characters that can
633 be expressed in certain programming languages using Universal Character
634 Names.
635
636 21. Unicode validation has been updated in the light of Unicode Corrigendum #9,
637 which points out that "non characters" are not "characters that may not
638 appear in Unicode strings" but rather "characters that are reserved for
639 internal use and have only local meaning".
640
641 22. When a pattern was compiled with automatic callouts (PCRE_AUTO_CALLOUT) and
642 there was a conditional group that depended on an assertion, if the
643 assertion was false, the callout that immediately followed the alternation
644 in the condition was skipped when pcre_exec() was used for matching.
645
646 23. Allow an explicit callout to be inserted before an assertion that is the
647 condition for a conditional group, for compatibility with automatic
648 callouts, which always insert a callout at this point.
649
650 24. In 8.31, (*COMMIT) was confined to within a recursive subpattern. Perl also
651 confines (*SKIP) and (*PRUNE) in the same way, and this has now been done.
652
653 25. (*PRUNE) is now supported by the JIT compiler.
654
655 26. Fix infinite loop when /(?<=(*SKIP)ac)a/ is matched against aa.
656
657 27. Fix the case where there are two or more SKIPs with arguments that may be
658 ignored.
659
660 28. (*SKIP) is now supported by the JIT compiler.
661
662 29. (*THEN) is now supported by the JIT compiler.
663
664 30. Update RunTest with additional test selector options.
665
666 31. The way PCRE handles backtracking verbs has been changed in two ways.
667
668 (1) Previously, in something like (*COMMIT)(*SKIP), COMMIT would override
669 SKIP. Now, PCRE acts on whichever backtracking verb is reached first by
670 backtracking. In some cases this makes it more Perl-compatible, but Perl's
671 rather obscure rules do not always do the same thing.
672
673 (2) Previously, backtracking verbs were confined within assertions. This is
674 no longer the case for positive assertions, except for (*ACCEPT). Again,
675 this sometimes improves Perl compatibility, and sometimes does not.
676
677 32. A number of tests that were in test 2 because Perl did things differently
678 have been moved to test 1, because either Perl or PCRE has changed, and
679 these tests are now compatible.
680
681 32. Backtracking control verbs are now handled in the same way in JIT and
682 interpreter.
683
684 33. An opening parenthesis in a MARK/PRUNE/SKIP/THEN name in a pattern that
685 contained a forward subroutine reference caused a compile error.
686
687 34. Auto-detect and optimize limited repetitions in JIT.
688
689 35. Implement PCRE_NEVER_UTF to lock out the use of UTF, in particular,
690 blocking (*UTF) etc.
691
692 36. In the interpreter, maximizing pattern repetitions for characters and
693 character types now use tail recursion, which reduces stack usage.
694
695 37. The value of the max lookbehind was not correctly preserved if a compiled
696 and saved regex was reloaded on a host of different endianness.
697
698 38. Implemented (*LIMIT_MATCH) and (*LIMIT_RECURSION). As part of the extension
699 of the compiled pattern block, expand the flags field from 16 to 32 bits
700 because it was almost full.
701
702 39. Try madvise first before posix_madvise.
703
704 40. Change 7 for PCRE 7.9 made it impossible for pcregrep to find empty lines
705 with a pattern such as ^$. It has taken 4 years for anybody to notice! The
706 original change locked out all matches of empty strings. This has been
707 changed so that one match of an empty string per line is recognized.
708 Subsequent searches on the same line (for colouring or for --only-matching,
709 for example) do not recognize empty strings.
710
711 41. Applied a user patch to fix a number of spelling mistakes in comments.
712
713 42. Data lines longer than 65536 caused pcretest to crash.
714
715 43. Clarified the data type for length and startoffset arguments for pcre_exec
716 and pcre_dfa_exec in the function-specific man pages, where they were
717 explicitly stated to be in bytes, never having been updated. I also added
718 some clarification to the pcreapi man page.
719
720 44. A call to pcre_dfa_exec() with an output vector size less than 2 caused
721 a segmentation fault.
722
723
724 Version 8.32 30-November-2012
725 -----------------------------
726
727 1. Improved JIT compiler optimizations for first character search and single
728 character iterators.
729
730 2. Supporting IBM XL C compilers for PPC architectures in the JIT compiler.
731 Patch by Daniel Richard G.
732
733 3. Single character iterator optimizations in the JIT compiler.
734
735 4. Improved JIT compiler optimizations for character ranges.
736
737 5. Rename the "leave" variable names to "quit" to improve WinCE compatibility.
738 Reported by Giuseppe D'Angelo.
739
740 6. The PCRE_STARTLINE bit, indicating that a match can occur only at the start
741 of a line, was being set incorrectly in cases where .* appeared inside
742 atomic brackets at the start of a pattern, or where there was a subsequent
743 *PRUNE or *SKIP.
744
745 7. Improved instruction cache flush for POWER/PowerPC.
746 Patch by Daniel Richard G.
747
748 8. Fixed a number of issues in pcregrep, making it more compatible with GNU
749 grep:
750
751 (a) There is now no limit to the number of patterns to be matched.
752
753 (b) An error is given if a pattern is too long.
754
755 (c) Multiple uses of --exclude, --exclude-dir, --include, and --include-dir
756 are now supported.
757
758 (d) --exclude-from and --include-from (multiple use) have been added.
759
760 (e) Exclusions and inclusions now apply to all files and directories, not
761 just to those obtained from scanning a directory recursively.
762
763 (f) Multiple uses of -f and --file-list are now supported.
764
765 (g) In a Windows environment, the default for -d has been changed from
766 "read" (the GNU grep default) to "skip", because otherwise the presence
767 of a directory in the file list provokes an error.
768
769 (h) The documentation has been revised and clarified in places.
770
771 9. Improve the matching speed of capturing brackets.
772
773 10. Changed the meaning of \X so that it now matches a Unicode extended
774 grapheme cluster.
775
776 11. Patch by Daniel Richard G to the autoconf files to add a macro for sorting
777 out POSIX threads when JIT support is configured.
778
779 12. Added support for PCRE_STUDY_EXTRA_NEEDED.
780
781 13. In the POSIX wrapper regcomp() function, setting re_nsub field in the preg
782 structure could go wrong in environments where size_t is not the same size
783 as int.
784
785 14. Applied user-supplied patch to pcrecpp.cc to allow PCRE_NO_UTF8_CHECK to be
786 set.
787
788 15. The EBCDIC support had decayed; later updates to the code had included
789 explicit references to (e.g.) \x0a instead of CHAR_LF. There has been a
790 general tidy up of EBCDIC-related issues, and the documentation was also
791 not quite right. There is now a test that can be run on ASCII systems to
792 check some of the EBCDIC-related things (but is it not a full test).
793
794 16. The new PCRE_STUDY_EXTRA_NEEDED option is now used by pcregrep, resulting
795 in a small tidy to the code.
796
797 17. Fix JIT tests when UTF is disabled and both 8 and 16 bit mode are enabled.
798
799 18. If the --only-matching (-o) option in pcregrep is specified multiple
800 times, each one causes appropriate output. For example, -o1 -o2 outputs the
801 substrings matched by the 1st and 2nd capturing parentheses. A separating
802 string can be specified by --om-separator (default empty).
803
804 19. Improving the first n character searches.
805
806 20. Turn case lists for horizontal and vertical white space into macros so that
807 they are defined only once.
808
809 21. This set of changes together give more compatible Unicode case-folding
810 behaviour for characters that have more than one other case when UCP
811 support is available.
812
813 (a) The Unicode property table now has offsets into a new table of sets of
814 three or more characters that are case-equivalent. The MultiStage2.py
815 script that generates these tables (the pcre_ucd.c file) now scans
816 CaseFolding.txt instead of UnicodeData.txt for character case
817 information.
818
819 (b) The code for adding characters or ranges of characters to a character
820 class has been abstracted into a generalized function that also handles
821 case-independence. In UTF-mode with UCP support, this uses the new data
822 to handle characters with more than one other case.
823
824 (c) A bug that is fixed as a result of (b) is that codepoints less than 256
825 whose other case is greater than 256 are now correctly matched
826 caselessly. Previously, the high codepoint matched the low one, but not
827 vice versa.
828
829 (d) The processing of \h, \H, \v, and \ in character classes now makes use
830 of the new class addition function, using character lists defined as
831 macros alongside the case definitions of 20 above.
832
833 (e) Caseless back references now work with characters that have more than
834 one other case.
835
836 (f) General caseless matching of characters with more than one other case
837 is supported.
838
839 22. Unicode character properties were updated from Unicode 6.2.0
840
841 23. Improved CMake support under Windows. Patch by Daniel Richard G.
842
843 24. Add support for 32-bit character strings, and UTF-32
844
845 25. Major JIT compiler update (code refactoring and bugfixing).
846 Experimental Sparc 32 support is added.
847
848 26. Applied a modified version of Daniel Richard G's patch to create
849 pcre.h.generic and config.h.generic by "make" instead of in the
850 PrepareRelease script.
851
852 27. Added a definition for CHAR_NULL (helpful for the z/OS port), and use it in
853 pcre_compile.c when checking for a zero character.
854
855 28. Introducing a native interface for JIT. Through this interface, the compiled
856 machine code can be directly executed. The purpose of this interface is to
857 provide fast pattern matching, so several sanity checks are not performed.
858 However, feature tests are still performed. The new interface provides
859 1.4x speedup compared to the old one.
860
861 29. If pcre_exec() or pcre_dfa_exec() was called with a negative value for
862 the subject string length, the error given was PCRE_ERROR_BADOFFSET, which
863 was confusing. There is now a new error PCRE_ERROR_BADLENGTH for this case.
864
865 30. In 8-bit UTF-8 mode, pcretest failed to give an error for data codepoints
866 greater than 0x7fffffff (which cannot be represented in UTF-8, even under
867 the "old" RFC 2279). Instead, it ended up passing a negative length to
868 pcre_exec().
869
870 31. Add support for GCC's visibility feature to hide internal functions.
871
872 32. Running "pcretest -C pcre8" or "pcretest -C pcre16" gave a spurious error
873 "unknown -C option" after outputting 0 or 1.
874
875 33. There is now support for generating a code coverage report for the test
876 suite in environments where gcc is the compiler and lcov is installed. This
877 is mainly for the benefit of the developers.
878
879 34. If PCRE is built with --enable-valgrind, certain memory regions are marked
880 unaddressable using valgrind annotations, allowing valgrind to detect
881 invalid memory accesses. This is mainly for the benefit of the developers.
882
883 25. (*UTF) can now be used to start a pattern in any of the three libraries.
884
885 26. Give configure error if --enable-cpp but no C++ compiler found.
886
887
888 Version 8.31 06-July-2012
889 -------------------------
890
891 1. Fixing a wrong JIT test case and some compiler warnings.
892
893 2. Removed a bashism from the RunTest script.
894
895 3. Add a cast to pcre_exec.c to fix the warning "unary minus operator applied
896 to unsigned type, result still unsigned" that was given by an MS compiler
897 on encountering the code "-sizeof(xxx)".
898
899 4. Partial matching support is added to the JIT compiler.
900
901 5. Fixed several bugs concerned with partial matching of items that consist
902 of more than one character:
903
904 (a) /^(..)\1/ did not partially match "aba" because checking references was
905 done on an "all or nothing" basis. This also applied to repeated
906 references.
907
908 (b) \R did not give a hard partial match if \r was found at the end of the
909 subject.
910
911 (c) \X did not give a hard partial match after matching one or more
912 characters at the end of the subject.
913
914 (d) When newline was set to CRLF, a pattern such as /a$/ did not recognize
915 a partial match for the string "\r".
916
917 (e) When newline was set to CRLF, the metacharacter "." did not recognize
918 a partial match for a CR character at the end of the subject string.
919
920 6. If JIT is requested using /S++ or -s++ (instead of just /S+ or -s+) when
921 running pcretest, the text "(JIT)" added to the output whenever JIT is
922 actually used to run the match.
923
924 7. Individual JIT compile options can be set in pcretest by following -s+[+]
925 or /S+[+] with a digit between 1 and 7.
926
927 8. OP_NOT now supports any UTF character not just single-byte ones.
928
929 9. (*MARK) control verb is now supported by the JIT compiler.
930
931 10. The command "./RunTest list" lists the available tests without actually
932 running any of them. (Because I keep forgetting what they all are.)
933
934 11. Add PCRE_INFO_MAXLOOKBEHIND.
935
936 12. Applied a (slightly modified) user-supplied patch that improves performance
937 when the heap is used for recursion (compiled with --disable-stack-for-
938 recursion). Instead of malloc and free for each heap frame each time a
939 logical recursion happens, frames are retained on a chain and re-used where
940 possible. This sometimes gives as much as 30% improvement.
941
942 13. As documented, (*COMMIT) is now confined to within a recursive subpattern
943 call.
944
945 14. As documented, (*COMMIT) is now confined to within a positive assertion.
946
947 15. It is now possible to link pcretest with libedit as an alternative to
948 libreadline.
949
950 16. (*COMMIT) control verb is now supported by the JIT compiler.
951
952 17. The Unicode data tables have been updated to Unicode 6.1.0.
953
954 18. Added --file-list option to pcregrep.
955
956 19. Added binary file support to pcregrep, including the -a, --binary-files,
957 -I, and --text options.
958
959 20. The madvise function is renamed for posix_madvise for QNX compatibility
960 reasons. Fixed by Giuseppe D'Angelo.
961
962 21. Fixed a bug for backward assertions with REVERSE 0 in the JIT compiler.
963
964 22. Changed the option for creating symbolic links for 16-bit man pages from
965 -s to -sf so that re-installing does not cause issues.
966
967 23. Support PCRE_NO_START_OPTIMIZE in JIT as (*MARK) support requires it.
968
969 24. Fixed a very old bug in pcretest that caused errors with restarted DFA
970 matches in certain environments (the workspace was not being correctly
971 retained). Also added to pcre_dfa_exec() a simple plausibility check on
972 some of the workspace data at the beginning of a restart.
973
974 25. \s*\R was auto-possessifying the \s* when it should not, whereas \S*\R
975 was not doing so when it should - probably a typo introduced by SVN 528
976 (change 8.10/14).
977
978 26. When PCRE_UCP was not set, \w+\x{c4} was incorrectly auto-possessifying the
979 \w+ when the character tables indicated that \x{c4} was a word character.
980 There were several related cases, all because the tests for doing a table
981 lookup were testing for characters less than 127 instead of 255.
982
983 27. If a pattern contains capturing parentheses that are not used in a match,
984 their slots in the ovector are set to -1. For those that are higher than
985 any matched groups, this happens at the end of processing. In the case when
986 there were back references that the ovector was too small to contain
987 (causing temporary malloc'd memory to be used during matching), and the
988 highest capturing number was not used, memory off the end of the ovector
989 was incorrectly being set to -1. (It was using the size of the temporary
990 memory instead of the true size.)
991
992 28. To catch bugs like 27 using valgrind, when pcretest is asked to specify an
993 ovector size, it uses memory at the end of the block that it has got.
994
995 29. Check for an overlong MARK name and give an error at compile time. The
996 limit is 255 for the 8-bit library and 65535 for the 16-bit library.
997
998 30. JIT compiler update.
999
1000 31. JIT is now supported on jailbroken iOS devices. Thanks for Ruiger
1001 Rill for the patch.
1002
1003 32. Put spaces around SLJIT_PRINT_D in the JIT compiler. Required by CXX11.
1004
1005 33. Variable renamings in the PCRE-JIT compiler. No functionality change.
1006
1007 34. Fixed typos in pcregrep: in two places there was SUPPORT_LIBZ2 instead of
1008 SUPPORT_LIBBZ2. This caused a build problem when bzip2 but not gzip (zlib)
1009 was enabled.
1010
1011 35. Improve JIT code generation for greedy plus quantifier.
1012
1013 36. When /((?:a?)*)*c/ or /((?>a?)*)*c/ was matched against "aac", it set group
1014 1 to "aa" instead of to an empty string. The bug affected repeated groups
1015 that could potentially match an empty string.
1016
1017 37. Optimizing single character iterators in JIT.
1018
1019 38. Wide characters specified with \uxxxx in JavaScript mode are now subject to
1020 the same checks as \x{...} characters in non-JavaScript mode. Specifically,
1021 codepoints that are too big for the mode are faulted, and in a UTF mode,
1022 disallowed codepoints are also faulted.
1023
1024 39. If PCRE was compiled with UTF support, in three places in the DFA
1025 matcher there was code that should only have been obeyed in UTF mode, but
1026 was being obeyed unconditionally. In 8-bit mode this could cause incorrect
1027 processing when bytes with values greater than 127 were present. In 16-bit
1028 mode the bug would be provoked by values in the range 0xfc00 to 0xdc00. In
1029 both cases the values are those that cannot be the first data item in a UTF
1030 character. The three items that might have provoked this were recursions,
1031 possessively repeated groups, and atomic groups.
1032
1033 40. Ensure that libpcre is explicitly listed in the link commands for pcretest
1034 and pcregrep, because some OS require shared objects to be explicitly
1035 passed to ld, causing the link step to fail if they are not.
1036
1037 41. There were two incorrect #ifdefs in pcre_study.c, meaning that, in 16-bit
1038 mode, patterns that started with \h* or \R* might be incorrectly matched.
1039
1040
1041 Version 8.30 04-February-2012
1042 -----------------------------
1043
1044 1. Renamed "isnumber" as "is_a_number" because in some Mac environments this
1045 name is defined in ctype.h.
1046
1047 2. Fixed a bug in fixed-length calculation for lookbehinds that would show up
1048 only in quite long subpatterns.
1049
1050 3. Removed the function pcre_info(), which has been obsolete and deprecated
1051 since it was replaced by pcre_fullinfo() in February 2000.
1052
1053 4. For a non-anchored pattern, if (*SKIP) was given with a name that did not
1054 match a (*MARK), and the match failed at the start of the subject, a
1055 reference to memory before the start of the subject could occur. This bug
1056 was introduced by fix 17 of release 8.21.
1057
1058 5. A reference to an unset group with zero minimum repetition was giving
1059 totally wrong answers (in non-JavaScript-compatibility mode). For example,
1060 /(another)?(\1?)test/ matched against "hello world test". This bug was
1061 introduced in release 8.13.
1062
1063 6. Add support for 16-bit character strings (a large amount of work involving
1064 many changes and refactorings).
1065
1066 7. RunGrepTest failed on msys because \r\n was replaced by whitespace when the
1067 command "pattern=`printf 'xxx\r\njkl'`" was run. The pattern is now taken
1068 from a file.
1069
1070 8. Ovector size of 2 is also supported by JIT based pcre_exec (the ovector size
1071 rounding is not applied in this particular case).
1072
1073 9. The invalid Unicode surrogate codepoints U+D800 to U+DFFF are now rejected
1074 if they appear, or are escaped, in patterns.
1075
1076 10. Get rid of a number of -Wunused-but-set-variable warnings.
1077
1078 11. The pattern /(?=(*:x))(q|)/ matches an empty string, and returns the mark
1079 "x". The similar pattern /(?=(*:x))((*:y)q|)/ did not return a mark at all.
1080 Oddly, Perl behaves the same way. PCRE has been fixed so that this pattern
1081 also returns the mark "x". This bug applied to capturing parentheses,
1082 non-capturing parentheses, and atomic parentheses. It also applied to some
1083 assertions.
1084
1085 12. Stephen Kelly's patch to CMakeLists.txt allows it to parse the version
1086 information out of configure.ac instead of relying on pcre.h.generic, which
1087 is not stored in the repository.
1088
1089 13. Applied Dmitry V. Levin's patch for a more portable method for linking with
1090 -lreadline.
1091
1092 14. ZH added PCRE_CONFIG_JITTARGET; added its output to pcretest -C.
1093
1094 15. Applied Graycode's patch to put the top-level frame on the stack rather
1095 than the heap when not using the stack for recursion. This gives a
1096 performance improvement in many cases when recursion is not deep.
1097
1098 16. Experimental code added to "pcretest -C" to output the stack frame size.
1099
1100
1101 Version 8.21 12-Dec-2011
1102 ------------------------
1103
1104 1. Updating the JIT compiler.
1105
1106 2. JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
1107 are added as well.
1108
1109 3. Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
1110 PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
1111 calling _pcre_jit_exec. Some extra comments are added.
1112
1113 4. (*MARK) settings inside atomic groups that do not contain any capturing
1114 parentheses, for example, (?>a(*:m)), were not being passed out. This bug
1115 was introduced by change 18 for 8.20.
1116
1117 5. Supporting of \x, \U and \u in JavaScript compatibility mode based on the
1118 ECMA-262 standard.
1119
1120 6. Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
1121 erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
1122 This bug was probably introduced by change 9 of 8.13.
1123
1124 7. While fixing 6 above, I noticed that a number of other items were being
1125 incorrectly rejected as "not fixed length". This arose partly because newer
1126 opcodes had not been added to the fixed-length checking code. I have (a)
1127 corrected the bug and added tests for these items, and (b) arranged for an
1128 error to occur if an unknown opcode is encountered while checking for fixed
1129 length instead of just assuming "not fixed length". The items that were
1130 rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
1131 (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
1132 repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
1133
1134 8. A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
1135 being incorrectly compiled and would have given unpredicatble results.
1136
1137 9. A possessively repeated subpattern with minimum repeat count greater than
1138 one behaved incorrectly. For example, (A){2,}+ behaved as if it was
1139 (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
1140 the first (A) could occur when it should not.
1141
1142 10. Add a cast and remove a redundant test from the code.
1143
1144 11. JIT should use pcre_malloc/pcre_free for allocation.
1145
1146 12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
1147 best practice nowadays, and helps with cross-compiling. (If the exec_prefix
1148 is anything other than /usr, -L is still shown).
1149
1150 13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
1151
1152 14. Perl does not support \N without a following name in a [] class; PCRE now
1153 also gives an error.
1154
1155 15. If a forward reference was repeated with an upper limit of around 2000,
1156 it caused the error "internal error: overran compiling workspace". The
1157 maximum number of forward references (including repeats) was limited by the
1158 internal workspace, and dependent on the LINK_SIZE. The code has been
1159 rewritten so that the workspace expands (via pcre_malloc) if necessary, and
1160 the default depends on LINK_SIZE. There is a new upper limit (for safety)
1161 of around 200,000 forward references. While doing this, I also speeded up
1162 the filling in of repeated forward references.
1163
1164 16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
1165 incorrectly expecting the subject to contain another "a" after the start.
1166
1167 17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
1168 in the match, the SKIP should be ignored. This was not happening; instead
1169 the SKIP was being treated as NOMATCH. For patterns such as
1170 /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
1171 tested.
1172
1173 18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
1174 now much more compatible with Perl, in particular in cases where the result
1175 is a non-match for a non-anchored pattern. For example, if
1176 /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
1177 "m", where previously it did not return a name. A side effect of this
1178 change is that for partial matches, the last encountered mark name is
1179 returned, as for non matches. A number of tests that were previously not
1180 Perl-compatible have been moved into the Perl-compatible test files. The
1181 refactoring has had the pleasing side effect of removing one argument from
1182 the match() function, thus reducing its stack requirements.
1183
1184 19. If the /S+ option was used in pcretest to study a pattern using JIT,
1185 subsequent uses of /S (without +) incorrectly behaved like /S+.
1186
1187 21. Retrieve executable code size support for the JIT compiler and fixing
1188 some warnings.
1189
1190 22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
1191 not work when the shorter character appeared right at the end of the
1192 subject string.
1193
1194 23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
1195 systems.
1196
1197 24. Added PCRE_INFO_JITSIZE to pass on the value from (21) above, and also
1198 output it when the /M option is used in pcretest.
1199
1200 25. The CheckMan script was not being included in the distribution. Also, added
1201 an explicit "perl" to run Perl scripts from the PrepareRelease script
1202 because this is reportedly needed in Windows.
1203
1204 26. If study data was being save in a file and studying had not found a set of
1205 "starts with" bytes for the pattern, the data written to the file (though
1206 never used) was taken from uninitialized memory and so caused valgrind to
1207 complain.
1208
1209 27. Updated RunTest.bat as provided by Sheri Pierce.
1210
1211 28. Fixed a possible uninitialized memory bug in pcre_jit_compile.c.
1212
1213 29. Computation of memory usage for the table of capturing group names was
1214 giving an unnecessarily large value.
1215
1216
1217 Version 8.20 21-Oct-2011
1218 ------------------------
1219
1220 1. Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
1221 a POSIX class. After further experiments with Perl, which convinced me that
1222 Perl has bugs and confusions, a closing square bracket is no longer allowed
1223 in a POSIX name. This bug also affected patterns with classes that started
1224 with full stops.
1225
1226 2. If a pattern such as /(a)b|ac/ is matched against "ac", there is no
1227 captured substring, but while checking the failing first alternative,
1228 substring 1 is temporarily captured. If the output vector supplied to
1229 pcre_exec() was not big enough for this capture, the yield of the function
1230 was still zero ("insufficient space for captured substrings"). This cannot
1231 be totally fixed without adding another stack variable, which seems a lot
1232 of expense for a edge case. However, I have improved the situation in cases
1233 such as /(a)(b)x|abc/ matched against "abc", where the return code
1234 indicates that fewer than the maximum number of slots in the ovector have
1235 been set.
1236
1237 3. Related to (2) above: when there are more back references in a pattern than
1238 slots in the output vector, pcre_exec() uses temporary memory during
1239 matching, and copies in the captures as far as possible afterwards. It was
1240 using the entire output vector, but this conflicts with the specification
1241 that only 2/3 is used for passing back captured substrings. Now it uses
1242 only the first 2/3, for compatibility. This is, of course, another edge
1243 case.
1244
1245 4. Zoltan Herczeg's just-in-time compiler support has been integrated into the
1246 main code base, and can be used by building with --enable-jit. When this is
1247 done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
1248 runtime --no-jit option is given.
1249
1250 5. When the number of matches in a pcre_dfa_exec() run exactly filled the
1251 ovector, the return from the function was zero, implying that there were
1252 other matches that did not fit. The correct "exactly full" value is now
1253 returned.
1254
1255 6. If a subpattern that was called recursively or as a subroutine contained
1256 (*PRUNE) or any other control that caused it to give a non-standard return,
1257 invalid errors such as "Error -26 (nested recursion at the same subject
1258 position)" or even infinite loops could occur.
1259
1260 7. If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
1261 computing the minimum length on reaching *ACCEPT, and so ended up with the
1262 wrong value of 1 rather than 0. Further investigation indicates that
1263 computing a minimum subject length in the presence of *ACCEPT is difficult
1264 (think back references, subroutine calls), and so I have changed the code
1265 so that no minimum is registered for a pattern that contains *ACCEPT.
1266
1267 8. If (*THEN) was present in the first (true) branch of a conditional group,
1268 it was not handled as intended. [But see 16 below.]
1269
1270 9. Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
1271 Sheri Pierce.
1272
1273 10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
1274 the first byte in a match must be "a".
1275
1276 11. Change 17 for 8.13 increased the recursion depth for patterns like
1277 /a(?:.)*?a/ drastically. I've improved things by remembering whether a
1278 pattern contains any instances of (*THEN). If it does not, the old
1279 optimizations are restored. It would be nice to do this on a per-group
1280 basis, but at the moment that is not feasible.
1281
1282 12. In some environments, the output of pcretest -C is CRLF terminated. This
1283 broke RunTest's code that checks for the link size. A single white space
1284 character after the value is now allowed for.
1285
1286 13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
1287 For "fr", it uses the Windows-specific input and output files.
1288
1289 14. If (*THEN) appeared in a group that was called recursively or as a
1290 subroutine, it did not work as intended. [But see next item.]
1291
1292 15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
1293 pattern fragments (but not containing any | characters). If A and B are
1294 matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
1295 was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
1296 D. In other words, Perl considers parentheses that do not contain any |
1297 characters to be part of a surrounding alternative, whereas PCRE was
1298 treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
1299 differently. PCRE now behaves in the same way as Perl, except in the case
1300 of subroutine/recursion calls such as (?1) which have in any case always
1301 been different (but PCRE had them first :-).
1302
1303 16. Related to 15 above: Perl does not treat the | in a conditional group as
1304 creating alternatives. Such a group is treated in the same way as an
1305 ordinary group without any | characters when processing (*THEN). PCRE has
1306 been changed to match Perl's behaviour.
1307
1308 17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
1309 RunGrepTest script failed.
1310
1311 18. Change 22 for version 13 caused atomic groups to use more stack. This is
1312 inevitable for groups that contain captures, but it can lead to a lot of
1313 stack use in large patterns. The old behaviour has been restored for atomic
1314 groups that do not contain any capturing parentheses.
1315
1316 19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
1317 suppress the check for a minimum subject length at run time. (If it was
1318 given to pcre_exec() or pcre_dfa_exec() it did work.)
1319
1320 20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
1321 fail to work when decoding hex characters in data strings in EBCDIC
1322 environments.
1323
1324 21. It appears that in at least one Mac OS environment, the isxdigit() function
1325 is implemented as a macro that evaluates to its argument more than once,
1326 contravening the C 90 Standard (I haven't checked a later standard). There
1327 was an instance in pcretest which caused it to go wrong when processing
1328 \x{...} escapes in subject strings. The has been rewritten to avoid using
1329 things like p++ in the argument of isxdigit().
1330
1331
1332 Version 8.13 16-Aug-2011
1333 ------------------------
1334
1335 1. The Unicode data tables have been updated to Unicode 6.0.0.
1336
1337 2. Two minor typos in pcre_internal.h have been fixed.
1338
1339 3. Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
1340 pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
1341 in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
1342
1343 4. There were a number of related bugs in the code for matching backrefences
1344 caselessly in UTF-8 mode when codes for the characters concerned were
1345 different numbers of bytes. For example, U+023A and U+2C65 are an upper
1346 and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
1347 (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
1348 code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
1349 2-byte code at the end of the subject (it thought there wasn't enough data
1350 left).
1351
1352 5. Comprehensive information about what went wrong is now returned by
1353 pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
1354 as the output vector has at least 2 elements. The offset of the start of
1355 the failing character and a reason code are placed in the vector.
1356
1357 6. When the UTF-8 string check fails for pcre_compile(), the offset that is
1358 now returned is for the first byte of the failing character, instead of the
1359 last byte inspected. This is an incompatible change, but I hope it is small
1360 enough not to be a problem. It makes the returned offset consistent with
1361 pcre_exec() and pcre_dfa_exec().
1362
1363 7. pcretest now gives a text phrase as well as the error number when
1364 pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
1365 failure, the offset and reason code are output.
1366
1367 8. When \R was used with a maximizing quantifier it failed to skip backwards
1368 over a \r\n pair if the subsequent match failed. Instead, it just skipped
1369 back over a single character (\n). This seems wrong (because it treated the
1370 two characters as a single entity when going forwards), conflicts with the
1371 documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
1372 behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
1373 has been changed.
1374
1375 9. Some internal refactoring has changed the processing so that the handling
1376 of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
1377 time (the PCRE_DOTALL option was changed this way some time ago: version
1378 7.7 change 16). This has made it possible to abolish the OP_OPT op code,
1379 which was always a bit of a fudge. It also means that there is one less
1380 argument for the match() function, which reduces its stack requirements
1381 slightly. This change also fixes an incompatibility with Perl: the pattern
1382 (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
1383
1384 10. More internal refactoring has drastically reduced the number of recursive
1385 calls to match() for possessively repeated groups such as (abc)++ when
1386 using pcre_exec().
1387
1388 11. While implementing 10, a number of bugs in the handling of groups were
1389 discovered and fixed:
1390
1391 (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
1392 (a|)*(?1) gave a compile-time internal error.
1393 ((a|)+)+ did not notice that the outer group could match an empty string.
1394 (^a|^)+ was not marked as anchored.
1395 (.*a|.*)+ was not marked as matching at start or after a newline.
1396
1397 12. Yet more internal refactoring has removed another argument from the match()
1398 function. Special calls to this function are now indicated by setting a
1399 value in a variable in the "match data" data block.
1400
1401 13. Be more explicit in pcre_study() instead of relying on "default" for
1402 opcodes that mean there is no starting character; this means that when new
1403 ones are added and accidentally left out of pcre_study(), testing should
1404 pick them up.
1405
1406 14. The -s option of pcretest has been documented for ages as being an old
1407 synonym of -m (show memory usage). I have changed it to mean "force study
1408 for every regex", that is, assume /S for every regex. This is similar to -i
1409 and -d etc. It's slightly incompatible, but I'm hoping nobody is still
1410 using it. It makes it easier to run collections of tests with and without
1411 study enabled, and thereby test pcre_study() more easily. All the standard
1412 tests are now run with and without -s (but some patterns can be marked as
1413 "never study" - see 20 below).
1414
1415 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
1416 restoration of the capturing data to the outer values was not happening
1417 correctly.
1418
1419 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
1420 empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
1421 pattern had matched an empty string, and so incorrectly returned a no
1422 match.
1423
1424 17. There was optimizing code for the last branch of non-capturing parentheses,
1425 and also for the obeyed branch of a conditional subexpression, which used
1426 tail recursion to cut down on stack usage. Unfortunately, now that there is
1427 the possibility of (*THEN) occurring in these branches, tail recursion is
1428 no longer possible because the return has to be checked for (*THEN). These
1429 two optimizations have therefore been removed. [But see 8.20/11 above.]
1430
1431 18. If a pattern containing \R was studied, it was assumed that \R always
1432 matched two bytes, thus causing the minimum subject length to be
1433 incorrectly computed because \R can also match just one byte.
1434
1435 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
1436 was incorrectly computed.
1437
1438 20. If /S is present twice on a test pattern in pcretest input, it now
1439 *disables* studying, thereby overriding the use of -s on the command line
1440 (see 14 above). This is necessary for one or two tests to keep the output
1441 identical in both cases.
1442
1443 21. When (*ACCEPT) was used in an assertion that matched an empty string and
1444 PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
1445
1446 22. When an atomic group that contained a capturing parenthesis was
1447 successfully matched, but the branch in which it appeared failed, the
1448 capturing was not being forgotten if a higher numbered group was later
1449 captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
1450 group 1 to "a", when in fact it should be unset. This applied to multi-
1451 branched capturing and non-capturing groups, repeated or not, and also to
1452 positive assertions (capturing in negative assertions does not happen
1453 in PCRE) and also to nested atomic groups.
1454
1455 23. Add the ++ qualifier feature to pcretest, to show the remainder of the
1456 subject after a captured substring, to make it easier to tell which of a
1457 number of identical substrings has been captured.
1458
1459 24. The way atomic groups are processed by pcre_exec() has been changed so that
1460 if they are repeated, backtracking one repetition now resets captured
1461 values correctly. For example, if ((?>(a+)b)+aabab) is matched against
1462 "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
1463 "aaa". Previously, it would have been "a". As part of this code
1464 refactoring, the way recursive calls are handled has also been changed.
1465
1466 25. If an assertion condition captured any substrings, they were not passed
1467 back unless some other capturing happened later. For example, if
1468 (?(?=(a))a) was matched against "a", no capturing was returned.
1469
1470 26. When studying a pattern that contained subroutine calls or assertions,
1471 the code for finding the minimum length of a possible match was handling
1472 direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
1473 group 1 called group 2 while simultaneously a separate group 2 called group
1474 1). A stack overflow occurred in this case. I have fixed this by limiting
1475 the recursion depth to 10.
1476
1477 27. Updated RunTest.bat in the distribution to the version supplied by Tom
1478 Fortmann. This supports explicit test numbers on the command line, and has
1479 argument validation and error reporting.
1480
1481 28. An instance of \X with an unlimited repeat could fail if at any point the
1482 first character it looked at was a mark character.
1483
1484 29. Some minor code refactoring concerning Unicode properties and scripts
1485 should reduce the stack requirement of match() slightly.
1486
1487 30. Added the '=' option to pcretest to check the setting of unused capturing
1488 slots at the end of the pattern, which are documented as being -1, but are
1489 not included in the return count.
1490
1491 31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
1492 compiled something random. Now it gives a compile-time error (as does
1493 Perl).
1494
1495 32. A *MARK encountered during the processing of a positive assertion is now
1496 recorded and passed back (compatible with Perl).
1497
1498 33. If --only-matching or --colour was set on a pcregrep call whose pattern
1499 had alternative anchored branches, the search for a second match in a line
1500 was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
1501 matched the line "0102" twice. The same bug affected patterns that started
1502 with a backwards assertion. For example /\b01|\b02/ also matched "0102"
1503 twice.
1504
1505 34. Previously, PCRE did not allow quantification of assertions. However, Perl
1506 does, and because of capturing effects, quantifying parenthesized
1507 assertions may at times be useful. Quantifiers are now allowed for
1508 parenthesized assertions.
1509
1510 35. A minor code tidy in pcre_compile() when checking options for \R usage.
1511
1512 36. \g was being checked for fancy things in a character class, when it should
1513 just be a literal "g".
1514
1515 37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
1516 appearance of a nested POSIX class supersedes an apparent external class.
1517 For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
1518 unescaped square brackets may also appear as part of class names. For
1519 example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
1520 more like Perl. (But see 8.20/1 above.)
1521
1522 38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
1523 was because it thought it was \N{name}, which is not supported).
1524
1525 39. Add minix to OS list not supporting the -S option in pcretest.
1526
1527 40. PCRE tries to detect cases of infinite recursion at compile time, but it
1528 cannot analyze patterns in sufficient detail to catch mutual recursions
1529 such as ((?1))((?2)). There is now a runtime test that gives an error if a
1530 subgroup is called recursively as a subpattern for a second time at the
1531 same position in the subject string. In previous releases this might have
1532 been caught by the recursion limit, or it might have run out of stack.
1533
1534 41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
1535 happen only once. PCRE was, however incorrectly giving a compile time error
1536 "recursive call could loop indefinitely" because it cannot analyze the
1537 pattern in sufficient detail. The compile time test no longer happens when
1538 PCRE is compiling a conditional subpattern, but actual runaway loops are
1539 now caught at runtime (see 40 above).
1540
1541 42. It seems that Perl allows any characters other than a closing parenthesis
1542 to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
1543 has been changed to be the same.
1544
1545 43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
1546 as not to get warnings when autogen.sh is called. Also changed
1547 AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
1548
1549 44. To help people who use pcregrep to scan files containing exceedingly long
1550 lines, the following changes have been made:
1551
1552 (a) The default value of the buffer size parameter has been increased from
1553 8K to 20K. (The actual buffer used is three times this size.)
1554
1555 (b) The default can be changed by ./configure --with-pcregrep-bufsize when
1556 PCRE is built.
1557
1558 (c) A --buffer-size=n option has been added to pcregrep, to allow the size
1559 to be set at run time.
1560
1561 (d) Numerical values in pcregrep options can be followed by K or M, for
1562 example --buffer-size=50K.
1563
1564 (e) If a line being scanned overflows pcregrep's buffer, an error is now
1565 given and the return code is set to 2.
1566
1567 45. Add a pointer to the latest mark to the callout data block.
1568
1569 46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
1570 partial match of an empty string instead of no match. This was specific to
1571 the use of ".".
1572
1573 47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
1574 complete match instead of a partial match. This bug was dependent on both
1575 the PCRE_UTF8 and PCRE_DOTALL options being set.
1576
1577 48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
1578 starting byte set, because \b was not being ignored.
1579
1580
1581 Version 8.12 15-Jan-2011
1582 ------------------------
1583
1584 1. Fixed some typos in the markup of the man pages, and wrote a script that
1585 checks for such things as part of the documentation building process.
1586
1587 2. On a big-endian 64-bit system, pcregrep did not correctly process the
1588 --match-limit and --recursion-limit options (added for 8.11). In
1589 particular, this made one of the standard tests fail. (The integer value
1590 went into the wrong half of a long int.)
1591
1592 3. If the --colour option was given to pcregrep with -v (invert match), it
1593 did strange things, either producing crazy output, or crashing. It should,
1594 of course, ignore a request for colour when reporting lines that do not
1595 match.
1596
1597 4. Another pcregrep bug caused similar problems if --colour was specified with
1598 -M (multiline) and the pattern match finished with a line ending.
1599
1600 5. In pcregrep, when a pattern that ended with a literal newline sequence was
1601 matched in multiline mode, the following line was shown as part of the
1602 match. This seems wrong, so I have changed it.
1603
1604 6. Another pcregrep bug in multiline mode, when --colour was specified, caused
1605 the check for further matches in the same line (so they could be coloured)
1606 to overrun the end of the current line. If another match was found, it was
1607 incorrectly shown (and then shown again when found in the next line).
1608
1609 7. If pcregrep was compiled under Windows, there was a reference to the
1610 function pcregrep_exit() before it was defined. I am assuming this was
1611 the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
1612 reported by a user. I've moved the definition above the reference.
1613
1614
1615 Version 8.11 10-Dec-2010
1616 ------------------------
1617
1618 1. (*THEN) was not working properly if there were untried alternatives prior
1619 to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
1620 backtracked to try for "b" instead of moving to the next alternative branch
1621 at the same level (in this case, to look for "c"). The Perl documentation
1622 is clear that when (*THEN) is backtracked onto, it goes to the "next
1623 alternative in the innermost enclosing group".
1624
1625 2. (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
1626 such as (A(*COMMIT)B(*THEN)C|D) any failure after matching A should
1627 result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
1628 (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
1629 (*THEN).
1630
1631 3. If \s appeared in a character class, it removed the VT character from
1632 the class, even if it had been included by some previous item, for example
1633 in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
1634 of \s, but is part of the POSIX "space" class.)
1635
1636 4. A partial match never returns an empty string (because you can always
1637 match an empty string at the end of the subject); however the checking for
1638 an empty string was starting at the "start of match" point. This has been
1639 changed to the "earliest inspected character" point, because the returned
1640 data for a partial match starts at this character. This means that, for
1641 example, /(?<=abc)def/ gives a partial match for the subject "abc"
1642 (previously it gave "no match").
1643
1644 5. Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
1645 of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
1646 previously a full match would be given. However, setting PCRE_PARTIAL_HARD
1647 has an implication that the given string is incomplete (because a partial
1648 match is preferred over a full match). For this reason, these items now
1649 give a partial match in this situation. [Aside: previously, the one case
1650 /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
1651 match rather than a full match, which was wrong by the old rules, but is
1652 now correct.]
1653
1654 6. There was a bug in the handling of #-introduced comments, recognized when
1655 PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
1656 If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
1657 UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
1658 scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
1659 but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
1660 places in pcre_compile().
1661
1662 7. Related to (6) above, when pcre_compile() was skipping #-introduced
1663 comments when looking ahead for named forward references to subpatterns,
1664 the only newline sequence it recognized was NL. It now handles newlines
1665 according to the set newline convention.
1666
1667 8. SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
1668 former, but used strtoul(), whereas pcretest avoided strtoul() but did not
1669 cater for a lack of strerror(). These oversights have been fixed.
1670
1671 9. Added --match-limit and --recursion-limit to pcregrep.
1672
1673 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
1674
1675 11. When the -o option was used, pcregrep was setting a return code of 1, even
1676 when matches were found, and --line-buffered was not being honoured.
1677
1678 12. Added an optional parentheses number to the -o and --only-matching options
1679 of pcregrep.
1680
1681 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
1682 can match an empty string. The code to do it in pcretest and pcredemo
1683 needed fixing:
1684
1685 (a) When the newline convention was "crlf", pcretest got it wrong, skipping
1686 only one byte after an empty string match just before CRLF (this case
1687 just got forgotten; "any" and "anycrlf" were OK).
1688
1689 (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
1690 mode when an empty string match preceded an ASCII character followed by
1691 a non-ASCII character. (The code for advancing by one character rather
1692 than one byte was nonsense.)
1693
1694 (c) The pcredemo.c sample program did not have any code at all to handle
1695 the cases when CRLF is a valid newline sequence.
1696
1697 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
1698 as a starting offset was within the subject string. There is now a new
1699 error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
1700 negative or greater than the length of the string. In order to test this,
1701 pcretest is extended to allow the setting of negative starting offsets.
1702
1703 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
1704 starting offset points to the beginning of a UTF-8 character was
1705 unnecessarily clumsy. I tidied it up.
1706
1707 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
1708 bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
1709
1710 17. Nobody had reported that the --include_dir option, which was added in
1711 release 7.7 should have been called --include-dir (hyphen, not underscore)
1712 for compatibility with GNU grep. I have changed it to --include-dir, but
1713 left --include_dir as an undocumented synonym, and the same for
1714 --exclude-dir, though that is not available in GNU grep, at least as of
1715 release 2.5.4.
1716
1717 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
1718 characters from a string of bytes) have been redefined so as not to use
1719 loops, in order to improve performance in some environments. At the same
1720 time, I abstracted some of the common code into auxiliary macros to save
1721 repetition (this should not affect the compiled code).
1722
1723 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
1724 compile-time error is now given if \c is not followed by an ASCII
1725 character, that is, a byte less than 128. (In EBCDIC mode, the code is
1726 different, and any byte value is allowed.)
1727
1728 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
1729 START_OPTIMIZE option, which is now allowed at compile time - but just
1730 passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
1731 to pcregrep and other applications that have no direct access to PCRE
1732 options. The new /Y option in pcretest sets this option when calling
1733 pcre_compile().
1734
1735 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
1736 back references. Groups containing recursive back references were forced to
1737 be atomic by that change, but in the case of named groups, the amount of
1738 memory required was incorrectly computed, leading to "Failed: internal
1739 error: code overflow". This has been fixed.
1740
1741 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
1742 pcretest.c, to avoid build problems in some Borland environments.
1743
1744
1745 Version 8.10 25-Jun-2010
1746 ------------------------
1747
1748 1. Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
1749 THEN.
1750
1751 2. (*ACCEPT) was not working when inside an atomic group.
1752
1753 3. Inside a character class, \B is treated as a literal by default, but
1754 faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
1755 causes the error). The code is unchanged, but I tidied the documentation.
1756
1757 4. Inside a character class, PCRE always treated \R and \X as literals,
1758 whereas Perl faults them if its -w option is set. I have changed PCRE so
1759 that it faults them when PCRE_EXTRA is set.
1760
1761 5. Added support for \N, which always matches any character other than
1762 newline. (It is the same as "." when PCRE_DOTALL is not set.)
1763
1764 6. When compiling pcregrep with newer versions of gcc which may have
1765 FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
1766 declared with attribute warn_unused_result" were given. Just casting the
1767 result to (void) does not stop the warnings; a more elaborate fudge is
1768 needed. I've used a macro to implement this.
1769
1770 7. Minor change to pcretest.c to avoid a compiler warning.
1771
1772 8. Added four artifical Unicode properties to help with an option to make
1773 \s etc use properties (see next item). The new properties are: Xan
1774 (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
1775
1776 9. Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
1777 use Unicode properties. (*UCP) at the start of a pattern can be used to set
1778 this option. Modified pcretest to add /W to test this facility. Added
1779 REG_UCP to make it available via the POSIX interface.
1780
1781 10. Added --line-buffered to pcregrep.
1782
1783 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
1784 studied, and the match started with a letter with a code point greater than
1785 127 whose first byte was different to the first byte of the other case of
1786 the letter, the other case of this starting letter was not recognized
1787 (#976).
1788
1789 12. If a pattern that was studied started with a repeated Unicode property
1790 test, for example, \p{Nd}+, there was the theoretical possibility of
1791 setting up an incorrect bitmap of starting bytes, but fortunately it could
1792 not have actually happened in practice until change 8 above was made (it
1793 added property types that matched character-matching opcodes).
1794
1795 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
1796 possible starting bytes for non-anchored patterns.
1797
1798 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
1799 \R, and also a number of cases that involve Unicode properties, both
1800 explicit and implicit when PCRE_UCP is set.
1801
1802 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
1803 input, it could crash or give wrong results if characters with values
1804 greater than 0xc0 were present in the subject string. (Detail: it assumed
1805 UTF-8 input when processing these items.)
1806
1807 16. Added a lot of (int) casts to avoid compiler warnings in systems where
1808 size_t is 64-bit (#991).
1809
1810 17. Added a check for running out of memory when PCRE is compiled with
1811 --disable-stack-for-recursion (#990).
1812
1813 18. If the last data line in a file for pcretest does not have a newline on
1814 the end, a newline was missing in the output.
1815
1816 19. The default pcre_chartables.c file recognizes only ASCII characters (values
1817 less than 128) in its various bitmaps. However, there is a facility for
1818 generating tables according to the current locale when PCRE is compiled. It
1819 turns out that in some environments, 0x85 and 0xa0, which are Unicode space
1820 characters, are recognized by isspace() and therefore were getting set in
1821 these tables, and indeed these tables seem to approximate to ISO 8859. This
1822 caused a problem in UTF-8 mode when pcre_study() was used to create a list
1823 of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
1824 which of course cannot start UTF-8 characters. I have changed the code so
1825 that only real ASCII characters (less than 128) and the correct starting
1826 bytes for UTF-8 encodings are set for characters greater than 127 when in
1827 UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
1828 altogether.)
1829
1830 20. Added the /T option to pcretest so as to be able to run tests with non-
1831 standard character tables, thus making it possible to include the tests
1832 used for 19 above in the standard set of tests.
1833
1834 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
1835 reference to a subpattern the other side of a comment that contains an
1836 opening parenthesis caused either an internal compiling error, or a
1837 reference to the wrong subpattern.
1838
1839
1840 Version 8.02 19-Mar-2010
1841 ------------------------
1842
1843 1. The Unicode data tables have been updated to Unicode 5.2.0.
1844
1845 2. Added the option --libs-cpp to pcre-config, but only when C++ support is
1846 configured.
1847
1848 3. Updated the licensing terms in the pcregexp.pas file, as agreed with the
1849 original author of that file, following a query about its status.
1850
1851 4. On systems that do not have stdint.h (e.g. Solaris), check for and include
1852 inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
1853
1854 5. A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
1855 quantifier applied to a forward-referencing subroutine call, could compile
1856 incorrect code or give the error "internal error: previously-checked
1857 referenced subpattern not found".
1858
1859 6. Both MS Visual Studio and Symbian OS have problems with initializing
1860 variables to point to external functions. For these systems, therefore,
1861 pcre_malloc etc. are now initialized to local functions that call the
1862 relevant global functions.
1863
1864 7. There were two entries missing in the vectors called coptable and poptable
1865 in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
1866 I've fixed the data, and added a kludgy way of testing at compile time that
1867 the lengths are correct (equal to the number of opcodes).
1868
1869 8. Following on from 7, I added a similar kludge to check the length of the
1870 eint vector in pcreposix.c.
1871
1872 9. Error texts for pcre_compile() are held as one long string to avoid too
1873 much relocation at load time. To find a text, the string is searched,
1874 counting zeros. There was no check for running off the end of the string,
1875 which could happen if a new error number was added without updating the
1876 string.
1877
1878 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
1879
1880 11. \K was not working if it appeared in an atomic group or in a group that
1881 was called as a "subroutine", or in an assertion. Perl 5.11 documents that
1882 \K is "not well defined" if used in an assertion. PCRE now accepts it if
1883 the assertion is positive, but not if it is negative.
1884
1885 12. Change 11 fortuitously reduced the size of the stack frame used in the
1886 "match()" function of pcre_exec.c by one pointer. Forthcoming
1887 implementation of support for (*MARK) will need an extra pointer on the
1888 stack; I have reserved it now, so that the stack frame size does not
1889 decrease.
1890
1891 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
1892 item in branch that calls a recursion is a subroutine call - as in the
1893 second branch in the above example - was incorrectly given the compile-
1894 time error "recursive call could loop indefinitely" because pcre_compile()
1895 was not correctly checking the subroutine for matching a non-empty string.
1896
1897 14. The checks for overrunning compiling workspace could trigger after an
1898 overrun had occurred. This is a "should never occur" error, but it can be
1899 triggered by pathological patterns such as hundreds of nested parentheses.
1900 The checks now trigger 100 bytes before the end of the workspace.
1901
1902 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
1903
1904
1905 Version 8.01 19-Jan-2010
1906 ------------------------
1907
1908 1. If a pattern contained a conditional subpattern with only one branch (in
1909 particular, this includes all (*DEFINE) patterns), a call to pcre_study()
1910 computed the wrong minimum data length (which is of course zero for such
1911 subpatterns). This could cause incorrect "no match" results.
1912
1913 2. For patterns such as (?i)a(?-i)b|c where an option setting at the start of
1914 the pattern is reset in the first branch, pcre_compile() failed with
1915 "internal error: code overflow at offset...". This happened only when
1916 the reset was to the original external option setting. (An optimization
1917 abstracts leading options settings into an external setting, which was the
1918 cause of this.)
1919
1920 3. A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
1921 of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
1922 assertion pattern did not match (meaning that the assertion was true), it
1923 was incorrectly treated as false if the SKIP had been reached during the
1924 matching. This also applied to assertions used as conditions.
1925
1926 4. If an item that is not supported by pcre_dfa_exec() was encountered in an
1927 assertion subpattern, including such a pattern used as a condition,
1928 unpredictable results occurred, instead of the error return
1929 PCRE_ERROR_DFA_UITEM.
1930
1931 5. The C++ GlobalReplace function was not working like Perl for the special
1932 situation when an empty string is matched. It now does the fancy magic
1933 stuff that is necessary.
1934
1935 6. In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
1936 removed. (These were left over from very, very early versions of PCRE.)
1937
1938 7. Some cosmetic changes to the code to make life easier when compiling it
1939 as part of something else:
1940
1941 (a) Change DEBUG to PCRE_DEBUG.
1942
1943 (b) In pcre_compile(), rename the member of the "branch_chain" structure
1944 called "current" as "current_branch", to prevent a collision with the
1945 Linux macro when compiled as a kernel module.
1946
1947 (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
1948 prevent a collision with the Linux macro when compiled as a kernel
1949 module.
1950
1951 8. In pcre_compile() there are some checks for integer overflows that used to
1952 cast potentially large values to (double). This has been changed to that
1953 when building, a check for int64_t is made, and if it is found, it is used
1954 instead, thus avoiding the use of floating point arithmetic. (There is no
1955 other use of FP in PCRE.) If int64_t is not found, the fallback is to
1956 double.
1957
1958 9. Added two casts to avoid signed/unsigned warnings from VS Studio Express
1959 2005 (difference between two addresses compared to an unsigned value).
1960
1961 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
1962 custom one, because of the following reported problem in Windows:
1963
1964 - libbz2 uses the Pascal calling convention (WINAPI) for the functions
1965 under Win32.
1966 - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
1967 therefore missing the function definition.
1968 - The compiler thus generates a "C" signature for the test function.
1969 - The linker fails to find the "C" function.
1970 - PCRE fails to configure if asked to do so against libbz2.
1971
1972 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
1973 messages were output:
1974
1975 Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
1976 rerunning libtoolize, to keep the correct libtool macros in-tree.
1977 Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
1978
1979 I have done both of these things.
1980
1981 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
1982 most of the time, it *can* run out if it is given a pattern that contains a
1983 runaway infinite recursion. I updated the discussion in the pcrestack man
1984 page.
1985
1986 13. Now that we have gone to the x.xx style of version numbers, the minor
1987 version may start with zero. Using 08 or 09 is a bad idea because users
1988 might check the value of PCRE_MINOR in their code, and 08 or 09 may be
1989 interpreted as invalid octal numbers. I've updated the previous comment in
1990 configure.ac, and also added a check that gives an error if 08 or 09 are
1991 used.
1992
1993 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
1994 causing partial matching to fail when the end of the subject matched \W
1995 in a UTF-8 pattern where \W was quantified with a minimum of 3.
1996
1997 15. There were some discrepancies between the declarations in pcre_internal.h
1998 of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
1999 their definitions. The declarations used "const uschar *" and the
2000 definitions used USPTR. Even though USPTR is normally defined as "const
2001 unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
2002 reported that: "This difference in casting confuses some C++ compilers, for
2003 example, SunCC recognizes above declarations as different functions and
2004 generates broken code for hbpcre." I have changed the declarations to use
2005 USPTR.
2006
2007 16. GNU libtool is named differently on some systems. The autogen.sh script now
2008 tries several variants such as glibtoolize (MacOSX) and libtoolize1x
2009 (FreeBSD).
2010
2011 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
2012 (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
2013 comment: "Figure out how to create a longlong from a string: strtoll and
2014 equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
2015 instance, but it only takes 2 args instead of 3!"
2016
2017 18. A subtle bug concerned with back references has been fixed by a change of
2018 specification, with a corresponding code fix. A pattern such as
2019 ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
2020 refers, was giving matches when it shouldn't. For example, xa=xaaa would
2021 match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
2022 same bug. Such groups have to be quantified to be useful, or contained
2023 inside another quantified group. (If there's no repetition, the reference
2024 can never match.) The problem arises because, having left the group and
2025 moved on to the rest of the pattern, a later failure that backtracks into
2026 the group uses the captured value from the final iteration of the group
2027 rather than the correct earlier one. I have fixed this in PCRE by forcing
2028 any group that contains a reference to itself to be an atomic group; that
2029 is, there cannot be any backtracking into it once it has completed. This is
2030 similar to recursive and subroutine calls.
2031
2032
2033 Version 8.00 19-Oct-09
2034 ----------------------
2035
2036 1. The table for translating pcre_compile() error codes into POSIX error codes
2037 was out-of-date, and there was no check on the pcre_compile() error code
2038 being within the table. This could lead to an OK return being given in
2039 error.
2040
2041 2. Changed the call to open a subject file in pcregrep from fopen(pathname,
2042 "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
2043 in a Windows environment.
2044
2045 3. The pcregrep --count option prints the count for each file even when it is
2046 zero, as does GNU grep. However, pcregrep was also printing all files when
2047 --files-with-matches was added. Now, when both options are given, it prints
2048 counts only for those files that have at least one match. (GNU grep just
2049 prints the file name in this circumstance, but including the count seems
2050 more useful - otherwise, why use --count?) Also ensured that the
2051 combination -clh just lists non-zero counts, with no names.
2052
2053 4. The long form of the pcregrep -F option was incorrectly implemented as
2054 --fixed_strings instead of --fixed-strings. This is an incompatible change,
2055 but it seems right to fix it, and I didn't think it was worth preserving
2056 the old behaviour.
2057
2058 5. The command line items --regex=pattern and --regexp=pattern were not
2059 recognized by pcregrep, which required --regex pattern or --regexp pattern
2060 (with a space rather than an '='). The man page documented the '=' forms,
2061 which are compatible with GNU grep; these now work.
2062
2063 6. No libpcreposix.pc file was created for pkg-config; there was just
2064 libpcre.pc and libpcrecpp.pc. The omission has been rectified.
2065
2066 7. Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
2067 when UCP support is not needed, by modifying the Python script that
2068 generates it from Unicode data files. This should not matter if the module
2069 is correctly used as a library, but I received one complaint about 50K of
2070 unwanted data. My guess is that the person linked everything into his
2071 program rather than using a library. Anyway, it does no harm.
2072
2073 8. A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
2074 was a minimum greater than 1 for a wide character in a possessive
2075 repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
2076 which had an unlimited repeat of a nested, fixed maximum repeat of a wide
2077 character. Chaos in the form of incorrect output or a compiling loop could
2078 result.
2079
2080 9. The restrictions on what a pattern can contain when partial matching is
2081 requested for pcre_exec() have been removed. All patterns can now be
2082 partially matched by this function. In addition, if there are at least two
2083 slots in the offset vector, the offset of the earliest inspected character
2084 for the match and the offset of the end of the subject are set in them when
2085 PCRE_ERROR_PARTIAL is returned.
2086
2087 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
2088 synonymous with PCRE_PARTIAL, for backwards compatibility, and
2089 PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
2090 and may be more useful for multi-segment matching.
2091
2092 11. Partial matching with pcre_exec() is now more intuitive. A partial match
2093 used to be given if ever the end of the subject was reached; now it is
2094 given only if matching could not proceed because another character was
2095 needed. This makes a difference in some odd cases such as Z(*FAIL) with the
2096 string "Z", which now yields "no match" instead of "partial match". In the
2097 case of pcre_dfa_exec(), "no match" is given if every matching path for the
2098 final character ended with (*FAIL).
2099
2100 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
2101 if the pattern had a "must contain" character that was already found in the
2102 earlier partial match, unless partial matching was again requested. For
2103 example, with the pattern /dog.(body)?/, the "must contain" character is
2104 "g". If the first part-match was for the string "dog", restarting with
2105 "sbody" failed. This bug has been fixed.
2106
2107 13. The string returned by pcre_dfa_exec() after a partial match has been
2108 changed so that it starts at the first inspected character rather than the
2109 first character of the match. This makes a difference only if the pattern
2110 starts with a lookbehind assertion or \b or \B (\K is not supported by
2111 pcre_dfa_exec()). It's an incompatible change, but it makes the two
2112 matching functions compatible, and I think it's the right thing to do.
2113
2114 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
2115 so that the demonstration program is easily available in environments where
2116 PCRE has not been installed from source.
2117
2118 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
2119 libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
2120 library.
2121
2122 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
2123 It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
2124 is not the first non-POSIX option to be added. Clearly some people find
2125 these options useful.
2126
2127 17. If a caller to the POSIX matching function regexec() passes a non-zero
2128 value for nmatch with a NULL value for pmatch, the value of
2129 nmatch is forced to zero.
2130
2131 18. RunGrepTest did not have a test for the availability of the -u option of
2132 the diff command, as RunTest does. It now checks in the same way as
2133 RunTest, and also checks for the -b option.
2134
2135 19. If an odd number of negated classes containing just a single character
2136 interposed, within parentheses, between a forward reference to a named
2137 subpattern and the definition of the subpattern, compilation crashed with
2138 an internal error, complaining that it could not find the referenced
2139 subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
2140 [The bug was that it was starting one character too far in when skipping
2141 over the character class, thus treating the ] as data rather than
2142 terminating the class. This meant it could skip too much.]
2143
2144 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
2145 /g option in pcretest when the pattern contains \K, which makes it possible
2146 to have an empty string match not at the start, even when the pattern is
2147 anchored. Updated pcretest and pcredemo to use this option.
2148
2149 21. If the maximum number of capturing subpatterns in a recursion was greater
2150 than the maximum at the outer level, the higher number was returned, but
2151 with unset values at the outer level. The correct (outer level) value is
2152 now given.
2153
2154 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
2155 PCRE did not set those parentheses (unlike Perl). I have now found a way to
2156 make it do so. The string so far is captured, making this feature
2157 compatible with Perl.
2158
2159 23. The tests have been re-organized, adding tests 11 and 12, to make it
2160 possible to check the Perl 5.10 features against Perl 5.10.
2161
2162 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
2163 pattern matches a fixed length string. PCRE did not allow this; now it
2164 does. Neither allows recursion.
2165
2166 25. I finally figured out how to implement a request to provide the minimum
2167 length of subject string that was needed in order to match a given pattern.
2168 (It was back references and recursion that I had previously got hung up
2169 on.) This code has now been added to pcre_study(); it finds a lower bound
2170 to the length of subject needed. It is not necessarily the greatest lower
2171 bound, but using it to avoid searching strings that are too short does give
2172 some useful speed-ups. The value is available to calling programs via
2173 pcre_fullinfo().
2174
2175 26. While implementing 25, I discovered to my embarrassment that pcretest had
2176 not been passing the result of pcre_study() to pcre_dfa_exec(), so the
2177 study optimizations had never been tested with that matching function.
2178 Oops. What is worse, even when it was passed study data, there was a bug in
2179 pcre_dfa_exec() that meant it never actually used it. Double oops. There
2180 were also very few tests of studied patterns with pcre_dfa_exec().
2181
2182 27. If (?| is used to create subpatterns with duplicate numbers, they are now
2183 allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
2184 on the other side of the coin, they are no longer allowed to have different
2185 names, because these cannot be distinguished in PCRE, and this has caused
2186 confusion. (This is a difference from Perl.)
2187
2188 28. When duplicate subpattern names are present (necessarily with different
2189 numbers, as required by 27 above), and a test is made by name in a
2190 conditional pattern, either for a subpattern having been matched, or for
2191 recursion in such a pattern, all the associated numbered subpatterns are
2192 tested, and the overall condition is true if the condition is true for any
2193 one of them. This is the way Perl works, and is also more like the way
2194 testing by number works.
2195
2196
2197 Version 7.9 11-Apr-09
2198 ---------------------
2199
2200 1. When building with support for bzlib/zlib (pcregrep) and/or readline
2201 (pcretest), all targets were linked against these libraries. This included
2202 libpcre, libpcreposix, and libpcrecpp, even though they do not use these
2203 libraries. This caused unwanted dependencies to be created. This problem
2204 has been fixed, and now only pcregrep is linked with bzlib/zlib and only
2205 pcretest is linked with readline.
2206
2207 2. The "typedef int BOOL" in pcre_internal.h that was included inside the
2208 "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
2209 moved outside it again, because FALSE and TRUE are already defined in AIX,
2210 but BOOL is not.
2211
2212 3. The pcre_config() function was treating the PCRE_MATCH_LIMIT and
2213 PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
2214
2215 4. The pcregrep documentation said spaces were inserted as well as colons (or
2216 hyphens) following file names and line numbers when outputting matching
2217 lines. This is not true; no spaces are inserted. I have also clarified the
2218 wording for the --colour (or --color) option.
2219
2220 5. In pcregrep, when --colour was used with -o, the list of matching strings
2221 was not coloured; this is different to GNU grep, so I have changed it to be
2222 the same.
2223
2224 6. When --colo(u)r was used in pcregrep, only the first matching substring in
2225 each matching line was coloured. Now it goes on to look for further matches
2226 of any of the test patterns, which is the same behaviour as GNU grep.
2227
2228 7. A pattern that could match an empty string could cause pcregrep to loop; it
2229 doesn't make sense to accept an empty string match in pcregrep, so I have
2230 locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
2231 seems to be how GNU grep behaves. [But see later change 40 for release
2232 8.33.]
2233
2234 8. The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
2235 start or after a newline", because the conditional assertion was not being
2236 correctly handled. The rule now is that both the assertion and what follows
2237 in the first alternative must satisfy the test.
2238
2239 9. If auto-callout was enabled in a pattern with a conditional group whose
2240 condition was an assertion, PCRE could crash during matching, both with
2241 pcre_exec() and pcre_dfa_exec().
2242
2243 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
2244 used for matching.
2245
2246 11. Unicode property support in character classes was not working for
2247 characters (bytes) greater than 127 when not in UTF-8 mode.
2248
2249 12. Added the -M command line option to pcretest.
2250
2251 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
2252
2253 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
2254
2255 16. Added comments and documentation about mis-use of no_arg in the C++
2256 wrapper.
2257
2258 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
2259 from Martin Jerabek that uses macro names for all relevant character and
2260 string constants.
2261
2262 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
2263 SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
2264 SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
2265 these, but not everybody uses configure.
2266
2267 19. A conditional group that had only one branch was not being correctly
2268 recognized as an item that could match an empty string. This meant that an
2269 enclosing group might also not be so recognized, causing infinite looping
2270 (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
2271 with the subject "ab", where knowledge that the repeated group can match
2272 nothing is needed in order to break the loop.
2273
2274 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
2275 exec(), but without supplying a callout function, matching went wrong.
2276
2277 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
2278 leak if the size of the offset vector was greater than 30. When the vector
2279 is smaller, the saved offsets during recursion go onto a local stack
2280 vector, but for larger vectors malloc() is used. It was failing to free
2281 when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
2282 error, in fact).
2283
2284 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
2285 heapframe that is used only when UTF-8 support is enabled. This caused no
2286 problem, but was untidy.
2287
2288 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
2289 CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
2290 included within another project.
2291
2292 24. Steven Van Ingelgem's patches to add more options to the CMake support,
2293 slightly modified by me:
2294
2295 (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
2296 not building pcregrep.
2297
2298 (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
2299 if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
2300
2301 25. Forward references, both numeric and by name, in patterns that made use of
2302 duplicate group numbers, could behave incorrectly or give incorrect errors,
2303 because when scanning forward to find the reference group, PCRE was not
2304 taking into account the duplicate group numbers. A pattern such as
2305 ^X(?3)(a)(?|(b)|(q))(Y) is an example.
2306
2307 26. Changed a few more instances of "const unsigned char *" to USPTR, making
2308 the feature of a custom pointer more persuasive (as requested by a user).
2309
2310 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
2311 pcretest.c, inside #ifndefs, because it seems they are sometimes already
2312 pre-defined.
2313
2314 28. Added support for (*UTF8) at the start of a pattern.
2315
2316 29. Arrange for flags added by the "release type" setting in CMake to be shown
2317 in the configuration summary.
2318
2319
2320 Version 7.8 05-Sep-08
2321 ---------------------
2322
2323 1. Replaced UCP searching code with optimized version as implemented for Ad
2324 Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
2325 stage table and inline lookup instead of a function, giving speed ups of 2
2326 to 5 times on some simple patterns that I tested. Permission was given to
2327 distribute the MultiStage2.py script that generates the tables (it's not in
2328 the tarball, but is in the Subversion repository).
2329
2330 2. Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
2331 scripts.
2332
2333 3. Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
2334 a group with a zero qualifier. The result of the study could be incorrect,
2335 or the function might crash, depending on the pattern.
2336
2337 4. Caseless matching was not working for non-ASCII characters in back
2338 references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
2339 It now works when Unicode Property Support is available.
2340
2341 5. In pcretest, an escape such as \x{de} in the data was always generating
2342 a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
2343 non-UTF-8 mode. If the value is greater than 255, it gives a warning about
2344 truncation.
2345
2346 6. Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
2347
2348 7. Added two (int) casts to pcregrep when printing the difference of two
2349 pointers, in case they are 64-bit values.
2350
2351 8. Added comments about Mac OS X stack usage to the pcrestack man page and to
2352 test 2 if it fails.
2353
2354 9. Added PCRE_CALL_CONVENTION just before the names of all exported functions,
2355 and a #define of that name to empty if it is not externally set. This is to
2356 allow users of MSVC to set it if necessary.
2357
2358 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
2359 the convenience functions in the pcre_get.c source file.
2360
2361 11. An option change at the start of a pattern that had top-level alternatives
2362 could cause overwriting and/or a crash. This command provoked a crash in
2363 some environments:
2364
2365 printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
2366
2367 This potential security problem was recorded as CVE-2008-2371.
2368
2369 12. For a pattern where the match had to start at the beginning or immediately
2370 after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
2371 pcre_dfa_exec() could read past the end of the passed subject if there was
2372 no match. To help with detecting such bugs (e.g. with valgrind), I modified
2373 pcretest so that it places the subject at the end of its malloc-ed buffer.
2374
2375 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
2376 exec() might read past the end of the data buffer in UTF-8 mode.
2377
2378 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
2379 the data contained the byte 0x85 as part of a UTF-8 character within its
2380 first line. This applied both to normal and DFA matching.
2381
2382 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
2383 /^[^d]*?$/8 failed to match "abc".
2384
2385 16. Added a missing copyright notice to pcrecpp_internal.h.
2386
2387 17. Make it more clear in the documentation that values returned from
2388 pcre_exec() in ovector are byte offsets, not character counts.
2389
2390 18. Tidied a few places to stop certain compilers from issuing warnings.
2391
2392 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
2393 supplied by Stefan Weber. I made a further small update for 7.8 because
2394 there is a change of source arrangements: the pcre_searchfuncs.c module is
2395 replaced by pcre_ucd.c.
2396
2397
2398 Version 7.7 07-May-08
2399 ---------------------
2400
2401 1. Applied Craig's patch to sort out a long long problem: "If we can't convert
2402 a string to a long long, pretend we don't even have a long long." This is
2403 done by checking for the strtoq, strtoll, and _strtoi64 functions.
2404
2405 2. Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
2406 pre-7.6 versions, which defined a global no_arg variable instead of putting
2407 it in the RE class. (See also #8 below.)
2408
2409 3. Remove a line of dead code, identified by coverity and reported by Nuno
2410 Lopes.
2411
2412 4. Fixed two related pcregrep bugs involving -r with --include or --exclude:
2413
2414 (1) The include/exclude patterns were being applied to the whole pathnames
2415 of files, instead of just to the final components.
2416
2417 (2) If there was more than one level of directory, the subdirectories were
2418 skipped unless they satisfied the include/exclude conditions. This is
2419 inconsistent with GNU grep (and could even be seen as contrary to the
2420 pcregrep specification - which I improved to make it absolutely clear).
2421 The action now is always to scan all levels of directory, and just
2422 apply the include/exclude patterns to regular files.
2423
2424 5. Added the --include_dir and --exclude_dir patterns to pcregrep, and used
2425 --exclude_dir in the tests to avoid scanning .svn directories.
2426
2427 6. Applied Craig's patch to the QuoteMeta function so that it escapes the
2428 NUL character as backslash + 0 rather than backslash + NUL, because PCRE
2429 doesn't support NULs in patterns.
2430
2431 7. Added some missing "const"s to declarations of static tables in
2432 pcre_compile.c and pcre_dfa_exec.c.
2433
2434 8. Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
2435 caused by fix #2 above. (Subsequently also a second patch to fix the
2436 first patch. And a third patch - this was a messy problem.)
2437
2438 9. Applied Craig's patch to remove the use of push_back().
2439
2440 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
2441 matching function regexec().
2442
2443 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
2444 which, however, unlike Perl's \g{...}, are subroutine calls, not back
2445 references. PCRE supports relative numbers with this syntax (I don't think
2446 Oniguruma does).
2447
2448 12. Previously, a group with a zero repeat such as (...){0} was completely
2449 omitted from the compiled regex. However, this means that if the group
2450 was called as a subroutine from elsewhere in the pattern, things went wrong
2451 (an internal error was given). Such groups are now left in the compiled
2452 pattern, with a new opcode that causes them to be skipped at execution
2453 time.
2454
2455 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
2456 to the way PCRE behaves:
2457
2458 (a) A lone ] character is dis-allowed (Perl treats it as data).
2459
2460 (b) A back reference to an unmatched subpattern matches an empty string
2461 (Perl fails the current match path).
2462
2463 (c) A data ] in a character class must be notated as \] because if the
2464 first data character in a class is ], it defines an empty class. (In
2465 Perl it is not possible to have an empty class.) The empty class []
2466 never matches; it forces failure and is equivalent to (*FAIL) or (?!).
2467 The negative empty class [^] matches any one character, independently
2468 of the DOTALL setting.
2469
2470 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
2471 non-existent subpattern following a character class starting with ']' and
2472 containing () gave an internal compiling error instead of "reference to
2473 non-existent subpattern". Fortunately, when the pattern did exist, the
2474 compiled code was correct. (When scanning forwards to check for the
2475 existence of the subpattern, it was treating the data ']' as terminating
2476 the class, so got the count wrong. When actually compiling, the reference
2477 was subsequently set up correctly.)
2478
2479 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
2480 it was being rejected as not supported by pcre_dfa_exec(), even though
2481 other assertions are supported. I have made pcre_dfa_exec() support
2482 (*FAIL).
2483
2484 16. The implementation of 13c above involved the invention of a new opcode,
2485 OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
2486 cannot be changed at match time, I realized I could make a small
2487 improvement to matching performance by compiling OP_ALLANY instead of
2488 OP_ANY for "." when DOTALL was set, and then removing the runtime tests
2489 on the OP_ANY path.
2490
2491 17. Compiling pcretest on Windows with readline support failed without the
2492 following two fixes: (1) Make the unistd.h include conditional on
2493 HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
2494
2495 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
2496 ncurses library to be included for pcretest when ReadLine support is
2497 requested, but also to allow for it to be overridden. This patch came from
2498 Daniel Bergström.
2499
2500 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
2501 as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
2502 any errors with the current Unicode tables. Thanks to Peter Kankowski for
2503 spotting this.
2504
2505
2506 Version 7.6 28-Jan-08
2507 ---------------------
2508
2509 1. A character class containing a very large number of characters with
2510 codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
2511 overflow.
2512
2513 2. Patch to cut out the "long long" test in pcrecpp_unittest when
2514 HAVE_LONG_LONG is not defined.
2515
2516 3. Applied Christian Ehrlicher's patch to update the CMake build files to
2517 bring them up to date and include new features. This patch includes:
2518
2519 - Fixed PH's badly added libz and libbz2 support.
2520 - Fixed a problem with static linking.
2521 - Added pcredemo. [But later removed - see 7 below.]
2522 - Fixed dftables problem and added an option.
2523 - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
2524 HAVE_LONG_LONG.
2525 - Added readline support for pcretest.
2526 - Added an listing of the option settings after cmake has run.
2527
2528 4. A user submitted a patch to Makefile that makes it easy to create
2529 "pcre.dll" under mingw when using Configure/Make. I added stuff to
2530 Makefile.am that cause it to include this special target, without
2531 affecting anything else. Note that the same mingw target plus all
2532 the other distribution libraries and programs are now supported
2533 when configuring with CMake (see 6 below) instead of with
2534 Configure/Make.
2535
2536 5. Applied Craig's patch that moves no_arg into the RE class in the C++ code.
2537 This is an attempt to solve the reported problem "pcrecpp::no_arg is not
2538 exported in the Windows port". It has not yet been confirmed that the patch
2539 solves the problem, but it does no harm.
2540
2541 6. Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
2542 NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
2543 with CMake, and also correct the comment about stack recursion.
2544
2545 7. Remove the automatic building of pcredemo from the ./configure system and
2546 from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
2547 of a program that users should build themselves after PCRE is installed, so
2548 building it automatically is not really right. What is more, it gave
2549 trouble in some build environments.
2550
2551 8. Further tidies to CMakeLists.txt from Sheri and Christian.
2552
2553
2554 Version 7.5 10-Jan-08
2555 ---------------------
2556
2557 1. Applied a patch from Craig: "This patch makes it possible to 'ignore'
2558 values in parens when parsing an RE using the C++ wrapper."
2559
2560 2. Negative specials like \S did not work in character classes in UTF-8 mode.
2561 Characters greater than 255 were excluded from the class instead of being
2562 included.
2563
2564 3. The same bug as (2) above applied to negated POSIX classes such as
2565 [:^space:].
2566
2567 4. PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
2568 defined or documented. It seems to have been a typo for PCRE_STATIC, so
2569 I have changed it.
2570
2571 5. The construct (?&) was not diagnosed as a syntax error (it referenced the
2572 first named subpattern) and a construct such as (?&a) would reference the
2573 first named subpattern whose name started with "a" (in other words, the
2574 length check was missing). Both these problems are fixed. "Subpattern name
2575 expected" is now given for (?&) (a zero-length name), and this patch also
2576 makes it give the same error for \k'' (previously it complained that that
2577 was a reference to a non-existent subpattern).
2578
2579 6. The erroneous patterns (?+-a) and (?-+a) give different error messages;
2580 this is right because (?- can be followed by option settings as well as by
2581 digits. I have, however, made the messages clearer.
2582
2583 7. Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
2584 than the number used in the conditional) now cause a compile-time error.
2585 This is actually not compatible with Perl, which accepts such patterns, but
2586 treats the conditional as always being FALSE (as PCRE used to), but it
2587 seems to me that giving a diagnostic is better.
2588
2589 8. Change "alphameric" to the more common word "alphanumeric" in comments
2590 and messages.
2591
2592 9. Fix two occurrences of "backslash" in comments that should have been
2593 "backspace".
2594
2595 10. Remove two redundant lines of code that can never be obeyed (their function
2596 was moved elsewhere).
2597
2598 11. The program that makes PCRE's Unicode character property table had a bug
2599 which caused it to generate incorrect table entries for sequences of
2600 characters that have the same character type, but are in different scripts.
2601 It amalgamated them into a single range, with the script of the first of
2602 them. In other words, some characters were in the wrong script. There were
2603 thirteen such cases, affecting characters in the following ranges:
2604
2605 U+002b0 - U+002c1
2606 U+0060c - U+0060d
2607 U+0061e - U+00612
2608 U+0064b - U+0065e
2609 U+0074d - U+0076d
2610 U+01800 - U+01805
2611 U+01d00 - U+01d77
2612 U+01d9b - U+01dbf
2613 U+0200b - U+0200f
2614 U+030fc - U+030fe
2615 U+03260 - U+0327f
2616 U+0fb46 - U+0fbb1
2617 U+10450 - U+1049d
2618
2619 12. The -o option (show only the matching part of a line) for pcregrep was not
2620 compatible with GNU grep in that, if there was more than one match in a
2621 line, it showed only the first of them. It now behaves in the same way as
2622 GNU grep.
2623
2624 13. If the -o and -v options were combined for pcregrep, it printed a blank
2625 line for every non-matching line. GNU grep prints nothing, and pcregrep now
2626 does the same. The return code can be used to tell if there were any
2627 non-matching lines.
2628
2629 14. Added --file-offsets and --line-offsets to pcregrep.
2630
2631 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
2632 infinitely looping recursion. The bug was that positive lookaheads were not
2633 being skipped when checking for a possible empty match (negative lookaheads
2634 and both kinds of lookbehind were skipped).
2635
2636 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
2637 inclusion of <windows.h> to before rather than after the definition of
2638 INVALID_FILE_ATTRIBUTES (patch from David Byron).
2639
2640 17. Specifying a possessive quantifier with a specific limit for a Unicode
2641 character property caused pcre_compile() to compile bad code, which led at
2642 runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
2643 are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
2644 caused the error; without that there was no problem.
2645
2646 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
2647
2648 19. Added --enable-pcretest-libreadline.
2649
2650 20. In pcrecpp.cc, the variable 'count' was incremented twice in
2651 RE::GlobalReplace(). As a result, the number of replacements returned was
2652 double what it should be. I removed one of the increments, but Craig sent a
2653 later patch that removed the other one (the right fix) and added unit tests
2654 that check the return values (which was not done before).
2655
2656 21. Several CMake things:
2657
2658 (1) Arranged that, when cmake is used on Unix, the libraries end up with
2659 the names libpcre and libpcreposix, not just pcre and pcreposix.
2660
2661 (2) The above change means that pcretest and pcregrep are now correctly
2662 linked with the newly-built libraries, not previously installed ones.
2663
2664 (3) Added PCRE_SUPPORT_LIBREADLINE, PCRE_SUPPORT_LIBZ, PCRE_SUPPORT_LIBBZ2.
2665
2666 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
2667 crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
2668 UTF-8 newline character). The key issue is that the pattern starts .*;
2669 this means that the match must be either at the beginning, or after a
2670 newline. The bug was in the code for advancing after a failed match and
2671 checking that the new position followed a newline. It was not taking
2672 account of UTF-8 characters correctly.
2673
2674 23. PCRE was behaving differently from Perl in the way it recognized POSIX
2675 character classes. PCRE was not treating the sequence [:...:] as a
2676 character class unless the ... were all letters. Perl, however, seems to
2677 allow any characters between [: and :], though of course it rejects as
2678 unknown any "names" that contain non-letters, because all the known class
2679 names consist only of letters. Thus, Perl gives an error for [[:1234:]],
2680 for example, whereas PCRE did not - it did not recognize a POSIX character
2681 class. This seemed a bit dangerous, so the code has been changed to be
2682 closer to Perl. The behaviour is not identical to Perl, because PCRE will
2683 diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
2684 treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
2685 Perl does, and where it didn't before.
2686
2687 24. Rewrite so as to remove the single use of %n from pcregrep because in some
2688 Windows environments %n is disabled by default.
2689
2690
2691 Version 7.4 21-Sep-07
2692 ---------------------
2693
2694 1. Change 7.3/28 was implemented for classes by looking at the bitmap. This
2695 means that a class such as [\s] counted as "explicit reference to CR or
2696 LF". That isn't really right - the whole point of the change was to try to
2697 help when there was an actual mention of one of the two characters. So now
2698 the change happens only if \r or \n (or a literal CR or LF) character is
2699 encountered.
2700
2701 2. The 32-bit options word was also used for 6 internal flags, but the numbers
2702 of both had grown to the point where there were only 3 bits left.
2703 Fortunately, there was spare space in the data structure, and so I have
2704 moved the internal flags into a new 16-bit field to free up more option
2705 bits.
2706
2707 3. The appearance of (?J) at the start of a pattern set the DUPNAMES option,
2708 but did not set the internal JCHANGED flag - either of these is enough to
2709 control the way the "get" function works - but the PCRE_INFO_JCHANGED
2710 facility is supposed to tell if (?J) was ever used, so now (?J) at the
2711 start sets both bits.
2712
2713 4. Added options (at build time, compile time, exec time) to change \R from
2714 matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
2715
2716 5. doc/pcresyntax.html was missing from the distribution.
2717
2718 6. Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
2719 compatibility, even though it is no longer used.
2720
2721 7. Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
2722 strtoull to pcrecpp.cc to select the available functions in WIN32 when the
2723 windows.h file is present (where different names are used). [This was
2724 reversed later after testing - see 16 below.]
2725
2726 8. Changed all #include <config.h> to #include "config.h". There were also
2727 some further <pcre.h> cases that I changed to "pcre.h".
2728
2729 9. When pcregrep was used with the --colour option, it missed the line ending
2730 sequence off the lines that it output.
2731
2732 10. It was pointed out to me that arrays of string pointers cause lots of
2733 relocations when a shared library is dynamically loaded. A technique of
2734 using a single long string with a table of offsets can drastically reduce
2735 these. I have refactored PCRE in four places to do this. The result is
2736 dramatic:
2737
2738 Originally: 290
2739 After changing UCP table: 187
2740 After changing error message table: 43
2741 After changing table of "verbs" 36
2742 After changing table of Posix names 22
2743
2744 Thanks to the folks working on Gregex for glib for this insight.
2745
2746 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
2747 unicode-properties was also set.
2748
2749 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
2750
2751 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
2752 checked only for CRLF.
2753
2754 14. Added casts to pcretest.c to avoid compiler warnings.
2755
2756 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
2757
2758 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
2759 and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
2760 entirely. This removes changes made in 7 above.
2761
2762 17. The CMake files have been updated, and there is now more information about
2763 building with CMake in the NON-UNIX-USE document.
2764
2765
2766 Version 7.3 28-Aug-07
2767 ---------------------
2768
2769 1. In the rejigging of the build system that eventually resulted in 7.1, the
2770 line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
2771 brackets there is not right, since it causes compilers to look for an
2772 installed pcre.h, not the version that is in the source that is being
2773 compiled (which of course may be different). I have changed it back to:
2774
2775 #include "pcre.h"
2776
2777 I have a vague recollection that the change was concerned with compiling in
2778 different directories, but in the new build system, that is taken care of
2779 by the VPATH setting the Makefile.
2780
2781 2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
2782 when the subject happened to end in the byte 0x85 (e.g. if the last
2783 character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
2784 characters but of course it shouldn't be taken as a newline when it is part
2785 of another character. The bug was that, for an unlimited repeat of . in
2786 not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
2787 characters when looking for a newline.
2788
2789 3. A small performance improvement in the DOTALL UTF-8 mode .* case.
2790
2791 4. Debugging: adjusted the names of opcodes for different kinds of parentheses
2792 in debug output.
2793
2794 5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
2795 long printing in the pcrecpp unittest when running under MinGW.
2796
2797 6. ESC_K was left out of the EBCDIC table.
2798
2799 7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
2800 parentheses; I made it 1000, which seemed large enough. Unfortunately, the
2801 limit also applies to "virtual nesting" when a pattern is recursive, and in
2802 this case 1000 isn't so big. I have been able to remove this limit at the
2803 expense of backing off one optimization in certain circumstances. Normally,
2804 when pcre_exec() would call its internal match() function recursively and
2805 immediately return the result unconditionally, it uses a "tail recursion"
2806 feature to save stack. However, when a subpattern that can match an empty
2807 string has an unlimited repetition quantifier, it no longer makes this
2808 optimization. That gives it a stack frame in which to save the data for
2809 checking that an empty string has been matched. Previously this was taken
2810 from the 1000-entry workspace that had been reserved. So now there is no
2811 explicit limit, but more stack is used.
2812
2813 8. Applied Daniel's patches to solve problems with the import/export magic
2814 syntax that is required for Windows, and which was going wrong for the
2815 pcreposix and pcrecpp parts of the library. These were overlooked when this
2816 problem was solved for the main library.
2817
2818 9. There were some crude static tests to avoid integer overflow when computing
2819 the size of patterns that contain repeated groups with explicit upper
2820 limits. As the maximum quantifier is 65535, the maximum group length was
2821 set at 30,000 so that the product of these two numbers did not overflow a
2822 32-bit integer. However, it turns out that people want to use groups that
2823 are longer than 30,000 bytes (though not repeat them that many times).
2824 Change 7.0/17 (the refactoring of the way the pattern size is computed) has
2825 made it possible to implement the integer overflow checks in a much more
2826 dynamic way, which I have now done. The artificial limitation on group
2827 length has been removed - we now have only the limit on the total length of
2828 the compiled pattern, which depends on the LINK_SIZE setting.
2829
2830 10. Fixed a bug in the documentation for get/copy named substring when
2831 duplicate names are permitted. If none of the named substrings are set, the
2832 functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
2833 empty string.
2834
2835 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
2836 instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
2837 because the ] is interpreted as the first data character and the
2838 terminating ] is not found. PCRE has been made compatible with Perl in this
2839 regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
2840 cause memory overwriting.
2841
2842 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
2843 string has been matched (to stop an infinite loop). It was not recognizing
2844 a conditional subpattern that could match an empty string if that
2845 subpattern was within another subpattern. For example, it looped when
2846 trying to match (((?(1)X|))*) but it was OK with ((?(1)X|)*) where the
2847 condition was not nested. This bug has been fixed.
2848
2849 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
2850 past the start of the subject in the presence of bytes with the top bit
2851 set, for example "\x8aBCD".
2852
2853 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
2854 (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
2855
2856 14. Optimized (?!) to (*FAIL).
2857
2858 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
2859 This restricts code points to be within the range 0 to 0x10FFFF, excluding
2860 the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
2861 full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
2862 does: it's just the validity check that is more restrictive.
2863
2864 16. Inserted checks for integer overflows during escape sequence (backslash)
2865 processing, and also fixed erroneous offset values for syntax errors during
2866 backslash processing.
2867
2868 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
2869 for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
2870
2871 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
2872 caused an overrun.
2873
2874 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
2875 something other than just ASCII characters) inside a group that had an
2876 unlimited repeat caused a loop at compile time (while checking to see
2877 whether the group could match an empty string).
2878
2879 20. Debugging a pattern containing \p or \P could cause a crash. For example,
2880 [\P{Any}] did so. (Error in the code for printing property names.)
2881
2882 21. An orphan \E inside a character class could cause a crash.
2883
2884 22. A repeated capturing bracket such as (A)? could cause a wild memory
2885 reference during compilation.
2886
2887 23. There are several functions in pcre_compile() that scan along a compiled
2888 expression for various reasons (e.g. to see if it's fixed length for look
2889 behind). There were bugs in these functions when a repeated \p or \P was
2890 present in the pattern. These operators have additional parameters compared
2891 with \d, etc, and these were not being taken into account when moving along
2892 the compiled data. Specifically:
2893
2894 (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
2895 length.
2896
2897 (b) An item such as \pL+ within a repeated group could cause crashes or
2898 loops.
2899
2900 (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
2901 "reference to non-existent subpattern" error.
2902
2903 (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
2904
2905 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
2906 characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
2907
2908 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
2909
2910 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
2911 character were causing crashes (broken optimization).
2912
2913 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
2914 \p or \P) caused a compile-time loop.
2915
2916 28. More problems have arisen in unanchored patterns when CRLF is a valid line
2917 break. For example, the unstudied pattern [\r\n]A does not match the string
2918 "\r\nA" because change 7.0/46 below moves the current point on by two
2919 characters after failing to match at the start. However, the pattern \nA
2920 *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
2921 the same is true. There doesn't seem any very clean way out of this, but
2922 what I have chosen to do makes the common cases work: PCRE now takes note
2923 of whether there can be an explicit match for \r or \n anywhere in the
2924 pattern, and if so, 7.0/46 no longer applies. As part of this change,
2925 there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
2926 pattern has explicit CR or LF references.
2927
2928 29. Added (*CR) etc for changing newline setting at start of pattern.
2929
2930
2931 Version 7.2 19-Jun-07
2932 ---------------------
2933
2934 1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
2935 which is apparently normally available under Windows.
2936
2937 2. Re-jig the pcregrep tests with different newline settings in an attempt
2938 to make them independent of the local environment's newline setting.
2939
2940 3. Add code to configure.ac to remove -g from the CFLAGS default settings.
2941
2942 4. Some of the "internals" tests were previously cut out when the link size
2943 was not 2, because the output contained actual offsets. The recent new
2944 "Z" feature of pcretest means that these can be cut out, making the tests
2945 usable with all link sizes.
2946
2947 5. Implemented Stan Switzer's goto replacement for longjmp() when not using
2948 stack recursion. This gives a massive performance boost under BSD, but just
2949 a small improvement under Linux. However, it saves one field in the frame
2950 in all cases.
2951
2952 6. Added more features from the forthcoming Perl 5.10:
2953
2954 (a) (?-n) (where n is a string of digits) is a relative subroutine or
2955 recursion call. It refers to the nth most recently opened parentheses.
2956
2957 (b) (?+n) is also a relative subroutine call; it refers to the nth next
2958 to be opened parentheses.
2959
2960 (c) Conditions that refer to capturing parentheses can be specified
2961 relatively, for example, (?(-2)... or (?(+3)...
2962
2963 (d) \K resets the start of the current match so that everything before
2964 is not part of it.
2965
2966 (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
2967
2968 (f) \g{name} is another synonym - part of Perl 5.10's unification of
2969 reference syntax.
2970
2971 (g) (?| introduces a group in which the numbering of parentheses in each
2972 alternative starts with the same number.
2973
2974 (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
2975
2976 7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
2977 PCRE_INFO_JCHANGED.
2978
2979 8. A pattern such as (.*(.)?)* caused pcre_exec() to fail by either not
2980 terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
2981 for detecting groups that can match an empty string.
2982
2983 9. A pattern with a very large number of alternatives (more than several
2984 hundred) was running out of internal workspace during the pre-compile
2985 phase, where pcre_compile() figures out how much memory will be needed. A
2986 bit of new cunning has reduced the workspace needed for groups with
2987 alternatives. The 1000-alternative test pattern now uses 12 bytes of
2988 workspace instead of running out of the 4096 that are available.
2989
2990 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
2991
2992 11. Applied patch from Google to remove an optimization that didn't quite work.
2993 The report of the bug said:
2994
2995 pcrecpp::RE("a*").FullMatch("aaa") matches, while
2996 pcrecpp::RE("a*?").FullMatch("aaa") does not, and
2997 pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
2998
2999 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
3000 it matched the wrong number of bytes.
3001
3002
3003 Version 7.1 24-Apr-07
3004 ---------------------
3005
3006 1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
3007 that is more "standard", making use of automake and other Autotools. There
3008 is some re-arrangement of the files and adjustment of comments consequent
3009 on this.
3010
3011 2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
3012 for recursive directory scanning broke on some systems because the files
3013 are not scanned in any specific order and on different systems the order
3014 was different. A call to "sort" has been inserted into RunGrepTest for the
3015 approprate test as a short-term fix. In the longer term there may be an
3016 alternative.
3017
3018 3. I had an email from Eric Raymond about problems translating some of PCRE's
3019 man pages to HTML (despite the fact that I distribute HTML pages, some
3020 people do their own conversions for various reasons). The problems
3021 concerned the use of low-level troff macros .br and .in. I have therefore
3022 removed all such uses from the man pages (some were redundant, some could
3023 be replaced by .nf/.fi pairs). The 132html script that I use to generate
3024 HTML has been updated to handle .nf/.fi and to complain if it encounters
3025 .br or .in.
3026
3027 4. Updated comments in configure.ac that get placed in config.h.in and also
3028 arranged for config.h to be included in the distribution, with the name
3029 config.h.generic, for the benefit of those who have to compile without
3030 Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
3031
3032 5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
3033 Weber: (1) pcre_internal.h was missing some function renames; (2) updated
3034 makevp.bat for the current PCRE, using the additional files
3035 makevp_c.txt, makevp_l.txt, and pcregexp.pas.
3036
3037 6. A Windows user reported a minor discrepancy with test 2, which turned out
3038 to be caused by a trailing space on an input line that had got lost in his
3039 copy. The trailing space was an accident, so I've just removed it.
3040
3041 7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
3042 that is needed.
3043
3044 8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
3045 as "const" (a) because they are and (b) because it helps the PHP
3046 maintainers who have recently made a script to detect big data structures
3047 in the php code that should be moved to the .rodata section. I remembered
3048 to update Builducptable as well, so it won't revert if ucptable.h is ever
3049 re-created.
3050
3051 9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
3052 pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
3053 order to be able to cut out the UTF-8 tables in the latter when UTF-8
3054 support is not required. This saves 1.5-2K of code, which is important in
3055 some applications.
3056
3057 Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
3058 so as not to refer to the tables, even though these functions will never be
3059 called when UTF-8 support is disabled. Otherwise there are problems with a
3060 shared library.
3061
3062 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
3063
3064 (a) It was defining its arguments as char * instead of void *.
3065
3066 (b) It was assuming that all moves were upwards in memory; this was true
3067 a long time ago when I wrote it, but is no longer the case.
3068
3069 The emulated memove() is provided for those environments that have neither
3070 memmove() nor bcopy(). I didn't think anyone used it these days, but that
3071 is clearly not the case, as these two bugs were recently reported.
3072
3073 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
3074 and Detrail to create the HTML documentation, the .txt form of the man
3075 pages, and it removes trailing spaces from listed files. It also creates
3076 pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
3077 case, it wraps all the #defines with #ifndefs. This script should be run
3078 before "make dist".
3079
3080 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
3081 with Unicode property support.
3082
3083 (a) For a maximizing quantifier, if the two different cases of the
3084 character were of different lengths in their UTF-8 codings (there are
3085 some cases like this - I found 11), and the matching function had to
3086 back up over a mixture of the two cases, it incorrectly assumed they
3087 were both the same length.
3088
3089 (b) When PCRE was configured to use the heap rather than the stack for
3090 recursion during matching, it was not correctly preserving the data for
3091 the other case of a UTF-8 character when checking ahead for a match
3092 while processing a minimizing repeat. If the check also involved
3093 matching a wide character, but failed, corruption could cause an
3094 erroneous result when trying to check for a repeat of the original
3095 character.
3096
3097 13. Some tidying changes to the testing mechanism:
3098
3099 (a) The RunTest script now detects the internal link size and whether there
3100 is UTF-8 and UCP support by running ./pcretest -C instead of relying on
3101 values substituted by "configure". (The RunGrepTest script already did
3102 this for UTF-8.) The configure.ac script no longer substitutes the
3103 relevant variables.
3104
3105 (b) The debugging options /B and /D in pcretest show the compiled bytecode
3106 with length and offset values. This means that the output is different
3107 for different internal link sizes. Test 2 is skipped for link sizes
3108 other than 2 because of this, bypassing the problem. Unfortunately,
3109 there was also a test in test 3 (the locale tests) that used /B and
3110 failed for link sizes other than 2. Rather than cut the whole test out,
3111 I have added a new /Z option to pcretest that replaces the length and
3112 offset values with spaces. This is now used to make test 3 independent
3113 of link size. (Test 2 will be tidied up later.)
3114
3115 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
3116 segmentation fault instead of returning the appropriate error message.
3117
3118 15. In multiline mode when the newline sequence was set to "any", the pattern
3119 ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
3120 This doesn't seem right; it now treats the CRLF combination as the line
3121 ending, and so does not match in that case. It's only a pattern such as ^$
3122 that would hit this one: something like ^ABC$ would have failed after \r
3123 and then tried again after \r\n.
3124
3125 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
3126 in an attempt to make files that differ only in their line terminators
3127 compare equal. This works on Linux.
3128
3129 17. Under certain error circumstances pcregrep might try to free random memory
3130 as it exited. This is now fixed, thanks to valgrind.
3131
3132 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
3133 "abc\r\n\r\n", it found an unwanted second match after the second \r. This
3134 was because its rules for how to advance for /g after matching an empty
3135 string at the end of a line did not allow for this case. They now check for
3136 it specially.
3137
3138 20. pcretest is supposed to handle patterns and data of any length, by
3139 extending its buffers when necessary. It was getting this wrong when the
3140 buffer for a data line had to be extended.
3141
3142 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
3143 CRLF as a newline sequence.
3144
3145 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
3146 out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
3147 I have nevertheless tidied it up.
3148
3149 23. Added some casts to kill warnings from HP-UX ia64 compiler.
3150
3151 24. Added a man page for pcre-config.
3152
3153
3154 Version 7.0 19-Dec-06
3155 ---------------------
3156
3157 1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
3158 moving to gcc 4.1.1.
3159
3160 2. The -S option for pcretest uses setrlimit(); I had omitted to #include
3161 sys/time.h, which is documented as needed for this function. It doesn't
3162 seem to matter on Linux, but it showed up on some releases of OS X.
3163
3164 3. It seems that there are systems where bytes whose values are greater than
3165 127 match isprint() in the "C" locale. The "C" locale should be the
3166 default when a C program starts up. In most systems, only ASCII printing
3167 characters match isprint(). This difference caused the output from pcretest
3168 to vary, making some of the tests fail. I have changed pcretest so that:
3169
3170 (a) When it is outputting text in the compiled version of a pattern, bytes
3171 other than 32-126 are always shown as hex escapes.
3172
3173 (b) When it is outputting text that is a matched part of a subject string,
3174 it does the same, unless a different locale has been set for the match
3175 (using the /L modifier). In this case, it uses isprint() to decide.
3176
3177 4. Fixed a major bug that caused incorrect computation of the amount of memory
3178 required for a compiled pattern when options that changed within the
3179 pattern affected the logic of the preliminary scan that determines the
3180 length. The relevant options are -x, and -i in UTF-8 mode. The result was
3181 that the computed length was too small. The symptoms of this bug were
3182 either the PCRE error "internal error: code overflow" from pcre_compile(),
3183 or a glibc crash with a message such as "pcretest: free(): invalid next
3184 size (fast)". Examples of patterns that provoked this bug (shown in
3185 pcretest format) are:
3186
3187 /(?-x: )/x
3188 /(?x)(?-x: \s*#\s*)/
3189 /((?i)[\x{c0}])/8
3190 /(?i:[\x{c0}])/8
3191
3192 HOWEVER: Change 17 below makes this fix obsolete as the memory computation
3193 is now done differently.
3194
3195 5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
3196 wrapper classes; (b) implement a new function in the C++ scanner that is
3197 more efficient than the old way of doing things because it avoids levels of
3198 recursion in the regex matching; (c) add a paragraph to the documentation
3199 for the FullMatch() function.
3200
3201 6. The escape sequence \n was being treated as whatever was defined as
3202 "newline". Not only was this contrary to the documentation, which states
3203 that \n is character 10 (hex 0A), but it also went horribly wrong when
3204 "newline" was defined as CRLF. This has been fixed.
3205
3206 7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
3207 was being set to -1 for the "end of line" case (supposedly a value that no
3208 character can have). Though this value is never used (the check for end of
3209 line is "zero bytes in current character"), it caused compiler complaints.
3210 I've changed it to 0xffffffff.
3211
3212 8. In pcre_version.c, the version string was being built by a sequence of
3213 C macros that, in the event of PCRE_PRERELEASE being defined as an empty
3214 string (as it is for production releases) called a macro with an empty
3215 argument. The C standard says the result of this is undefined. The gcc
3216 compiler treats it as an empty string (which was what was wanted) but it is
3217 reported that Visual C gives an error. The source has been hacked around to
3218 avoid this problem.
3219
3220 9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
3221 builds of pcretest, and changed the call to _setmode() to use _O_BINARY
3222 instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
3223 of them did).
3224
3225 10. Originally, pcretest opened its input and output without "b"; then I was
3226 told that "b" was needed in some environments, so it was added for release
3227 5.0 to both the input and output. (It makes no difference on Unix-like
3228 systems.) Later I was told that it is wrong for the input on Windows. I've
3229 now abstracted the modes into two macros, to make it easier to fiddle with
3230 them, and removed "b" from the input mode under Windows.
3231
3232 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
3233
3234 12. Added -help and --help to pcretest as an official way of being reminded
3235 of the options.
3236
3237 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
3238 and pcrecpp.cc because they annoy compilers at high warning levels.
3239
3240 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
3241
3242 15. Fixed an occurrence of == in configure.ac that should have been = (shell
3243 scripts are not C programs :-) and which was not noticed because it works
3244 on Linux.
3245
3246 16. pcretest is supposed to handle any length of pattern and data line (as one
3247 line or as a continued sequence of lines) by extending its input buffer if
3248 necessary. This feature was broken for very long pattern lines, leading to
3249 a string of junk being passed to pcre_compile() if the pattern was longer
3250 than about 50K.
3251
3252 17. I have done a major re-factoring of the way pcre_compile() computes the
3253 amount of memory needed for a compiled pattern. Previously, there was code
3254 that made a preliminary scan of the pattern in order to do this. That was
3255 OK when PCRE was new, but as the facilities have expanded, it has become
3256 harder and harder to keep it in step with the real compile phase, and there
3257 have been a number of bugs (see for example, 4 above). I have now found a
3258 cunning way of running the real compile function in a "fake" mode that
3259 enables it to compute how much memory it would need, while actually only
3260 ever using a few hundred bytes of working memory and without too many
3261 tests of the mode. This should make future maintenance and development
3262 easier. A side effect of this work is that the limit of 200 on the nesting
3263 depth of parentheses has been removed (though this was never a serious
3264 limitation, I suspect). However, there is a downside: pcre_compile() now
3265 runs more slowly than before (30% or more, depending on the pattern). I
3266 hope this isn't a big issue. There is no effect on runtime performance.
3267
3268 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
3269 newline (only possible for the last line of a file) and it was a
3270 pattern that set a locale (followed by /Lsomething), pcretest crashed.
3271
3272 19. Added additional timing features to pcretest. (1) The -tm option now times
3273 matching only, not compiling. (2) Both -t and -tm can be followed, as a
3274 separate command line item, by a number that specifies the number of
3275 repeats to use when timing. The default is 50000; this gives better
3276 precision, but takes uncomfortably long for very large patterns.
3277
3278 20. Extended pcre_study() to be more clever in cases where a branch of a
3279 subpattern has no definite first character. For example, (a*|b*)[cd] would
3280 previously give no result from pcre_study(). Now it recognizes that the
3281 first character must be a, b, c, or d.
3282
3283 21. There was an incorrect error "recursive call could loop indefinitely" if
3284 a subpattern (or the entire pattern) that was being tested for matching an
3285 empty string contained only one non-empty item after a nested subpattern.
3286 For example, the pattern (?>\x{100}*)\d(?R) provoked this error
3287 incorrectly, because the \d was being skipped in the check.
3288
3289 22. The pcretest program now has a new pattern option /B and a command line
3290 option -b, which is equivalent to adding /B to every pattern. This causes
3291 it to show the compiled bytecode, without the additional information that
3292 -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
3293 is the same as /B/I).
3294
3295 23. A new optimization is now able automatically to treat some sequences such
3296 as a*b as a*+b. More specifically, if something simple (such as a character
3297 or a simple class like \d) has an unlimited quantifier, and is followed by
3298 something that cannot possibly match the quantified thing, the quantifier
3299 is automatically "possessified".
3300
3301 24. A recursive reference to a subpattern whose number was greater than 39
3302 went wrong under certain circumstances in UTF-8 mode. This bug could also
3303 have affected the operation of pcre_study().
3304
3305 25. Realized that a little bit of performance could be had by replacing
3306 (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
3307
3308 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
3309
3310 27. Possessive quantifiers such as a++ were previously implemented by turning
3311 them into atomic groups such as ($>a+). Now they have their own opcodes,
3312 which improves performance. This includes the automatically created ones
3313 from 23 above.
3314
3315 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
3316 lookahead was broken if it was not anchored. PCRE was mistakenly expecting
3317 the first matched character to be a colon. This applied both to named and
3318 numbered groups.
3319
3320 29. The ucpinternal.h header file was missing its idempotency #ifdef.
3321
3322 30. I was sent a "project" file called libpcre.a.dev which I understand makes
3323 building PCRE on Windows easier, so I have included it in the distribution.
3324
3325 31. There is now a check in pcretest against a ridiculously large number being
3326 returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
3327 loop, the loop is abandoned.
3328
3329 32. Forward references to subpatterns in conditions such as (?(2)...) where
3330 subpattern 2 is defined later cause pcre_compile() to search forwards in
3331 the pattern for the relevant set of parentheses. This search went wrong
3332 when there were unescaped parentheses in a character class, parentheses
3333 escaped with \Q...\E, or parentheses in a #-comment in /x mode.
3334
3335 33. "Subroutine" calls and backreferences were previously restricted to
3336 referencing subpatterns earlier in the regex. This restriction has now
3337 been removed.
3338
3339 34. Added a number of extra features that are going to be in Perl 5.10. On the
3340 whole, these are just syntactic alternatives for features that PCRE had
3341 previously implemented using the Python syntax or my own invention. The
3342 other formats are all retained for compatibility.
3343
3344 (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
3345 as (?P<name>...). The new forms, as well as being in Perl 5.10, are
3346 also .NET compatible.
3347
3348 (b) A recursion or subroutine call to a named group can now be defined as
3349 (?&name) as well as (?P>name).
3350
3351 (c) A backreference to a named group can now be defined as \k<name> or
3352 \k'name' as well as (?P=name). The new forms, as well as being in Perl
3353 5.10, are also .NET compatible.
3354
3355 (d) A conditional reference to a named group can now use the syntax
3356 (?(<name>) or (?('name') as well as (?(name).
3357
3358 (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
3359 groups (named and numbered) that are never evaluated inline, but can be
3360 called as "subroutines" from elsewhere. In effect, the DEFINE condition
3361 is always false. There may be only one alternative in such a group.
3362
3363 (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
3364 as the simple (?(R). The condition is true only if the most recent
3365 recursion is that of the given number or name. It does not search out
3366 through the entire recursion stack.
3367
3368 (g) The escape \gN or \g{N} has been added, where N is a positive or
3369 negative number, specifying an absolute or relative reference.
3370
3371 35. Tidied to get rid of some further signed/unsigned compiler warnings and
3372 some "unreachable code" warnings.
3373
3374 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
3375 things, this adds five new scripts.
3376
3377 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
3378 There were also incompatibilities regarding the handling of \Q..\E inside
3379 character classes, for example with patterns like [\Qa\E-\Qz\E] where the
3380 hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
3381
3382 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
3383 matches an empty string, and forcibly breaks the loop. There were bugs in
3384 this code in non-simple cases. For a pattern such as ^(a()*)* matched
3385 against aaaa the result was just "a" rather than "aaaa", for example. Two
3386 separate and independent bugs (that affected different cases) have been
3387 fixed.
3388
3389 39. Refactored the code to abolish the use of different opcodes for small
3390 capturing bracket numbers. This is a tidy that I avoided doing when I
3391 removed the limit on the number of capturing brackets for 3.5 back in 2001.
3392 The new approach is not only tidier, it makes it possible to reduce the
3393 memory needed to fix the previous bug (38).
3394
3395 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
3396 sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
3397 processing dot, circumflex, or dollar metacharacters, or #-comments in /x
3398 mode.
3399
3400 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
3401 report.
3402
3403 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
3404 copy construction and assignment in the C++ wrapper.
3405
3406 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
3407 couple of bugs that could have given wrong results in the "--newline=crlf"
3408 case.
3409
3410 44. Added a number of casts and did some reorganization of signed/unsigned int
3411 variables following suggestions from Dair Grant. Also renamed the variable
3412 "this" as "item" because it is a C++ keyword.
3413
3414 45. Arranged for dftables to add
3415
3416 #include "pcre_internal.h"
3417
3418 to pcre_chartables.c because without it, gcc 4.x may remove the array
3419 definition from the final binary if PCRE is built into a static library and
3420 dead code stripping is activated.
3421
3422 46. For an unanchored pattern, if a match attempt fails at the start of a
3423 newline sequence, and the newline setting is CRLF or ANY, and the next two
3424 characters are CRLF, advance by two characters instead of one.
3425
3426
3427 Version 6.7 04-Jul-06
3428 ---------------------
3429
3430 1. In order to handle tests when input lines are enormously long, pcretest has
3431 been re-factored so that it automatically extends its buffers when
3432 necessary. The code is crude, but this _is_ just a test program. The
3433 default size has been increased from 32K to 50K.
3434
3435 2. The code in pcre_study() was using the value of the re argument before
3436 testing it for NULL. (Of course, in any sensible call of the function, it
3437 won't be NULL.)
3438
3439 3. The memmove() emulation function in pcre_internal.h, which is used on
3440 systems that lack both memmove() and bcopy() - that is, hardly ever -
3441 was missing a "static" storage class specifier.
3442
3443 4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
3444 containing an extended class (one that cannot be represented by a bitmap
3445 because it contains high-valued characters or Unicode property items, e.g.
3446 [\pZ]). Almost always one would set UTF-8 mode when processing such a
3447 pattern, but PCRE should not loop if you do not (it no longer does).
3448 [Detail: two cases were found: (a) a repeated subpattern containing an
3449 extended class; (b) a recursive reference to a subpattern that followed a
3450 previous extended class. It wasn't skipping over the extended class
3451 correctly when UTF-8 mode was not set.]
3452
3453 5. A negated single-character class was not being recognized as fixed-length
3454 in lookbehind assertions such as (?<=[^f]), leading to an incorrect
3455 compile error "lookbehind assertion is not fixed length".
3456
3457 6. The RunPerlTest auxiliary script was showing an unexpected difference
3458 between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
3459 write a Perl script that can interpret lines of an input file either as
3460 byte characters or as UTF-8, which is what "perltest" was being required to
3461 do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
3462 can't do is switch easily at run time between having the "use utf8;" pragma
3463 or not. In the end, I fudged it by using the RunPerlTest script to insert
3464 "use utf8;" explicitly for the UTF-8 tests.
3465
3466 7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
3467 the end of the subject string, contrary to the documentation and to what
3468 Perl does. This was true of both matching functions. Now it matches only at
3469 the start of the subject and immediately after *internal* newlines.
3470
3471 8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
3472 a pointer to an int instead of a pointer to an unsigned long int. This
3473 caused problems on 64-bit systems.
3474
3475 9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
3476 instance of the 'standard' template library not being so standard".
3477
3478 10. There was no check on the number of named subpatterns nor the maximum
3479 length of a subpattern name. The product of these values is used to compute
3480 the size of the memory block for a compiled pattern. By supplying a very
3481 long subpattern name and a large number of named subpatterns, the size
3482 computation could be caused to overflow. This is now prevented by limiting
3483 the length of names to 32 characters, and the number of named subpatterns
3484 to 10,000.
3485
3486 11. Subpatterns that are repeated with specific counts have to be replicated in
3487 the compiled pattern. The size of memory for this was computed from the
3488 length of the subpattern and the repeat count. The latter is limited to
3489 65535, but there was no limit on the former, meaning that integer overflow
3490 could in principle occur. The compiled length of a repeated subpattern is
3491 now limited to 30,000 bytes in order to prevent this.
3492
3493 12. Added the optional facility to have named substrings with the same name.
3494
3495 13. Added the ability to use a named substring as a condition, using the
3496 Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
3497 are numbers (not recommended). Forward references are permitted.
3498
3499 14. Added forward references in named backreferences (if you see what I mean).
3500
3501 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
3502 pattern could run off the end of the subject. For example, the pattern
3503 "(?s)(.{1,5})"8 did this with the subject "ab".
3504
3505 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
3506 PCRE_CASELESS was set when matching characters that were quantified with ?
3507 or *.
3508
3509 17. A character class other than a single negated character that had a minimum
3510 but no maximum quantifier - for example [ab]{6,} - was not handled
3511 correctly by pce_dfa_exec(). It would match only one character.
3512
3513 18. A valid (though odd) pattern that looked like a POSIX character
3514 class but used an invalid character after [ (for example [[,abc,]]) caused
3515 pcre_compile() to give the error "Failed: internal error: code overflow" or
3516 in some cases to crash with a glibc free() error. This could even happen if
3517 the pattern terminated after [[ but there just happened to be a sequence of
3518 letters, a binary zero, and a closing ] in the memory that followed.
3519
3520 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
3521 over the years. Originally (before any Unicode support), just the bottom 8
3522 bits were taken. Thus, for example, \500 really meant \100. Nowadays the
3523 output from "man perlunicode" includes this:
3524
3525 The regular expression compiler produces polymorphic opcodes. That
3526 is, the pattern adapts to the data and automatically switches to
3527 the Unicode character scheme when presented with Unicode data--or
3528 instead uses a traditional byte scheme when presented with byte
3529 data.
3530
3531 Sadly, a wide octal escape does not cause a switch, and in a string with
3532 no other multibyte characters, these octal escapes are treated as before.
3533 Thus, in Perl, the pattern /\500/ actually matches \100 but the pattern
3534 /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
3535 Unicode string.
3536
3537 I have not perpetrated such confusion in PCRE. Up till now, it took just
3538 the bottom 8 bits, as in old Perl. I have now made octal escapes with
3539 values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
3540 translate to the appropriate multibyte character.
3541
3542 29. Applied some refactoring to reduce the number of warnings from Microsoft
3543 and Borland compilers. This has included removing the fudge introduced
3544 seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
3545 a warning about an unused variable.
3546
3547 21. PCRE has not included VT (character 0x0b) in the set of whitespace
3548 characters since release 4.0, because Perl (from release 5.004) does not.
3549 [Or at least, is documented not to: some releases seem to be in conflict
3550 with the documentation.] However, when a pattern was studied with
3551 pcre_study() and all its branches started with \s, PCRE still included VT
3552 as a possible starting character. Of course, this did no harm; it just
3553 caused an unnecessary match attempt.
3554
3555 22. Removed a now-redundant internal flag bit that recorded the fact that case
3556 dependency changed within the pattern. This was once needed for "required
3557 byte" processing, but is no longer used. This recovers a now-scarce options
3558 bit. Also moved the least significant internal flag bit to the most-
3559 significant bit of the word, which was not previously used (hangover from
3560 the days when it was an int rather than a uint) to free up another bit for
3561 the future.
3562
3563 23. Added support for CRLF line endings as well as CR and LF. As well as the
3564 default being selectable at build time, it can now be changed at runtime
3565 via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
3566 specify that it is scanning data with non-default line endings.
3567
3568 24. Changed the definition of CXXLINK to make it agree with the definition of
3569 LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
3570
3571 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
3572 recursions. This makes a big different to stack usage for some patterns.
3573
3574 26. If a subpattern containing a named recursion or subroutine reference such
3575 as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
3576 the space required for the compiled pattern went wrong and gave too small a
3577 value. Depending on the environment, this could lead to "Failed: internal
3578 error: code overflow at offset 49" or "glibc detected double free or
3579 corruption" errors.
3580
3581 27. Applied patches from Google (a) to support the new newline modes and (b) to
3582 advance over multibyte UTF-8 characters in GlobalReplace.
3583
3584 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
3585 difference for some implementation of PCRE in some Windows version.
3586
3587 29. Added some extra testing facilities to pcretest:
3588
3589 \q<number> in a data line sets the "match limit" value
3590 \Q<number> in a data line sets the "match recursion limt" value
3591 -S <number> sets the stack size, where <number> is in megabytes
3592
3593 The -S option isn't available for Windows.
3594
3595
3596 Version 6.6 06-Feb-06
3597 ---------------------
3598
3599 1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
3600 in pcreposix.h. I have copied the definition from pcre.h.
3601
3602 2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
3603 because pcre.h is no longer a built file.
3604
3605 3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
3606 not normally included in the compiled code.
3607
3608
3609 Version 6.5 01-Feb-06
3610 ---------------------
3611
3612 1. When using the partial match feature with pcre_dfa_exec(), it was not
3613 anchoring the second and subsequent partial matches at the new starting
3614 point. This could lead to incorrect results. For example, with the pattern
3615 /1234/, partially matching against "123" and then "a4" gave a match.
3616
3617 2. Changes to pcregrep:
3618
3619 (a) All non-match returns from pcre_exec() were being treated as failures
3620 to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
3621 error message is output. Some extra information is given for the
3622 PCRE_ERROR_MATCHLIMIT and PCRE_ERROR_RECURSIONLIMIT errors, which are
3623 probably the only errors that are likely to be caused by users (by
3624 specifying a regex that has nested indefinite repeats, for instance).
3625 If there are more than 20 of these errors, pcregrep is abandoned.
3626
3627 (b) A binary zero was treated as data while matching, but terminated the
3628 output line if it was written out. This has been fixed: binary zeroes
3629 are now no different to any other data bytes.
3630
3631 (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
3632 used to set a locale for matching. The --locale=xxxx long option has
3633 been added (no short equivalent) to specify a locale explicitly on the
3634 pcregrep command, overriding the environment variables.
3635
3636 (d) When -B was used with -n, some line numbers in the output were one less
3637 than they should have been.
3638
3639 (e) Added the -o (--only-matching) option.
3640
3641 (f) If -A or -C was used with -c (count only), some lines of context were
3642 accidentally printed for the final match.
3643
3644 (g) Added the -H (--with-filename) option.
3645
3646 (h) The combination of options -rh failed to suppress file names for files
3647 that were found from directory arguments.
3648
3649 (i) Added the -D (--devices) and -d (--directories) options.
3650
3651 (j) Added the -F (--fixed-strings) option.
3652
3653 (k) Allow "-" to be used as a file name for -f as well as for a data file.
3654
3655 (l) Added the --colo(u)r option.
3656
3657 (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
3658 is not present by default.
3659
3660 3. A nasty bug was discovered in the handling of recursive patterns, that is,
3661 items such as (?R) or (?1), when the recursion could match a number of
3662 alternatives. If it matched one of the alternatives, but subsequently,
3663 outside the recursion, there was a failure, the code tried to back up into
3664 the recursion. However, because of the way PCRE is implemented, this is not
3665 possible, and the result was an incorrect result from the match.
3666
3667 In order to prevent this happening, the specification of recursion has
3668 been changed so that all such subpatterns are automatically treated as
3669 atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
3670
3671 4. I had overlooked the fact that, in some locales, there are characters for
3672 which isalpha() is true but neither isupper() nor islower() are true. In
3673 the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
3674 and ordfeminine) are like this. This affected the treatment of \w and \W
3675 when they appeared in character classes, but not when they appeared outside
3676 a character class. The bit map for "word" characters is now created
3677 separately from the results of isalnum() instead of just taking it from the
3678 upper, lower, and digit maps. (Plus the underscore character, of course.)
3679
3680 5. The above bug also affected the handling of POSIX character classes such as
3681 [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
3682 permanent tables. Instead, the bit maps for such a class were previously
3683 created as the appropriate unions of the upper, lower, and digit bitmaps.
3684 Now they are created by subtraction from the [[:word:]] class, which has
3685 its own bitmap.
3686
3687 6. The [[:blank:]] character class matches horizontal, but not vertical space.
3688 It is created by subtracting the vertical space characters (\x09, \x0a,
3689 \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
3690 subtraction was done in the overall bitmap for a character class, meaning
3691 that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
3692 be recognized. This bug has been fixed.
3693
3694 7. Patches from the folks at Google:
3695
3696 (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
3697 real life, but is still worth protecting against".
3698
3699 (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
3700 regular expressions".
3701
3702 (c) pcre_scanner.cc: avoid use of std::count() because not all systems
3703 have it.
3704
3705 (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
3706 "configure" and the latter not, in order to fix a problem somebody had
3707 with compiling the Arg class on HP-UX.
3708
3709 (e) Improve the error-handling of the C++ wrapper a little bit.
3710
3711 (f) New tests for checking recursion limiting.
3712
3713 8. The pcre_memmove() function, which is used only if the environment does not
3714 have a standard memmove() function (and is therefore rarely compiled),
3715 contained two bugs: (a) use of int instead of size_t, and (b) it was not
3716 returning a result (though PCRE never actually uses the result).
3717
3718 9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
3719 large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
3720 returned instead of calling malloc() with an overflowing number that would
3721 most likely cause subsequent chaos.
3722
3723 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
3724
3725 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
3726 with this option is matched, the nmatch and pmatch options of regexec() are
3727 ignored.
3728
3729 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
3730 provided in case anyone wants to the the POSIX interface with UTF-8
3731 strings.
3732
3733 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
3734 C++ linking (needed for some HP-UX environments).
3735
3736 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
3737 (unused parameter) and in the pcre_printint() function (omitted "default"
3738 switch label when the default is to do nothing).
3739
3740 15. Added some code to make it possible, when PCRE is compiled as a C++
3741 library, to replace subject pointers for pcre_exec() with a smart pointer
3742 class, thus making it possible to process discontinuous strings.
3743
3744 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
3745 much the same function. They were added by different people who were trying
3746 to make PCRE easy to compile on non-Unix systems. It has been suggested
3747 that PCRE_EXPORT be abolished now that there is more automatic apparatus
3748 for compiling on Windows systems. I have therefore replaced it with
3749 PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
3750 defaults to "extern" for C or "extern C" for C++, which works fine on
3751 Unix-like systems. It is now possible to override the value of PCRE_DATA_
3752 SCOPE with something explicit in config.h. In addition:
3753
3754 (a) pcreposix.h still had just "extern" instead of either of these macros;
3755 I have replaced it with PCRE_DATA_SCOPE.
3756
3757 (b) Functions such as _pcre_xclass(), which are internal to the library,
3758 but external in the C sense, all had PCRE_EXPORT in their definitions.
3759 This is apparently wrong for the Windows case, so I have removed it.
3760 (It makes no difference on Unix-like systems.)
3761
3762 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
3763 of recursive calls to match(). This is different to MATCH_LIMIT because
3764 that limits the total number of calls to match(), not all of which increase
3765 the depth of recursion. Limiting the recursion depth limits the amount of
3766 stack (or heap if NO_RECURSE is set) that is used. The default can be set
3767 when PCRE is compiled, and changed at run time. A patch from Google adds
3768 this functionality to the C++ interface.
3769
3770 18. Changes to the handling of Unicode character properties:
3771
3772 (a) Updated the table to Unicode 4.1.0.
3773
3774 (b) Recognize characters that are not in the table as "Cn" (undefined).
3775
3776 (c) I revised the way the table is implemented to a much improved format
3777 which includes recognition of ranges. It now supports the ranges that
3778 are defined in UnicodeData.txt, and it also amalgamates other
3779 characters into ranges. This has reduced the number of entries in the
3780 table from around 16,000 to around 3,000, thus reducing its size
3781 considerably. I realized I did not need to use a tree structure after
3782 all - a binary chop search is just as efficient. Having reduced the
3783 number of entries, I extended their size from 6 bytes to 8 bytes to
3784 allow for more data.
3785
3786 (d) Added support for Unicode script names via properties such as \p{Han}.
3787
3788 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
3789 matching that character.
3790
3791 20. When matching a repeated Unicode property with a minimum greater than zero,
3792 (for example \pL{2,}), PCRE could look past the end of the subject if it
3793 reached it while seeking the minimum number of characters. This could
3794 happen only if some of the characters were more than one byte long, because
3795 there is a check for at least the minimum number of bytes.
3796
3797 21. Refactored the implementation of \p and \P so as to be more general, to
3798 allow for more different types of property in future. This has changed the
3799 compiled form incompatibly. Anybody with saved compiled patterns that use
3800 \p or \P will have to recompile them.
3801
3802 22. Added "Any" and "L&" to the supported property types.
3803
3804 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
3805 but give a compile time error if the value is greater than 0xff.
3806
3807 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
3808 accidentally not being installed or uninstalled.
3809
3810 25. The pcre.h file was built from pcre.h.in, but the only changes that were
3811 made were to insert the current release number. This seemed silly, because
3812 it made things harder for people building PCRE on systems that don't run
3813 "configure". I have turned pcre.h into a distributed file, no longer built
3814 by "configure", with the version identification directly included. There is
3815 no longer a pcre.h.in file.
3816
3817 However, this change necessitated a change to the pcre-config script as
3818 well. It is built from pcre-config.in, and one of the substitutions was the
3819 release number. I have updated configure.ac so that ./configure now finds
3820 the release number by grepping pcre.h.
3821
3822 26. Added the ability to run the tests under valgrind.
3823
3824
3825 Version 6.4 05-Sep-05
3826 ---------------------
3827
3828 1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
3829 "--" to be printed when multiple files were scanned, even when none of the
3830 -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
3831 consider it to be a bug, and have restored the previous behaviour.
3832
3833 2. A couple of code tidies to get rid of compiler warnings.
3834
3835 3. The pcretest program used to cheat by referring to symbols in the library
3836 whose names begin with _pcre_. These are internal symbols that are not
3837 really supposed to be visible externally, and in some environments it is
3838 possible to suppress them. The cheating is now confined to including
3839 certain files from the library's source, which is a bit cleaner.
3840
3841 4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
3842 file's purpose clearer.
3843
3844 5. Reorganized pcre_ucp_findchar().
3845
3846
3847 Version 6.3 15-Aug-05
3848 ---------------------
3849
3850 1. The file libpcre.pc.in did not have general read permission in the tarball.
3851
3852 2. There were some problems when building without C++ support:
3853
3854 (a) If C++ support was not built, "make install" and "make test" still
3855 tried to test it.
3856
3857 (b) There were problems when the value of CXX was explicitly set. Some
3858 changes have been made to try to fix these, and ...
3859
3860 (c) --disable-cpp can now be used to explicitly disable C++ support.
3861
3862 (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
3863 backslash in a target when C++ was disabled. This confuses some
3864 versions of "make", apparently. Using an intermediate variable solves
3865 this. (Same for CPP_LOBJ.)
3866
3867 3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
3868 (non-Windows) now includes $(CFLAGS) because these flags are sometimes
3869 necessary on certain architectures.
3870
3871 4. Added a setting of -export-symbols-regex to the link command to remove
3872 those symbols that are exported in the C sense, but actually are local
3873 within the library, and not documented. Their names all begin with
3874 "_pcre_". This is not a perfect job, because (a) we have to except some
3875 symbols that pcretest ("illegally") uses, and (b) the facility isn't always
3876 available (and never for static libraries). I have made a note to try to
3877 find a way round (a) in the future.
3878
3879
3880 Version 6.2 01-Aug-05
3881 ---------------------
3882
3883 1. There was no test for integer overflow of quantifier values. A construction
3884 such as {1111111111111111} would give undefined results. What is worse, if
3885 a minimum quantifier for a parenthesized subpattern overflowed and became
3886 negative, the calculation of the memory size went wrong. This could have
3887 led to memory overwriting.
3888
3889 2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
3890
3891 3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
3892 operating environments where this matters.
3893
3894 4. Applied Giuseppe Maxia's patch to add additional features for controlling
3895 PCRE options from within the C++ wrapper.
3896
3897 5. Named capturing subpatterns were not being correctly counted when a pattern
3898 was compiled. This caused two problems: (a) If there were more than 100
3899 such subpatterns, the calculation of the memory needed for the whole
3900 compiled pattern went wrong, leading to an overflow error. (b) Numerical
3901 back references of the form \12, where the number was greater than 9, were
3902 not recognized as back references, even though there were sufficient
3903 previous subpatterns.
3904
3905 6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
3906 versions of gcc, e.g. 2.95.4.
3907
3908
3909 Version 6.1 21-Jun-05
3910 ---------------------
3911
3912 1. There was one reference to the variable "posix" in pcretest.c that was not
3913 surrounded by "#if !defined NOPOSIX".
3914
3915 2. Make it possible to compile pcretest without DFA support, UTF8 support, or
3916 the cross-check on the old pcre_info() function, for the benefit of the
3917 cut-down version of PCRE that is currently imported into Exim.
3918
3919 3. A (silly) pattern starting with (?i)(?-i) caused an internal space
3920 allocation error. I've done the easy fix, which wastes 2 bytes for sensible
3921 patterns that start (?i) but I don't think that matters. The use of (?i) is
3922 just an example; this all applies to the other options as well.
3923
3924 4. Since libtool seems to echo the compile commands it is issuing, the output
3925 from "make" can be reduced a bit by putting "@" in front of each libtool
3926 compile command.
3927
3928 5. Patch from the folks at Google for configure.in to be a bit more thorough
3929 in checking for a suitable C++ installation before trying to compile the
3930 C++ stuff. This should fix a reported problem when a compiler was present,
3931 but no suitable headers.
3932
3933 6. The man pages all had just "PCRE" as their title. I have changed them to
3934 be the relevant file name. I have also arranged that these names are
3935 retained in the file doc/pcre.txt, which is a concatenation in text format
3936 of all the man pages except the little individual ones for each function.
3937
3938 7. The NON-UNIX-USE file had not been updated for the different set of source
3939 files that come with release 6. I also added a few comments about the C++
3940 wrapper.
3941
3942
3943 Version 6.0 07-Jun-05
3944 ---------------------
3945
3946 1. Some minor internal re-organization to help with my DFA experiments.
3947
3948 2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
3949 didn't matter for the library itself when fully configured, but did matter
3950 when compiling without UCP support, or within Exim, where the ucp files are
3951 not imported.
3952
3953 3. Refactoring of the library code to split up the various functions into
3954 different source modules. The addition of the new DFA matching code (see
3955 below) to a single monolithic source would have made it really too
3956 unwieldy, quite apart from causing all the code to be include in a
3957 statically linked application, when only some functions are used. This is
3958 relevant even without the DFA addition now that patterns can be compiled in
3959 one application and matched in another.
3960
3961 The downside of splitting up is that there have to be some external
3962 functions and data tables that are used internally in different modules of
3963 the library but which are not part of the API. These have all had their
3964 names changed to start with "_pcre_" so that they are unlikely to clash
3965 with other external names.
3966
3967 4. Added an alternate matching function, pcre_dfa_exec(), which matches using
3968 a different (DFA) algorithm. Although it is slower than the original
3969 function, it does have some advantages for certain types of matching
3970 problem.
3971
3972 5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
3973 including restarting after a partial match.
3974
3975 6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
3976 defined when compiling for Windows was sent to me. I have put it into the
3977 code, though I have no means of testing or verifying it.
3978
3979 7. Added the pcre_refcount() auxiliary function.
3980
3981 8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
3982 match before or at the first newline in the subject string. In pcretest,
3983 the /f option on a pattern can be used to set this.
3984
3985 9. A repeated \w when used in UTF-8 mode with characters greater than 256
3986 would behave wrongly. This has been present in PCRE since release 4.0.
3987
3988 10. A number of changes to the pcregrep command:
3989
3990 (a) Refactored how -x works; insert ^(...)$ instead of setting
3991 PCRE_ANCHORED and checking the length, in preparation for adding
3992 something similar for -w.
3993
3994 (b) Added the -w (match as a word) option.
3995
3996 (c) Refactored the way lines are read and buffered so as to have more
3997 than one at a time available.
3998
3999 (d) Implemented a pcregrep test script.
4000
4001 (e) Added the -M (multiline match) option. This allows patterns to match
4002 over several lines of the subject. The buffering ensures that at least
4003 8K, or the rest of the document (whichever is the shorter) is available
4004 for matching (and similarly the previous 8K for lookbehind assertions).
4005
4006 (f) Changed the --help output so that it now says
4007
4008 -w, --word-regex(p)
4009
4010 instead of two lines, one with "regex" and the other with "regexp"
4011 because that confused at least one person since the short forms are the
4012 same. (This required a bit of code, as the output is generated
4013 automatically from a table. It wasn't just a text change.)
4014
4015 (g) -- can be used to terminate pcregrep options if the next thing isn't an
4016 option but starts with a hyphen. Could be a pattern or a path name
4017 starting with a hyphen, for instance.
4018
4019 (h) "-" can be given as a file name to represent stdin.
4020
4021 (i) When file names are being printed, "(standard input)" is used for
4022 the standard input, for compatibility with GNU grep. Previously
4023 "<stdin>" was used.
4024
4025 (j) The option --label=xxx can be used to supply a name to be used for
4026 stdin when file names are being printed. There is no short form.
4027
4028 (k) Re-factored the options decoding logic because we are going to add
4029 two more options that take data. Such options can now be given in four
4030 different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
4031
4032 (l) Added the -A, -B, and -C options for requesting that lines of context
4033 around matches be printed.
4034
4035 (m) Added the -L option to print the names of files that do not contain
4036 any matching lines, that is, the complement of -l.
4037
4038 (n) The return code is 2 if any file cannot be opened, but pcregrep does
4039 continue to scan other files.
4040
4041 (o) The -s option was incorrectly implemented. For compatibility with other
4042 greps, it now suppresses the error message for a non-existent or non-
4043 accessible file (but not the return code). There is a new option called
4044 -q that suppresses the output of matching lines, which was what -s was
4045 previously doing.
4046
4047 (p) Added --include and --exclude options to specify files for inclusion
4048 and exclusion when recursing.
4049
4050 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
4051 Hopefully, it now does.
4052
4053 12. Missing cast in pcre_study().
4054
4055 13. Added an "uninstall" target to the makefile.
4056
4057 14. Replaced "extern" in the function prototypes in Makefile.in with
4058 "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
4059 world, but is set differently for Windows.
4060
4061 15. Added a second compiling function called pcre_compile2(). The only
4062 difference is that it has an extra argument, which is a pointer to an
4063 integer error code. When there is a compile-time failure, this is set
4064 non-zero, in addition to the error test pointer being set to point to an
4065 error message. The new argument may be NULL if no error number is required
4066 (but then you may as well call pcre_compile(), which is now just a
4067 wrapper). This facility is provided because some applications need a
4068 numeric error indication, but it has also enabled me to tidy up the way
4069 compile-time errors are handled in the POSIX wrapper.
4070
4071 16. Added VPATH=.libs to the makefile; this should help when building with one
4072 prefix path and installing with another. (Or so I'm told by someone who
4073 knows more about this stuff than I do.)
4074
4075 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
4076 passes PCRE_DOTALL to the pcre_compile() function, making the "." character
4077 match everything, including newlines. This is not POSIX-compatible, but
4078 somebody wanted the feature. From pcretest it can be activated by using
4079 both the P and the s flags.
4080
4081 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
4082
4083 19. libpcre.pc was being incorrectly installed as executable.
4084
4085 20. A couple of places in pcretest check for end-of-line by looking for '\n';
4086 it now also looks for '\r' so that it will work unmodified on Windows.
4087
4088 21. Added Google's contributed C++ wrapper to the distribution.
4089
4090 22. Added some untidy missing memory free() calls in pcretest, to keep
4091 Electric Fence happy when testing.
4092
4093
4094
4095 Version 5.0 13-Sep-04
4096 ---------------------
4097
4098 1. Internal change: literal characters are no longer packed up into items
4099 containing multiple characters in a single byte-string. Each character
4100 is now matched using a separate opcode. However, there may be more than one
4101 byte in the character in UTF-8 mode.
4102
4103 2. The pcre_callout_block structure has two new fields: pattern_position and
4104 next_item_length. These contain the offset in the pattern to the next match
4105 item, and its length, respectively.
4106
4107 3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
4108 insertion of callouts before each pattern item. Added the /C option to
4109 pcretest to make use of this.
4110
4111 4. On the advice of a Windows user, the lines
4112
4113 #if defined(_WIN32) || defined(WIN32)
4114 _setmode( _fileno( stdout ), 0x8000 );
4115 #endif /* defined(_WIN32) || defined(WIN32) */
4116
4117 have been added to the source of pcretest. This apparently does useful
4118 magic in relation to line terminators.
4119
4120 5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
4121 for the benefit of those environments where the "b" makes a difference.
4122
4123 6. The icc compiler has the same options as gcc, but "configure" doesn't seem
4124 to know about it. I have put a hack into configure.in that adds in code
4125 to set GCC=yes if CC=icc. This seems to end up at a point in the
4126 generated configure script that is early enough to affect the setting of
4127 compiler options, which is what is needed, but I have no means of testing
4128 whether it really works. (The user who reported this had patched the
4129 generated configure script, which of course I cannot do.)
4130
4131 LATER: After change 22 below (new libtool files), the configure script
4132 seems to know about icc (and also ecc). Therefore, I have commented out
4133 this hack in configure.in.
4134
4135 7. Added support for pkg-config (2 patches were sent in).
4136
4137 8. Negated POSIX character classes that used a combination of internal tables
4138 were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
4139 [[:^ascii]]. Typically, they would match almost any characters. The other
4140 POSIX classes were not broken in this way.
4141
4142 9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
4143 to find the match, as PCRE was deluded into thinking that the match had to
4144 start at the start point or following a newline. The same bug applied to
4145 patterns with negative forward assertions or any backward assertions
4146 preceding ".*" at the start, unless the pattern required a fixed first
4147 character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
4148
4149 10. In UTF-8 mode, when moving forwards in the subject after a failed match
4150 starting at the last subject character, bytes beyond the end of the subject
4151 string were read.
4152
4153 11. Renamed the variable "class" as "classbits" to make life easier for C++
4154 users. (Previously there was a macro definition, but it apparently wasn't
4155 enough.)
4156
4157 12. Added the new field "tables" to the extra data so that tables can be passed
4158 in at exec time, or the internal tables can be re-selected. This allows
4159 a compiled regex to be saved and re-used at a later time by a different
4160 program that might have everything at different addresses.
4161
4162 13. Modified the pcre-config script so that, when run on Solaris, it shows a
4163 -R library as well as a -L library.
4164
4165 14. The debugging options of pcretest (-d on the command line or D on a
4166 pattern) showed incorrect output for anything following an extended class
4167 that contained multibyte characters and which was followed by a quantifier.
4168
4169 15. Added optional support for general category Unicode character properties
4170 via the \p, \P, and \X escapes. Unicode property support implies UTF-8
4171 support. It adds about 90K to the size of the library. The meanings of the
4172 inbuilt class escapes such as \d and \s have NOT been changed.
4173
4174 16. Updated pcredemo.c to include calls to free() to release the memory for the
4175 compiled pattern.
4176
4177 17. The generated file chartables.c was being created in the source directory
4178 instead of in the building directory. This caused the build to fail if the
4179 source directory was different from the building directory, and was
4180 read-only.
4181
4182 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
4183 file. No doubt somebody will tell me if they don't make sense... Also added
4184 Dan Mooney's comments about building on OpenVMS.
4185
4186 19. Added support for partial matching via the PCRE_PARTIAL option for
4187 pcre_exec() and the \P data escape in pcretest.
4188
4189 20. Extended pcretest with 3 new pattern features:
4190
4191 (i) A pattern option of the form ">rest-of-line" causes pcretest to
4192 write the compiled pattern to the file whose name is "rest-of-line".
4193 This is a straight binary dump of the data, with the saved pointer to
4194 the character tables forced to be NULL. The study data, if any, is
4195 written too. After writing, pcretest reads a new pattern.
4196
4197 (ii) If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
4198 compiled pattern from the given file. There must not be any
4199 occurrences of "<" in the file name (pretty unlikely); if there are,
4200 pcretest will instead treat the initial "<" as a pattern delimiter.
4201 After reading in the pattern, pcretest goes on to read data lines as
4202 usual.
4203
4204 (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
4205 and 16-bit fields in a compiled pattern, to simulate a pattern that
4206 was compiled on a host of opposite endianness.
4207
4208 21. The pcre-exec() function can now cope with patterns that were compiled on
4209 hosts of opposite endianness, with this restriction:
4210
4211 As for any compiled expression that is saved and used later, the tables
4212 pointer field cannot be preserved; the extra_data field in the arguments
4213 to pcre_exec() should be used to pass in a tables address if a value
4214 other than the default internal tables were used at compile time.
4215
4216 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
4217 now diagnosed as an error. Previously, most of the time, a negative number
4218 would have been treated as zero, but if in addition "ovector" was passed as
4219 NULL, a crash could occur.
4220
4221 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
4222 new versions from the libtool 1.5 distribution (the last one is a copy of
4223 a file called libtool.m4). This seems to have fixed the need to patch
4224 "configure" to support Darwin 1.3 (which I used to do). However, I still
4225 had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
4226 workstation).
4227
4228 24. Changed the PCRE licence to be the more standard "BSD" licence.
4229
4230
4231 Version 4.5 01-Dec-03
4232 ---------------------
4233
4234 1. There has been some re-arrangement of the code for the match() function so
4235 that it can be compiled in a version that does not call itself recursively.
4236 Instead, it keeps those local variables that need separate instances for
4237 each "recursion" in a frame on the heap, and gets/frees frames whenever it
4238 needs to "recurse". Keeping track of where control must go is done by means
4239 of setjmp/longjmp. The whole thing is implemented by a set of macros that
4240 hide most of the details from the main code, and operates only if
4241 NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
4242 "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
4243 operating.
4244
4245 To make it easier for callers to provide specially tailored get/free
4246 functions for this usage, two new functions, pcre_stack_malloc, and
4247 pcre_stack_free, are used. They are always called in strict stacking order,
4248 and the size of block requested is always the same.
4249
4250 The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
4251 PCRE has been compiled to use the stack or the heap for recursion. The
4252 -C option of pcretest uses this to show which version is compiled.
4253
4254 A new data escape \S, is added to pcretest; it causes the amounts of store
4255 obtained and freed by both kinds of malloc/free at match time to be added
4256 to the output.
4257
4258 2. Changed the locale test to use "fr_FR" instead of "fr" because that's
4259 what's available on my current Linux desktop machine.
4260
4261 3. When matching a UTF-8 string, the test for a valid string at the start has
4262 been extended. If start_offset is not zero, PCRE now checks that it points
4263 to a byte that is the start of a UTF-8 character. If not, it returns
4264 PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
4265 this is necessary because there may be backward assertions in the pattern.
4266 When matching the same subject several times, it may save resources to use
4267 PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
4268
4269 4. The code for checking the validity of UTF-8 strings has been tightened so
4270 that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
4271 containing "overlong sequences".
4272
4273 5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
4274 I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
4275 should have been "&", but it just so happened that all the cases this let
4276 through by mistake were picked up later in the function.
4277
4278 6. I had used a variable called "isblank" - this is a C99 function, causing
4279 some compilers to warn. To avoid this, I renamed it (as "blankclass").
4280
4281 7. Cosmetic: (a) only output another newline at the end of pcretest if it is
4282 prompting; (b) run "./pcretest /dev/null" at the start of the test script
4283 so the version is shown; (c) stop "make test" echoing "./RunTest".
4284
4285 8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
4286
4287 9. The prototype for memmove() for systems that don't have it was using
4288 size_t, but the inclusion of the header that defines size_t was later. I've
4289 moved the #includes for the C headers earlier to avoid this.
4290
4291 10. Added some adjustments to the code to make it easier to compiler on certain
4292 special systems:
4293
4294 (a) Some "const" qualifiers were missing.
4295 (b) Added the macro EXPORT before all exported functions; by default this
4296 is defined to be empty.
4297 (c) Changed the dftables auxiliary program (that builds chartables.c) so
4298 that it reads its output file name as an argument instead of writing
4299 to the standard output and assuming this can be redirected.
4300
4301 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
4302 class containing characters with values greater than 255, PCRE compilation
4303 went into a loop.
4304
4305 12. A recursive reference to a subpattern that was within another subpattern
4306 that had a minimum quantifier of zero caused PCRE to crash. For example,
4307 (x(y(?2))z)? provoked this bug with a subject that got as far as the
4308 recursion. If the recursively-called subpattern itself had a zero repeat,
4309 that was OK.
4310
4311 13. In pcretest, the buffer for reading a data line was set at 30K, but the
4312 buffer into which it was copied (for escape processing) was still set at
4313 1024, so long lines caused crashes.
4314
4315 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
4316 "internal error: code overflow...". This applied to any character class
4317 that was followed by a possessive quantifier.
4318
4319 15. Modified the Makefile to add libpcre.la as a prerequisite for
4320 libpcreposix.la because I was told this is needed for a parallel build to
4321 work.
4322
4323 16. If a pattern that contained .* following optional items at the start was
4324 studied, the wrong optimizing data was generated, leading to matching
4325 errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
4326 matching string must start with a or b or c. The correct conclusion for
4327 this pattern is that a match can start with any character.
4328
4329
4330 Version 4.4 13-Aug-03
4331 ---------------------
4332
4333 1. In UTF-8 mode, a character class containing characters with values between
4334 127 and 255 was not handled correctly if the compiled pattern was studied.
4335 In fixing this, I have also improved the studying algorithm for such
4336 classes (slightly).
4337
4338 2. Three internal functions had redundant arguments passed to them. Removal
4339 might give a very teeny performance improvement.
4340
4341 3. Documentation bug: the value of the capture_top field in a callout is *one
4342 more than* the number of the hightest numbered captured substring.
4343
4344 4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
4345 in incorrectly linking with a previously installed version. They now link
4346 explicitly with libpcre.la.
4347
4348 5. configure.in no longer needs to recognize Cygwin specially.
4349
4350 6. A problem in pcre.in for Windows platforms is fixed.
4351
4352 7. If a pattern was successfully studied, and the -d (or /D) flag was given to
4353 pcretest, it used to include the size of the study block as part of its
4354 output. Unfortunately, the structure contains a field that has a different
4355 size on different hardware architectures. This meant that the tests that
4356 showed this size failed. As the block is currently always of a fixed size,
4357 this information isn't actually particularly useful in pcretest output, so
4358 I have just removed it.
4359
4360 8. Three pre-processor statements accidentally did not start in column 1.
4361 Sadly, there are *still* compilers around that complain, even though
4362 standard C has not required this for well over a decade. Sigh.
4363
4364 9. In pcretest, the code for checking callouts passed small integers in the
4365 callout_data field, which is a void * field. However, some picky compilers
4366 complained about the casts involved for this on 64-bit systems. Now
4367 pcretest passes the address of the small integer instead, which should get
4368 rid of the warnings.
4369
4370 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
4371 both compile and run time, and gives an error if an invalid UTF-8 sequence
4372 is found. There is a option for disabling this check in cases where the
4373 string is known to be correct and/or the maximum performance is wanted.
4374
4375 11. In response to a bug report, I changed one line in Makefile.in from
4376
4377 -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
4378 to
4379 -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
4380
4381 to look similar to other lines, but I have no way of telling whether this
4382 is the right thing to do, as I do not use Windows. No doubt I'll get told
4383 if it's wrong...
4384
4385
4386 Version 4.3 21-May-03
4387 ---------------------
4388
4389 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
4390 Makefile.
4391
4392 2. Some refactoring to improve the quality of the code:
4393
4394 (i) The utf8_table... variables are now declared "const".
4395
4396 (ii) The code for \cx, which used the "case flipping" table to upper case
4397 lower case letters, now just substracts 32. This is ASCII-specific,
4398 but the whole concept of \cx is ASCII-specific, so it seems
4399 reasonable.
4400
4401 (iii) PCRE was using its character types table to recognize decimal and
4402 hexadecimal digits in the pattern. This is silly, because it handles
4403 only 0-9, a-f, and A-F, but the character types table is locale-
4404 specific, which means strange things might happen. A private
4405 table is now used for this - though it costs 256 bytes, a table is
4406 much faster than multiple explicit tests. Of course, the standard
4407 character types table is still used for matching digits in subject
4408 strings against \d.
4409
4410 (iv) Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
4411 ending in _t are). So I've renamed it as ESC_tee.
4412
4413 3. The first argument for regexec() in the POSIX wrapper should have been
4414 defined as "const".
4415
4416 4. Changed pcretest to use malloc() for its buffers so that they can be
4417 Electric Fenced for debugging.
4418
4419 5. There were several places in the code where, in UTF-8 mode, PCRE would try
4420 to read one or more bytes before the start of the subject string. Often this
4421 had no effect on PCRE's behaviour, but in some circumstances it could
4422 provoke a segmentation fault.
4423
4424 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
4425 to try to read one or more bytes before the start of the subject string.
4426
4427 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
4428 UTF-8 support could misbehave in various ways if the subject string
4429 contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
4430 area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
4431 back over UTF-8 characters.)
4432
4433
4434 Version 4.2 14-Apr-03
4435 ---------------------
4436
4437 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
4438
4439 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
4440 [ON_WINDOWS]: new variable, "#" on non-Windows platforms
4441 [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
4442 [WIN_PREFIX]: new variable, "cyg" for Cygwin
4443 * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
4444 and BUILD_EXEEXT
4445 Note: automatic setting of the BUILD variables is not yet working
4446 set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
4447 compile-time but not at link-time
4448 [LINK]: use for linking executables only
4449 make different versions for Windows and non-Windows
4450 [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
4451 libraries
4452 [LINK_FOR_BUILD]: new variable
4453 [OBJEXT]: use throughout
4454 [EXEEXT]: use throughout
4455 <winshared>: new target
4456 <wininstall>: new target
4457 <dftables.o>: use native compiler
4458 <dftables>: use native linker
4459 <install>: handle Windows platform correctly
4460 <clean>: ditto
4461 <check>: ditto
4462 copy DLL to top builddir before testing
4463
4464 As part of these changes, -no-undefined was removed again. This was reported
4465 to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
4466 in any case.
4467
4468 3. Some tidies to get rid of compiler warnings:
4469
4470 . In the match_data structure, match_limit was an unsigned long int, whereas
4471 match_call_count was an int. I've made them both unsigned long ints.
4472
4473 . In pcretest the fact that a const uschar * doesn't automatically cast to
4474 a void * provoked a warning.
4475
4476 . Turning on some more compiler warnings threw up some "shadow" variables
4477 and a few more missing casts.
4478
4479 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
4480 option, a class that contained a single character with a value between 128
4481 and 255 (e.g. /[\xFF]/) caused PCRE to crash.
4482
4483 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
4484 option, a class that contained several characters, but with at least one
4485 whose value was between 128 and 255 caused PCRE to crash.
4486
4487
4488 Version 4.1 12-Mar-03
4489 ---------------------
4490
4491 1. Compiling with gcc -pedantic found a couple of places where casts were
4492 needed, and a string in dftables.c that was longer than standard compilers are
4493 required to support.
4494
4495 2. Compiling with Sun's compiler found a few more places where the code could
4496 be tidied up in order to avoid warnings.
4497
4498 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
4499 first of these names is deprecated in the latest Autoconf in favour of the name
4500 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
4501 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
4502 analogy I have changed it to CFLAGS_FOR_BUILD.
4503
4504 4. Added -no-undefined to the linking command in the Makefile, because this is
4505 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
4506 linking step for the pcreposix library.
4507
4508 5. PCRE was failing to diagnose the case of two named groups with the same
4509 name.
4510
4511 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
4512 literal character that is needed in the subject for a match, and scans along to
4513 ensure that it is present before embarking on the full matching process. This
4514 saves time in cases of nested unlimited repeats that are never going to match.
4515 Problem: the scan can take a lot of time if the subject is very long (e.g.
4516 megabytes), thus penalizing straightforward matches. It is now done only if the
4517 amount of subject to be scanned is less than 1000 bytes.
4518
4519 7. A lesser problem with the same optimization is that it was recording the
4520 first character of an anchored pattern as "needed", thus provoking a search
4521 right along the subject, even when the first match of the pattern was going to
4522 fail. The "needed" character is now not set for anchored patterns, unless it
4523 follows something in the pattern that is of non-fixed length. Thus, it still
4524 fulfils its original purpose of finding quick non-matches in cases of nested
4525 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
4526
4527
4528 Version 4.0 17-Feb-03
4529 ---------------------
4530
4531 1. If a comment in an extended regex that started immediately after a meta-item
4532 extended to the end of string, PCRE compiled incorrect data. This could lead to
4533 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
4534
4535 2. Moved to autoconf 2.53 and libtool 1.4.2.
4536
4537 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
4538 the special perltest8 script is no longer needed - all the tests can be run
4539 from a single perltest script.
4540
4541 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
4542 by \s. It has now been removed in PCRE. This means it isn't recognized as
4543 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
4544 class [:space:] *does* include VT, thereby creating a mess.
4545
4546 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
4547 space and tab.
4548
4549 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
4550 its new features into the main test script, reducing the number of scripts.
4551
4552 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
4553 were backward compatible, and made the (?i) apply to the whole pattern, as if
4554 /i were given. Now it behaves more logically, and applies the option setting
4555 only to what follows. PCRE has been changed to follow suit. However, if it
4556 finds options settings right at the start of the pattern, it extracts them into
4557 the global options, as before. Thus, they show up in the info data.
4558
4559 8. Added support for the \Q...\E escape sequence. Characters in between are
4560 treated as literals. This is slightly different from Perl in that $ and @ are
4561 also handled as literals inside the quotes. In Perl, they will cause variable
4562 interpolation. Note the following examples:
4563
4564 Pattern PCRE matches Perl matches
4565
4566 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
4567 \Qabc\$xyz\E abc\$xyz abc\$xyz
4568 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
4569
4570 For compatibility with Perl, \Q...\E sequences are recognized inside character
4571 classes as well as outside them.
4572
4573 9. Re-organized 3 code statements in pcretest to avoid "overflow in
4574 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
4575 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
4576 signed/unsigned warnings.
4577
4578 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
4579 option for pcretest, so I've replaced it by a simple function that does just
4580 that job.
4581
4582 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
4583 "pcregrep -".
4584
4585 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
4586 Java package. This provides some syntactic sugar for simple cases of what my
4587 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
4588 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
4589 item, you can use this simplified notation. Note that only makes sense with
4590 greedy quantifiers. Consequently, the use of the possessive quantifier forces
4591 greediness, whatever the setting of the PCRE_UNGREEDY option.
4592
4593 13. A change of greediness default within a pattern was not taking effect at
4594 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
4595 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
4596 was abstracted outside.
4597
4598 14. PCRE now supports the \G assertion. It is true when the current matching
4599 position is at the start point of the match. This differs from \A when the
4600 starting offset is non-zero. Used with the /g option of pcretest (or similar
4601 code), it works in the same way as it does for Perl's /g option. If all
4602 alternatives of a regex begin with \G, the expression is anchored to the start
4603 match position, and the "anchored" flag is set in the compiled expression.
4604
4605 15. Some bugs concerning the handling of certain option changes within patterns
4606 have been fixed. These applied to options other than (?ims). For example,
4607 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
4608 been the other way round. Some of this was related to change 7 above.
4609
4610 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
4611 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
4612 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
4613 POSIX classes only within a class (e.g. /[[:alpha:]]/).
4614
4615 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
4616 mode. Unlike ".", it always matches newline, whatever the setting of
4617 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
4618 assertions. Perl allows it, but it doesn't (in general) work because it can't
4619 calculate the length of the lookbehind. At least, that's the case for Perl
4620 5.8.0 - I've been told they are going to document that it doesn't work in
4621 future.
4622
4623 18. Added an error diagnosis for escapes that PCRE does not support: these are
4624 \L, \l, \N, \P, \p, \U, \u, and \X.
4625
4626 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
4627 reading past the end of the pattern in cases such as /[abcd/.
4628
4629 20. PCRE was getting more memory than necessary for patterns with classes that
4630 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
4631
4632 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
4633 compiling PCRE for use with Virtual Pascal.
4634
4635 22. Small fix to the Makefile to make it work properly if the build is done
4636 outside the source tree.
4637
4638 23. Added a new extension: a condition to go with recursion. If a conditional
4639 subpattern starts with (?(R) the "true" branch is used if recursion has
4640 happened, whereas the "false" branch is used only at the top level.
4641
4642 24. When there was a very long string of literal characters (over 255 bytes
4643 without UTF support, over 250 bytes with UTF support), the computation of how
4644 much memory was required could be incorrect, leading to segfaults or other
4645 strange effects.
4646
4647 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
4648 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
4649 there was a subsequent back reference to those brackets. This meant that, for
4650 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
4651 possible to check for precisely this case. All we can do is abandon the
4652 optimization if .* occurs inside capturing brackets when there are any back
4653 references whatsoever. (See below for a better fix that came later.)
4654
4655 26. The handling of the optimization for finding the first character of a
4656 non-anchored pattern, and for finding a character that is required later in the
4657 match were failing in some cases. This didn't break the matching; it just
4658 failed to optimize when it could. The way this is done has been re-implemented.
4659
4660 27. Fixed typo in error message for invalid (?R item (it said "(?p").
4661
4662 28. Added a new feature that provides some of the functionality that Perl
4663 provides with (?{...}). The facility is termed a "callout". The way it is done
4664 in PCRE is for the caller to provide an optional function, by setting
4665 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
4666 global variable. By default it is unset, which disables all calling out. To get
4667 the function called, the regex must include (?C) at appropriate points. This
4668 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
4669 This provides a means of identifying different callout points. When PCRE
4670 reaches such a point in the regex, if pcre_callout has been set, the external
4671 function is called. It is provided with data in a structure called
4672 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
4673 matching continues; if it returns a non-zero value, the match at the current
4674 point fails. However, backtracking will occur if possible. [This was changed
4675 later and other features added - see item 49 below.]
4676
4677 29. pcretest is upgraded to test the callout functionality. It provides a
4678 callout function that displays information. By default, it shows the start of
4679 the match and the current position in the text. There are some new data escapes
4680 to vary what happens:
4681
4682 \C+ in addition, show current contents of captured substrings
4683 \C- do not supply a callout function
4684 \C!n return 1 when callout number n is reached
4685 \C!n!m return 1 when callout number n is reached for the mth time
4686
4687 30. If pcregrep was called with the -l option and just a single file name, it
4688 output "<stdin>" if a match was found, instead of the file name.
4689
4690 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
4691 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
4692 pcre_exec(). This saves a malloc/free per call. The default value of
4693 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
4694 when configuring.
4695
4696 32. The default maximum size of a compiled pattern is 64K. There have been a
4697 few cases of people hitting this limit. The code now uses macros to handle the
4698 storing of links as offsets within the compiled pattern. It defaults to 2-byte
4699 links, but this can be changed to 3 or 4 bytes by --with-link-size when
4700 configuring. Tests 2 and 5 work only with 2-byte links because they output
4701 debugging information about compiled patterns.
4702
4703 33. Internal code re-arrangements:
4704
4705 (a) Moved the debugging function for printing out a compiled regex into
4706 its own source file (printint.c) and used #include to pull it into
4707 pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
4708 separate copies.
4709
4710 (b) Defined the list of op-code names for debugging as a macro in
4711 internal.h so that it is next to the definition of the opcodes.
4712
4713 (c) Defined a table of op-code lengths for simpler skipping along compiled
4714 code. This is again a macro in internal.h so that it is next to the
4715 definition of the opcodes.
4716
4717 34. Added support for recursive calls to individual subpatterns, along the
4718 lines of Robin Houston's patch (but implemented somewhat differently).
4719
4720 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
4721 allow it to read and process whole directories in Win32. This code was
4722 contributed by Lionel Fourquaux; it has not been tested by me.
4723
4724 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
4725 used to name a group. Names consist of alphanumerics and underscores, and must
4726 be unique. Back references use the syntax (?P=name) and recursive calls use
4727 (?P>name) which is a PCRE extension to the Python extension. Groups still have
4728 numbers. The function pcre_fullinfo() can be used after compilation to extract
4729 a name/number map. There are three relevant calls:
4730
4731 PCRE_INFO_NAMEENTRYSIZE yields the size of each entry in the map
4732 PCRE_INFO_NAMECOUNT yields the number of entries
4733 PCRE_INFO_NAMETABLE yields a pointer to the map.
4734
4735 The map is a vector of fixed-size entries. The size of each entry depends on
4736 the length of the longest name used. The first two bytes of each entry are the
4737 group number, most significant byte first. There follows the corresponding
4738 name, zero terminated. The names are in alphabetical order.
4739
4740 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
4741 case instead of 255. Making it the same both with and without UTF-8 support
4742 means that the same test output works with both.
4743
4744 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
4745 calling malloc() with a zero argument.
4746
4747 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
4748 optimization. I've improved things by keeping a bitmap of backreferences with
4749 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
4750 fact referenced, the optimization can be applied. It is unlikely that a
4751 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
4752 the match to follow \n) will appear inside brackets with a number greater than
4753 31, but if it does, any back reference > 31 suppresses the optimization.
4754
4755 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
4756 of disabling numbered capturing parentheses. Any opening parenthesis that is
4757 not followed by ? behaves as if it were followed by ?: but named parentheses
4758 can still be used for capturing (and they will acquire numbers in the usual
4759 way).
4760
4761 41. Redesigned the return codes from the match() function into yes/no/error so
4762 that errors can be passed back from deep inside the nested calls. A malloc
4763 failure while inside a recursive subpattern call now causes the
4764 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
4765
4766 42. It is now possible to set a limit on the number of times the match()
4767 function is called in a call to pcre_exec(). This facility makes it possible to
4768 limit the amount of recursion and backtracking, though not in a directly
4769 obvious way, because the match() function is used in a number of different
4770 circumstances. The count starts from zero for each position in the subject
4771 string (for non-anchored patterns). The default limit is, for compatibility, a
4772 large number, namely 10 000 000. You can change this in two ways:
4773
4774 (a) When configuring PCRE before making, you can use --with-match-limit=n
4775 to set a default value for the compiled library.
4776
4777 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
4778 a different value is set. See 45 below.
4779
4780 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
4781
4782 43. Added a new function pcre_config(int, void *) to enable run-time extraction
4783 of things that can be changed at compile time. The first argument specifies
4784 what is wanted and the second points to where the information is to be placed.
4785 The current list of available information is:
4786
4787 PCRE_CONFIG_UTF8
4788
4789 The output is an integer that is set to one if UTF-8 support is available;
4790 otherwise it is set to zero.
4791
4792 PCRE_CONFIG_NEWLINE
4793
4794 The output is an integer that it set to the value of the code that is used for
4795 newline. It is either LF (10) or CR (13).
4796
4797 PCRE_CONFIG_LINK_SIZE
4798
4799 The output is an integer that contains the number of bytes used for internal
4800 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
4801
4802 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
4803
4804 The output is an integer that contains the threshold above which the POSIX
4805 interface uses malloc() for output vectors. See item 31 above.
4806
4807 PCRE_CONFIG_MATCH_LIMIT
4808
4809 The output is an unsigned integer that contains the default limit of the number
4810 of match() calls in a pcre_exec() execution. See 42 above.
4811
4812 44. pcretest has been upgraded by the addition of the -C option. This causes it
4813 to extract all the available output from the new pcre_config() function, and to
4814 output it. The program then exits immediately.
4815
4816 45. A need has arisen to pass over additional data with calls to pcre_exec() in
4817 order to support additional features. One way would have been to define
4818 pcre_exec2() (for example) with extra arguments, but this would not have been
4819 extensible, and would also have required all calls to the original function to
4820 be mapped to the new one. Instead, I have chosen to extend the mechanism that
4821 is used for passing in "extra" data from pcre_study().
4822
4823 The pcre_extra structure is now exposed and defined in pcre.h. It currently
4824 contains the following fields:
4825
4826 flags a bitmap indicating which of the following fields are set
4827 study_data opaque data from pcre_study()
4828 match_limit a way of specifying a limit on match() calls for a specific
4829 call to pcre_exec()
4830 callout_data data for callouts (see 49 below)
4831
4832 The flag bits are also defined in pcre.h, and are
4833
4834 PCRE_EXTRA_STUDY_DATA
4835 PCRE_EXTRA_MATCH_LIMIT
4836 PCRE_EXTRA_CALLOUT_DATA
4837
4838 The pcre_study() function now returns one of these new pcre_extra blocks, with
4839 the actual study data pointed to by the study_data field, and the
4840 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
4841 before. That is, this change is entirely upwards-compatible and requires no
4842 change to existing code.
4843
4844 If you want to pass in additional data to pcre_exec(), you can either place it
4845 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
4846 block.
4847
4848 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
4849 data string contains the escape sequence \M, pcretest calls pcre_exec() several
4850 times with different match limits, until it finds the minimum value needed for
4851 pcre_exec() to complete. The value is then output. This can be instructive; for
4852 most simple matches the number is quite small, but for pathological cases it
4853 gets very large very quickly.
4854
4855 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
4856 returns the size of the data block pointed to by the study_data field in a
4857 pcre_extra block, that is, the value that was passed as the argument to
4858 pcre_malloc() when PCRE was getting memory in which to place the information
4859 created by pcre_study(). The fourth argument should point to a size_t variable.
4860 pcretest has been extended so that this information is shown after a successful
4861 pcre_study() call when information about the compiled regex is being displayed.
4862
4863 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
4864 because what follows is always an absolute path. (Later: it turns out that this
4865 is more than cosmetic for MinGW, because it doesn't like empty path
4866 components.)
4867
4868 49. Some changes have been made to the callout feature (see 28 above):
4869
4870 (i) A callout function now has three choices for what it returns:
4871
4872 0 => success, carry on matching
4873 > 0 => failure at this point, but backtrack if possible
4874 < 0 => serious error, return this value from pcre_exec()
4875
4876 Negative values should normally be chosen from the set of PCRE_ERROR_xxx
4877 values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
4878 "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
4879 use by callout functions. It will never be used by PCRE itself.
4880
4881 (ii) The pcre_extra structure (see 45 above) has a void * field called
4882 callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
4883 pcre_callout_block structure has a field of the same name. The contents of
4884 the field passed in the pcre_extra structure are passed to the callout
4885 function in the corresponding field in the callout block. This makes it
4886 easier to use the same callout-containing regex from multiple threads. For
4887 testing, the pcretest program has a new data escape
4888
4889 \C*n pass the number n (may be negative) as callout_data
4890
4891 If the callout function in pcretest receives a non-zero value as
4892 callout_data, it returns that value.
4893
4894 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
4895 there were some redundant $(CFLAGS) in commands that are now specified as
4896 $(LINK), which already includes $(CFLAGS).
4897
4898 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
4899 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
4900 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
4901 one-byte characters throughout. Note that case-insensitive matching applies
4902 only to characters whose values are less than 256. PCRE doesn't support the
4903 notion of cases for higher-valued characters.
4904
4905 (i) A character class whose characters are all within 0-255 is handled as
4906 a bit map, and the map is inverted for negative classes. Previously, a
4907 character > 255 always failed to match such a class; however it should
4908 match if the class was a negative one (e.g. [^ab]). This has been fixed.
4909
4910 (ii) A negated character class with a single character < 255 is coded as
4911 "not this character" (OP_NOT). This wasn't working properly when the test
4912 character was multibyte, either singly or repeated.
4913
4914 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
4915 mode, for example: \x{100}{2,3}.
4916
4917 (iv) The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
4918 singly or repeated) now correctly test multibyte characters. However,
4919 PCRE doesn't recognize any characters with values greater than 255 as
4920 digits, spaces, or word characters. Such characters always match \D, \S,
4921 and \W, and never match \d, \s, or \w.
4922
4923 (v) Classes may now contain characters and character ranges with values
4924 greater than 255. For example: [ab\x{100}-\x{400}].
4925
4926 (vi) pcregrep now has a --utf-8 option (synonym -u) which makes it call
4927 PCRE in UTF-8 mode.
4928
4929 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
4930 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
4931 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
4932 value.)
4933
4934 53. The single man page has become too large. I have therefore split it up into
4935 a number of separate man pages. These also give rise to individual HTML pages;
4936 these are now put in a separate directory, and there is an index.html page that
4937 lists them all. Some hyperlinking between the pages has been installed.
4938
4939 54. Added convenience functions for handling named capturing parentheses.
4940
4941 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
4942 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
4943 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
4944 are faulted.
4945
4946 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
4947 calling configure. These values are used when compiling the dftables.c program
4948 which is run to generate the source of the default character tables. They
4949 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
4950 you will need to set these values.
4951
4952 57. Updated the building process for Windows DLL, as provided by Fred Cox.
4953
4954
4955 Version 3.9 02-Jan-02
4956 ---------------------
4957
4958 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
4959
4960 2. If --disable-static was given, the building process failed when trying to
4961 build pcretest and pcregrep. (For some reason it was using libtool to compile
4962 them, which is not right, as they aren't part of the library.)
4963
4964
4965 Version 3.8 18-Dec-01
4966 ---------------------
4967
4968 1. The experimental UTF-8 code was completely screwed up. It was packing the
4969 bytes in the wrong order. How dumb can you get?
4970
4971
4972 Version 3.7 29-Oct-01
4973 ---------------------
4974
4975 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
4976 This caused pcretest, when used on the test data, to segfault. Unfortunately,
4977 this didn't happen under Solaris 8, where I normally test things.
4978
4979 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
4980 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
4981 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
4982 this fix an hour or so after the initial 3.7 release.)
4983
4984
4985 Version 3.6 23-Oct-01
4986 ---------------------
4987
4988 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
4989 offsets passed as NULL with zero offset count.
4990
4991 2. The config.guess and config.sub files had not been updated when I moved to
4992 the latest autoconf.
4993
4994
4995 Version 3.5 15-Aug-01
4996 ---------------------
4997
4998 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
4999 had been forgotten.
5000
5001 2. By using declared but undefined structures, we can avoid using "void"
5002 definitions in pcre.h while keeping the internal definitions of the structures
5003 private.
5004
5005 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
5006 user point of view, this means that both static and shared libraries are built
5007 by default, but this can be individually controlled. More of the work of
5008 handling this static/shared cases is now inside libtool instead of PCRE's make
5009 file.
5010
5011 4. The pcretest utility is now installed along with pcregrep because it is
5012 useful for users (to test regexs) and by doing this, it automatically gets
5013 relinked by libtool. The documentation has been turned into a man page, so
5014 there are now .1, .txt, and .html versions in /doc.
5015
5016 5. Upgrades to pcregrep:
5017 (i) Added long-form option names like gnu grep.
5018 (ii) Added --help to list all options with an explanatory phrase.
5019 (iii) Added -r, --recursive to recurse into sub-directories.
5020 (iv) Added -f, --file to read patterns from a file.
5021
5022 6. pcre_exec() was referring to its "code" argument before testing that
5023 argument for NULL (and giving an error if it was NULL).
5024
5025 7. Upgraded Makefile.in to allow for compiling in a different directory from
5026 the source directory.
5027
5028 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
5029 options bits, the pointer it was passed was to an int instead of to an unsigned
5030 long int. This mattered only on 64-bit systems.
5031
5032 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
5033 generated) instead of pcre.in, which it its source. Also made the same change
5034 in several of the .c files.
5035
5036 10. A new release of gcc defines printf() as a macro, which broke pcretest
5037 because it had an ifdef in the middle of a string argument for printf(). Fixed
5038 by using separate calls to printf().
5039
5040 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
5041 script, to force use of CR or LF instead of \n in the source. On non-Unix
5042 systems, the value can be set in config.h.
5043
5044 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
5045 absolute limit. Changed the text of the error message to make this clear, and
5046 likewise updated the man page.
5047
5048 13. The limit of 99 on the number of capturing subpatterns has been removed.
5049 The new limit is 65535, which I hope will not be a "real" limit.
5050
5051
5052 Version 3.4 22-Aug-00
5053 ---------------------
5054
5055 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
5056
5057 2. Diagnose condition (?(0) as an error instead of crashing on matching.
5058
5059
5060 Version 3.3 01-Aug-00
5061 ---------------------
5062
5063 1. If an octal character was given, but the value was greater than \377, it
5064 was not getting masked to the least significant bits, as documented. This could
5065 lead to crashes in some systems.
5066
5067 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
5068 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
5069
5070 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
5071 These just pass their arguments on to (pcre_free)(), but they are provided
5072 because some uses of PCRE bind it to non-C systems that can call its functions,
5073 but cannot call free() or pcre_free() directly.
5074
5075 4. Add "make test" as a synonym for "make check". Corrected some comments in
5076 the Makefile.
5077
5078 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
5079 Makefile.
5080
5081 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
5082 command called pgrep for grepping around the active processes.
5083
5084 7. Added the beginnings of support for UTF-8 character strings.
5085
5086 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
5087 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
5088 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
5089 out for the ar command.)
5090
5091
5092 Version 3.2 12-May-00
5093 ---------------------
5094
5095 This is purely a bug fixing release.
5096
5097 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
5098 of ZA. This was just one example of several cases that could provoke this bug,
5099 which was introduced by change 9 of version 2.00. The code for breaking
5100 infinite loops after an iteration that matches an empty string was't working
5101 correctly.
5102
5103 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
5104 when matched against abbab (for example). After matching an empty string, it
5105 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
5106 caused it to match further down the string than it should.
5107
5108 3. The code contained an inclusion of sys/types.h. It isn't clear why this
5109 was there because it doesn't seem to be needed, and it causes trouble on some
5110 systems, as it is not a Standard C header. It has been removed.
5111
5112 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
5113 were reported on the Macintosh. The changes were from
5114
5115 while ((c = *(++ptr)) != 0 && c != '\n');
5116 to
5117 while ((c = *(++ptr)) != 0 && c != '\n') ;
5118
5119 Totally extraordinary, but if that's what it takes...
5120
5121 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
5122 available. Added HAVE_BCOPY and an autoconf test for it; if neither
5123 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
5124 assumes the way PCRE uses memmove() (always moving upwards).
5125
5126 6. PCRE is being used in one environment where strchr() is not available. There
5127 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
5128 faster code anyway.
5129
5130
5131 Version 3.1 09-Feb-00
5132 ---------------------
5133
5134 The only change in this release is the fixing of some bugs in Makefile.in for
5135 the "install" target:
5136
5137 (1) It was failing to install pcreposix.h.
5138
5139 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
5140
5141
5142 Version 3.0 01-Feb-00
5143 ---------------------
5144
5145 1. Add support for the /+ modifier to perltest (to output $` like it does in
5146 pcretest).
5147
5148 2. Add support for the /g modifier to perltest.
5149
5150 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
5151 matches null strings.
5152
5153 4. Fix perltest so that it doesn't do unwanted things when fed an empty
5154 pattern. Perl treats empty patterns specially - it reuses the most recent
5155 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
5156 effect.
5157
5158 5. The POSIX interface was broken in that it was just handing over the POSIX
5159 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
5160 required a bigger vector, with some working space on the end. This means that
5161 the POSIX wrapper now has to get and free some memory, and copy the results.
5162
5163 6. Added some simple autoconf support, placing the test data and the
5164 documentation in separate directories, re-organizing some of the
5165 information files, and making it build pcre-config (a GNU standard). Also added
5166 libtool support for building PCRE as a shared library, which is now the
5167 default.
5168
5169 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
5170 09 are not valid octal constants. Single digits will be used for minor values
5171 less than 10.
5172
5173 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
5174 existing programs that set these in the POSIX interface can use PCRE without
5175 modification.
5176
5177 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
5178 return all that pcre_info() returns, plus additional data. The pcre_info()
5179 function is retained for compatibility, but is considered to be obsolete.
5180
5181 10. Added experimental recursion feature (?R) to handle one common case that
5182 Perl 5.6 will be able to do with (?p{...}).
5183
5184 11. Added support for POSIX character classes like [:alpha:], which Perl is
5185 adopting.
5186
5187
5188 Version 2.08 31-Aug-99
5189 ----------------------
5190
5191 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
5192 trying to match at the startoffset position, but instead was moving forward to
5193 the next newline as if a previous match had failed.
5194
5195 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
5196 and could get into a loop if a null string was matched other than at the start
5197 of the subject.
5198
5199 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
5200 be distinguished at compile time, and for completeness also added PCRE_DATE.
5201
5202 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
5203 in GnuWin32 environments.
5204
5205
5206 Version 2.07 29-Jul-99
5207 ----------------------
5208
5209 1. The documentation is now supplied in plain text form and HTML as well as in
5210 the form of man page sources.
5211
5212 2. C++ compilers don't like assigning (void *) values to other pointer types.
5213 In particular this affects malloc(). Although there is no problem in Standard
5214 C, I've put in casts to keep C++ compilers happy.
5215
5216 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
5217 should be (const char *).
5218
5219 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
5220 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
5221 However, I haven't made this a standard facility. The documentation doesn't
5222 mention it, and the Makefile doesn't support it.
5223
5224 5. The Makefile now contains an "install" target, with editable destinations at
5225 the top of the file. The pcretest program is not installed.
5226
5227 6. pgrep -V now gives the PCRE version number and date.
5228
5229 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
5230 causing the entire string to be ignored, instead of just the last character.
5231
5232 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
5233 non-matching string, it can take a very, very long time, even for strings of
5234 quite modest length, because of the nested recursion. PCRE now does better in
5235 some of these cases. It does this by remembering the last required literal
5236 character in the pattern, and pre-searching the subject to ensure it is present
5237 before running the real match. In other words, it applies a heuristic to detect
5238 some types of certain failure quickly, and in the above example, if presented
5239 with a string that has no trailing " it gives "no match" very quickly.
5240
5241 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
5242 other alternatives are tried instead.
5243
5244
5245 Version 2.06 09-Jun-99
5246 ----------------------
5247
5248 1. Change pcretest's output for amount of store used to show just the code
5249 space, because the remainder (the data block) varies in size between 32-bit and
5250 64-bit systems.
5251
5252 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
5253 start matching at. This allows lookbehinds to work when searching for multiple
5254 occurrences in a string.
5255
5256 3. Added additional options to pcretest for testing multiple occurrences:
5257
5258 /+ outputs the rest of the string that follows a match
5259 /g loops for multiple occurrences, using the new startoffset argument
5260 /G loops for multiple occurrences by passing an incremented pointer
5261
5262 4. PCRE wasn't doing the "first character" optimization for patterns starting
5263 with \b or \B, though it was doing it for other lookbehind assertions. That is,
5264 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
5265 the letter 'x'. On long subject strings, this gives a significant speed-up.
5266
5267
5268 Version 2.05 21-Apr-99
5269 ----------------------
5270
5271 1. Changed the type of magic_number from int to long int so that it works
5272 properly on 16-bit systems.
5273
5274 2. Fixed a bug which caused patterns starting with .* not to work correctly
5275 when the subject string contained newline characters. PCRE was assuming
5276 anchoring for such patterns in all cases, which is not correct because .* will
5277 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
5278 DOTALL is set at top level; otherwise it knows that patterns starting with .*
5279 must be retried after every newline in the subject.
5280
5281
5282 Version 2.04 18-Feb-99
5283 ----------------------
5284
5285 1. For parenthesized subpatterns with repeats whose minimum was zero, the
5286 computation of the store needed to hold the pattern was incorrect (too large).
5287 If such patterns were nested a few deep, this could multiply and become a real
5288 problem.
5289
5290 2. Added /M option to pcretest to show the memory requirement of a specific
5291 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
5292
5293 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
5294 compiled in such a way that the backtracking after subsequent failure was
5295 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
5296 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
5297
5298
5299 Version 2.03 02-Feb-99
5300 ----------------------
5301
5302 1. Fixed typo and small mistake in man page.
5303
5304 2. Added 4th condition (GPL supersedes if conflict) and created separate
5305 LICENCE file containing the conditions.
5306
5307 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
5308 Perl, that is the internal \ allows the delimiter to be included in the
5309 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
5310 the final delimiter, add \ to the end of the pattern (to test the error).
5311
5312 4. Added the convenience functions for extracting substrings after a successful
5313 match. Updated pcretest to make it able to test these functions.
5314
5315
5316 Version 2.02 14-Jan-99
5317 ----------------------
5318
5319 1. Initialized the working variables associated with each extraction so that
5320 their saving and restoring doesn't refer to uninitialized store.
5321
5322 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
5323 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
5324 fix the problem.
5325
5326 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
5327 calls, and wasn't printing the correct value for compiling calls. Increased the
5328 default value of LOOPREPEAT, and the number of significant figures in the
5329 times.
5330
5331 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
5332
5333 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
5334 a building problem on Windows NT with a FAT file system.
5335
5336
5337 Version 2.01 21-Oct-98
5338 ----------------------
5339
5340 1. Changed the API for pcre_compile() to allow for the provision of a pointer
5341 to character tables built by pcre_maketables() in the current locale. If NULL
5342 is passed, the default tables are used.
5343
5344
5345 Version 2.00 24-Sep-98
5346 ----------------------
5347
5348 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
5349 it any more.
5350
5351 2. Allow quantification of (?>) groups, and make it work correctly.
5352
5353 3. The first character computation wasn't working for (?>) groups.
5354
5355 4. Correct the implementation of \Z (it is permitted to match on the \n at the
5356 end of the subject) and add 5.005's \z, which really does match only at the
5357 very end of the subject.
5358
5359 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
5360
5361 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
5362 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
5363 localized options. All options to pcre_study() were also removed.
5364
5365 7. Add other new features from 5.005:
5366
5367 $(?<= positive lookbehind
5368 $(?<! negative lookbehind
5369 (?imsx-imsx) added the unsetting capability
5370 such a setting is global if at outer level; local otherwise
5371 (?imsx-imsx:) non-capturing groups with option setting
5372 (?(cond)re|re) conditional pattern matching
5373
5374 A backreference to itself in a repeated group matches the previous
5375 captured string.
5376
5377 8. General tidying up of studying (both automatic and via "study")
5378 consequential on the addition of new assertions.
5379
5380 9. As in 5.005, unlimited repeated groups that could match an empty substring
5381 are no longer faulted at compile time. Instead, the loop is forcibly broken at
5382 runtime if any iteration does actually match an empty substring.
5383
5384 10. Include the RunTest script in the distribution.
5385
5386 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
5387 discrepancies, some of which were old and were also with respect to 5.004. They
5388 have now been fixed.
5389
5390
5391 Version 1.09 28-Apr-98
5392 ----------------------
5393
5394 1. A negated single character class followed by a quantifier with a minimum
5395 value of one (e.g. [^x]{1,6} ) was not compiled correctly. This could lead to
5396 program crashes, or just wrong answers. This did not apply to negated classes
5397 containing more than one character, or to minima other than one.
5398
5399
5400 Version 1.08 27-Mar-98
5401 ----------------------
5402
5403 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
5404
5405 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
5406 latter must appear before anything that relies on it in the pattern.
5407
5408
5409 Version 1.07 16-Feb-98
5410 ----------------------
5411
5412 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
5413 repeat of a potentially empty string).
5414
5415
5416 Version 1.06 23-Jan-98
5417 ----------------------
5418
5419 1. Added Markus Oberhumer's little patches for C++.
5420
5421 2. Literal strings longer than 255 characters were broken.
5422
5423
5424 Version 1.05 23-Dec-97
5425 ----------------------
5426
5427 1. Negated character classes containing more than one character were failing if
5428 PCRE_CASELESS was set at run time.
5429
5430
5431 Version 1.04 19-Dec-97
5432 ----------------------
5433
5434 1. Corrected the man page, where some "const" qualifiers had been omitted.
5435
5436 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
5437 input syntax.
5438
5439 3. Fixed memory leak which occurred when a regex with back references was
5440 matched with an offsets vector that wasn't big enough. The temporary memory
5441 that is used in this case wasn't being freed if the match failed.
5442
5443 4. Tidied pcretest to ensure it frees memory that it gets.
5444
5445 5. Temporary memory was being obtained in the case where the passed offsets
5446 vector was exactly big enough.
5447
5448 6. Corrected definition of offsetof() from change 5 below.
5449
5450 7. I had screwed up change 6 below and broken the rules for the use of
5451 setjmp(). Now fixed.
5452
5453
5454 Version 1.03 18-Dec-97
5455 ----------------------
5456
5457 1. A erroneous regex with a missing opening parenthesis was correctly
5458 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
5459 on some systems.
5460
5461 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
5462 it was reported that one broken compiler failed on the former because "code" is
5463 also an independent variable.
5464
5465 3. The erroneous regex a[]b caused an array overrun reference.
5466
5467 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
5468 fail on data ending with that character. (It was going on too far, and checking
5469 the next character, typically a binary zero.) This was specific to the
5470 optimized code for single-character negative classes.
5471
5472 5. Added a contributed patch from the TIN world which does the following:
5473
5474 + Add an undef for memmove, in case the the system defines a macro for it.
5475
5476 + Add a definition of offsetof(), in case there isn't one. (I don't know
5477 the reason behind this - offsetof() is part of the ANSI standard - but
5478 it does no harm).
5479
5480 + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
5481 most of the places where whitespace preceded '#'. I have given up and
5482 allowed the remaining 2 cases to be at the margin.
5483
5484 + Rename some variables in pcre to eliminate shadowing. This seems very
5485 pedantic, but does no harm, of course.
5486
5487 6. Moved the call to setjmp() into its own function, to get rid of warnings
5488 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
5489
5490 7. Constructs such as \d{8,} were compiling into the equivalent of
5491 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
5492 outcome, but in this particular case used more store than had been allocated,
5493 which caused the bug to be discovered because it threw up an internal error.
5494
5495 8. The debugging code in both pcre and pcretest for outputting the compiled
5496 form of a regex was going wrong in the case of back references followed by
5497 curly-bracketed repeats.
5498
5499
5500 Version 1.02 12-Dec-97
5501 ----------------------
5502
5503 1. Typos in pcre.3 and comments in the source fixed.
5504
5505 2. Applied a contributed patch to get rid of places where it used to remove
5506 'const' from variables, and fixed some signed/unsigned and uninitialized
5507 variable warnings.
5508
5509 3. Added the "runtest" target to Makefile.
5510
5511 4. Set default compiler flag to -O2 rather than just -O.
5512
5513
5514 Version 1.01 19-Nov-97
5515 ----------------------
5516
5517 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
5518 like /([ab]*)*/, that is, for classes with more than one character in them.
5519
5520 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
5521 as /((?>a*))*/ (a PCRE_EXTRA facility).
5522
5523
5524 Version 1.00 18-Nov-97
5525 ----------------------
5526
5527 1. Added compile-time macros to support systems such as SunOS4 which don't have
5528 memmove() or strerror() but have other things that can be used instead.
5529
5530 2. Arranged that "make clean" removes the executables.
5531
5532
5533 Version 0.99 27-Oct-97
5534 ----------------------
5535
5536 1. Fixed bug in code for optimizing classes with only one character. It was
5537 initializing a 32-byte map regardless, which could cause it to run off the end
5538 of the memory it had got.
5539
5540 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
5541
5542
5543 Version 0.98 22-Oct-97
5544 ----------------------
5545
5546 1. Fixed bug in code for handling temporary memory usage when there are more
5547 back references than supplied space in the ovector. This could cause segfaults.
5548
5549
5550 Version 0.97 21-Oct-97
5551 ----------------------
5552
5553 1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
5554
5555 2. Optimized negated single characters not to use a bit map.
5556
5557 3. Brought error texts together as macro definitions; clarified some of them;
5558 fixed one that was wrong - it said "range out of order" when it meant "invalid
5559 escape sequence".
5560
5561 4. Changed some char * arguments to const char *.
5562
5563 5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
5564
5565 6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
5566 pcretest.
5567
5568
5569 Version 0.96 16-Oct-97
5570 ----------------------
5571
5572 1. Added a simple "pgrep" utility to the distribution.
5573
5574 2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
5575 unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
5576 where "ddd" means "one or more decimal digits".
5577
5578 3. Fixed serious bug. If a pattern had a back reference, but the call to
5579 pcre_exec() didn't supply a large enough ovector to record the related
5580 identifying subpattern, the match always failed. PCRE now remembers the number
5581 of the largest back reference, and gets some temporary memory in which to save
5582 the offsets during matching if necessary, in order to ensure that
5583 backreferences always work.
5584
5585 4. Increased the compatibility with Perl in a number of ways:
5586
5587 (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
5588 to request this handling. The option can be set at compile or exec time.
5589
5590 (b) $ matches before a terminating newline by default; an option
5591 PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
5592 mode). The option can be set at compile or exec time.
5593
5594 (c) The handling of \ followed by a digit other than 0 is now supposed to be
5595 the same as Perl's. If the decimal number it represents is less than 10
5596 or there aren't that many previous left capturing parentheses, an octal
5597 escape is read. Inside a character class, it's always an octal escape,
5598 even if it is a single digit.
5599
5600 (d) An escaped but undefined alphabetic character is taken as a literal,
5601 unless PCRE_EXTRA is set. Currently this just reserves the remaining
5602 escapes.
5603
5604 (e) {0} is now permitted. (The previous item is removed from the compiled
5605 pattern).
5606
5607 5. Changed all the names of code files so that the basic parts are no longer
5608 than 10 characters, and abolished the teeny "globals.c" file.
5609
5610 6. Changed the handling of character classes; they are now done with a 32-byte
5611 bit map always.
5612
5613 7. Added the -d and /D options to pcretest to make it possible to look at the
5614 internals of compilation without having to recompile pcre.
5615
5616
5617 Version 0.95 23-Sep-97
5618 ----------------------
5619
5620 1. Fixed bug in pre-pass concerning escaped "normal" characters such as \x5c or
5621 \x20 at the start of a run of normal characters. These were being treated as
5622 real characters, instead of the source characters being re-checked.
5623
5624
5625 Version 0.94 18-Sep-97
5626 ----------------------
5627
5628 1. The functions are now thread-safe, with the caveat that the global variables
5629 containing pointers to malloc() and free() or alternative functions are the
5630 same for all threads.
5631
5632 2. Get pcre_study() to generate a bitmap of initial characters for non-
5633 anchored patterns when this is possible, and use it if passed to pcre_exec().
5634
5635
5636 Version 0.93 15-Sep-97
5637 ----------------------
5638
5639 1. /(b)|(:+)/ was computing an incorrect first character.
5640
5641 2. Add pcre_study() to the API and the passing of pcre_extra to pcre_exec(),
5642 but not actually doing anything yet.
5643
5644 3. Treat "-" characters in classes that cannot be part of ranges as literals,
5645 as Perl does (e.g. [-az] or [az-]).
5646
5647 4. Set the anchored flag if a branch starts with .* or .*? because that tests
5648 all possible positions.
5649
5650 5. Split up into different modules to avoid including unneeded functions in a
5651 compiled binary. However, compile and exec are still in one module. The "study"
5652 function is split off.
5653
5654 6. The character tables are now in a separate module whose source is generated
5655 by an auxiliary program - but can then be edited by hand if required. There are
5656 now no calls to isalnum(), isspace(), isdigit(), isxdigit(), tolower() or
5657 toupper() in the code.
5658
5659 7. Turn the malloc/free funtions variables into pcre_malloc and pcre_free and
5660 make them global. Abolish the function for setting them, as the caller can now
5661 set them directly.
5662
5663
5664 Version 0.92 11-Sep-97
5665 ----------------------
5666
5667 1. A repeat with a fixed maximum and a minimum of 1 for an ordinary character
5668 (e.g. /a{1,3}/) was broken (I mis-optimized it).
5669
5670 2. Caseless matching was not working in character classes if the characters in
5671 the pattern were in upper case.
5672
5673 3. Make ranges like [W-c] work in the same way as Perl for caseless matching.
5674
5675 4. Make PCRE_ANCHORED public and accept as a compile option.
5676
5677 5. Add an options word to pcre_exec() and accept PCRE_ANCHORED and
5678 PCRE_CASELESS at run time. Add escapes \A and \I to pcretest to cause it to
5679 pass them.
5680
5681 6. Give an error if bad option bits passed at compile or run time.
5682
5683 7. Add PCRE_MULTILINE at compile and exec time, and (?m) as well. Add \M to
5684 pcretest to cause it to pass that flag.
5685
5686 8. Add pcre_info(), to get the number of identifying subpatterns, the stored
5687 options, and the first character, if set.
5688
5689 9. Recognize C+ or C{n,m} where n >= 1 as providing a fixed starting character.
5690
5691
5692 Version 0.91 10-Sep-97
5693 ----------------------
5694
5695 1. PCRE was failing to diagnose unlimited repeats of subpatterns that could
5696 match the empty string as in /(a*)*/. It was looping and ultimately crashing.
5697
5698 2. PCRE was looping on encountering an indefinitely repeated back reference to
5699 a subpattern that had matched an empty string, e.g. /(a|)\1*/. It now does what
5700 Perl does - treats the match as successful.
5701
5702 ****

Properties

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

  ViewVC Help
Powered by ViewVC 1.1.5