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

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1536 - (show annotations)
Fri Mar 27 16:44:50 2015 UTC (4 years, 8 months ago) by ph10
File size: 274988 byte(s)
Fix memory bug for [\S\V\H] compile.
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
127 Version 8.36 26-September-2014
128 ------------------------------
129
130 1. Got rid of some compiler warnings in the C++ modules that were shown up by
131 -Wmissing-field-initializers and -Wunused-parameter.
132
133 2. The tests for quantifiers being too big (greater than 65535) were being
134 applied after reading the number, and stupidly assuming that integer
135 overflow would give a negative number. The tests are now applied as the
136 numbers are read.
137
138 3. Tidy code in pcre_exec.c where two branches that used to be different are
139 now the same.
140
141 4. The JIT compiler did not generate match limit checks for certain
142 bracketed expressions with quantifiers. This may lead to exponential
143 backtracking, instead of returning with PCRE_ERROR_MATCHLIMIT. This
144 issue should be resolved now.
145
146 5. Fixed an issue, which occures when nested alternatives are optimized
147 with table jumps.
148
149 6. Inserted two casts and changed some ints to size_t in the light of some
150 reported 64-bit compiler warnings (Bugzilla 1477).
151
152 7. Fixed a bug concerned with zero-minimum possessive groups that could match
153 an empty string, which sometimes were behaving incorrectly in the
154 interpreter (though correctly in the JIT matcher). This pcretest input is
155 an example:
156
157 '\A(?:[^"]++|"(?:[^"]*+|"")*+")++'
158 NON QUOTED "QUOT""ED" AFTER "NOT MATCHED
159
160 the interpreter was reporting a match of 'NON QUOTED ' only, whereas the
161 JIT matcher and Perl both matched 'NON QUOTED "QUOT""ED" AFTER '. The test
162 for an empty string was breaking the inner loop and carrying on at a lower
163 level, when possessive repeated groups should always return to a higher
164 level as they have no backtrack points in them. The empty string test now
165 occurs at the outer level.
166
167 8. Fixed a bug that was incorrectly auto-possessifying \w+ in the pattern
168 ^\w+(?>\s*)(?<=\w) which caused it not to match "test test".
169
170 9. Give a compile-time error for \o{} (as Perl does) and for \x{} (which Perl
171 doesn't).
172
173 10. Change 8.34/15 introduced a bug that caused the amount of memory needed
174 to hold a pattern to be incorrectly computed (too small) when there were
175 named back references to duplicated names. This could cause "internal
176 error: code overflow" or "double free or corruption" or other memory
177 handling errors.
178
179 11. When named subpatterns had the same prefixes, back references could be
180 confused. For example, in this pattern:
181
182 /(?P<Name>a)?(?P<Name2>b)?(?(<Name>)c|d)*l/
183
184 the reference to 'Name' was incorrectly treated as a reference to a
185 duplicate name.
186
187 12. A pattern such as /^s?c/mi8 where the optional character has more than
188 one "other case" was incorrectly compiled such that it would only try to
189 match starting at "c".
190
191 13. When a pattern starting with \s was studied, VT was not included in the
192 list of possible starting characters; this should have been part of the
193 8.34/18 patch.
194
195 14. If a character class started [\Qx]... where x is any character, the class
196 was incorrectly terminated at the ].
197
198 15. If a pattern that started with a caseless match for a character with more
199 than one "other case" was studied, PCRE did not set up the starting code
200 unit bit map for the list of possible characters. Now it does. This is an
201 optimization improvement, not a bug fix.
202
203 16. The Unicode data tables have been updated to Unicode 7.0.0.
204
205 17. Fixed a number of memory leaks in pcregrep.
206
207 18. Avoid a compiler warning (from some compilers) for a function call with
208 a cast that removes "const" from an lvalue by using an intermediate
209 variable (to which the compiler does not object).
210
211 19. Incorrect code was compiled if a group that contained an internal recursive
212 back reference was optional (had quantifier with a minimum of zero). This
213 example compiled incorrect code: /(((a\2)|(a*)\g<-1>))*/ and other examples
214 caused segmentation faults because of stack overflows at compile time.
215
216 20. A pattern such as /((?(R)a|(?1)))+/, which contains a recursion within a
217 group that is quantified with an indefinite repeat, caused a compile-time
218 loop which used up all the system stack and provoked a segmentation fault.
219 This was not the same bug as 19 above.
220
221 21. Add PCRECPP_EXP_DECL declaration to operator<< in pcre_stringpiece.h.
222 Patch by Mike Frysinger.
223
224
225 Version 8.35 04-April-2014
226 --------------------------
227
228 1. A new flag is set, when property checks are present in an XCLASS.
229 When this flag is not set, PCRE can perform certain optimizations
230 such as studying these XCLASS-es.
231
232 2. The auto-possessification of character sets were improved: a normal
233 and an extended character set can be compared now. Furthermore
234 the JIT compiler optimizes more character set checks.
235
236 3. Got rid of some compiler warnings for potentially uninitialized variables
237 that show up only when compiled with -O2.
238
239 4. A pattern such as (?=ab\K) that uses \K in an assertion can set the start
240 of a match later then the end of the match. The pcretest program was not
241 handling the case sensibly - it was outputting from the start to the next
242 binary zero. It now reports this situation in a message, and outputs the
243 text from the end to the start.
244
245 5. Fast forward search is improved in JIT. Instead of the first three
246 characters, any three characters with fixed position can be searched.
247 Search order: first, last, middle.
248
249 6. Improve character range checks in JIT. Characters are read by an inprecise
250 function now, which returns with an unknown value if the character code is
251 above a certain threshold (e.g: 256). The only limitation is that the value
252 must be bigger than the threshold as well. This function is useful when
253 the characters above the threshold are handled in the same way.
254
255 7. The macros whose names start with RAWUCHAR are placeholders for a future
256 mode in which only the bottom 21 bits of 32-bit data items are used. To
257 make this more memorable for those maintaining the code, the names have
258 been changed to start with UCHAR21, and an extensive comment has been added
259 to their definition.
260
261 8. Add missing (new) files sljitNativeTILEGX.c and sljitNativeTILEGX-encoder.c
262 to the export list in Makefile.am (they were accidentally omitted from the
263 8.34 tarball).
264
265 9. The informational output from pcretest used the phrase "starting byte set"
266 which is inappropriate for the 16-bit and 32-bit libraries. As the output
267 for "first char" and "need char" really means "non-UTF-char", I've changed
268 "byte" to "char", and slightly reworded the output. The documentation about
269 these values has also been (I hope) clarified.
270
271 10. Another JIT related optimization: use table jumps for selecting the correct
272 backtracking path, when more than four alternatives are present inside a
273 bracket.
274
275 11. Empty match is not possible, when the minimum length is greater than zero,
276 and there is no \K in the pattern. JIT should avoid empty match checks in
277 such cases.
278
279 12. In a caseless character class with UCP support, when a character with more
280 than one alternative case was not the first character of a range, not all
281 the alternative cases were added to the class. For example, s and \x{17f}
282 are both alternative cases for S: the class [RST] was handled correctly,
283 but [R-T] was not.
284
285 13. The configure.ac file always checked for pthread support when JIT was
286 enabled. This is not used in Windows, so I have put this test inside a
287 check for the presence of windows.h (which was already tested for).
288
289 14. Improve pattern prefix search by a simplified Boyer-Moore algorithm in JIT.
290 The algorithm provides a way to skip certain starting offsets, and usually
291 faster than linear prefix searches.
292
293 15. Change 13 for 8.20 updated RunTest to check for the 'fr' locale as well
294 as for 'fr_FR' and 'french'. For some reason, however, it then used the
295 Windows-specific input and output files, which have 'french' screwed in.
296 So this could never have worked. One of the problems with locales is that
297 they aren't always the same. I have now updated RunTest so that it checks
298 the output of the locale test (test 3) against three different output
299 files, and it allows the test to pass if any one of them matches. With luck
300 this should make the test pass on some versions of Solaris where it was
301 failing. Because of the uncertainty, the script did not used to stop if
302 test 3 failed; it now does. If further versions of a French locale ever
303 come to light, they can now easily be added.
304
305 16. If --with-pcregrep-bufsize was given a non-integer value such as "50K",
306 there was a message during ./configure, but it did not stop. This now
307 provokes an error. The invalid example in README has been corrected.
308 If a value less than the minimum is given, the minimum value has always
309 been used, but now a warning is given.
310
311 17. If --enable-bsr-anycrlf was set, the special 16/32-bit test failed. This
312 was a bug in the test system, which is now fixed. Also, the list of various
313 configurations that are tested for each release did not have one with both
314 16/32 bits and --enable-bar-anycrlf. It now does.
315
316 18. pcretest was missing "-C bsr" for displaying the \R default setting.
317
318 19. Little endian PowerPC systems are supported now by the JIT compiler.
319
320 20. The fast forward newline mechanism could enter to an infinite loop on
321 certain invalid UTF-8 input. Although we don't support these cases
322 this issue can be fixed by a performance optimization.
323
324 21. Change 33 of 8.34 is not sufficient to ensure stack safety because it does
325 not take account if existing stack usage. There is now a new global
326 variable called pcre_stack_guard that can be set to point to an external
327 function to check stack availability. It is called at the start of
328 processing every parenthesized group.
329
330 22. A typo in the code meant that in ungreedy mode the max/min qualifier
331 behaved like a min-possessive qualifier, and, for example, /a{1,3}b/U did
332 not match "ab".
333
334 23. When UTF was disabled, the JIT program reported some incorrect compile
335 errors. These messages are silenced now.
336
337 24. Experimental support for ARM-64 and MIPS-64 has been added to the JIT
338 compiler.
339
340 25. Change all the temporary files used in RunGrepTest to be different to those
341 used by RunTest so that the tests can be run simultaneously, for example by
342 "make -j check".
343
344
345 Version 8.34 15-December-2013
346 -----------------------------
347
348 1. Add pcre[16|32]_jit_free_unused_memory to forcibly free unused JIT
349 executable memory. Patch inspired by Carsten Klein.
350
351 2. ./configure --enable-coverage defined SUPPORT_GCOV in config.h, although
352 this macro is never tested and has no effect, because the work to support
353 coverage involves only compiling and linking options and special targets in
354 the Makefile. The comment in config.h implied that defining the macro would
355 enable coverage support, which is totally false. There was also support for
356 setting this macro in the CMake files (my fault, I just copied it from
357 configure). SUPPORT_GCOV has now been removed.
358
359 3. Make a small performance improvement in strlen16() and strlen32() in
360 pcretest.
361
362 4. Change 36 for 8.33 left some unreachable statements in pcre_exec.c,
363 detected by the Solaris compiler (gcc doesn't seem to be able to diagnose
364 these cases). There was also one in pcretest.c.
365
366 5. Cleaned up a "may be uninitialized" compiler warning in pcre_exec.c.
367
368 6. In UTF mode, the code for checking whether a group could match an empty
369 string (which is used for indefinitely repeated groups to allow for
370 breaking an infinite loop) was broken when the group contained a repeated
371 negated single-character class with a character that occupied more than one
372 data item and had a minimum repetition of zero (for example, [^\x{100}]* in
373 UTF-8 mode). The effect was undefined: the group might or might not be
374 deemed as matching an empty string, or the program might have crashed.
375
376 7. The code for checking whether a group could match an empty string was not
377 recognizing that \h, \H, \v, \V, and \R must match a character.
378
379 8. Implemented PCRE_INFO_MATCH_EMPTY, which yields 1 if the pattern can match
380 an empty string. If it can, pcretest shows this in its information output.
381
382 9. Fixed two related bugs that applied to Unicode extended grapheme clusters
383 that were repeated with a maximizing qualifier (e.g. \X* or \X{2,5}) when
384 matched by pcre_exec() without using JIT:
385
386 (a) If the rest of the pattern did not match after a maximal run of
387 grapheme clusters, the code for backing up to try with fewer of them
388 did not always back up over a full grapheme when characters that do not
389 have the modifier quality were involved, e.g. Hangul syllables.
390
391 (b) If the match point in a subject started with modifier character, and
392 there was no match, the code could incorrectly back up beyond the match
393 point, and potentially beyond the first character in the subject,
394 leading to a segfault or an incorrect match result.
395
396 10. A conditional group with an assertion condition could lead to PCRE
397 recording an incorrect first data item for a match if no other first data
398 item was recorded. For example, the pattern (?(?=ab)ab) recorded "a" as a
399 first data item, and therefore matched "ca" after "c" instead of at the
400 start.
401
402 11. Change 40 for 8.33 (allowing pcregrep to find empty strings) showed up a
403 bug that caused the command "echo a | ./pcregrep -M '|a'" to loop.
404
405 12. The source of pcregrep now includes z/OS-specific code so that it can be
406 compiled for z/OS as part of the special z/OS distribution.
407
408 13. Added the -T and -TM options to pcretest.
409
410 14. The code in pcre_compile.c for creating the table of named capturing groups
411 has been refactored. Instead of creating the table dynamically during the
412 actual compiling pass, the information is remembered during the pre-compile
413 pass (on the stack unless there are more than 20 named groups, in which
414 case malloc() is used) and the whole table is created before the actual
415 compile happens. This has simplified the code (it is now nearly 150 lines
416 shorter) and prepared the way for better handling of references to groups
417 with duplicate names.
418
419 15. A back reference to a named subpattern when there is more than one of the
420 same name now checks them in the order in which they appear in the pattern.
421 The first one that is set is used for the reference. Previously only the
422 first one was inspected. This change makes PCRE more compatible with Perl.
423
424 16. Unicode character properties were updated from Unicode 6.3.0.
425
426 17. The compile-time code for auto-possessification has been refactored, based
427 on a patch by Zoltan Herczeg. It now happens after instead of during
428 compilation. The code is cleaner, and more cases are handled. The option
429 PCRE_NO_AUTO_POSSESS is added for testing purposes, and the -O and /O
430 options in pcretest are provided to set it. It can also be set by
431 (*NO_AUTO_POSSESS) at the start of a pattern.
432
433 18. The character VT has been added to the default ("C" locale) set of
434 characters that match \s and are generally treated as white space,
435 following this same change in Perl 5.18. There is now no difference between
436 "Perl space" and "POSIX space". Whether VT is treated as white space in
437 other locales depends on the locale.
438
439 19. The code for checking named groups as conditions, either for being set or
440 for being recursed, has been refactored (this is related to 14 and 15
441 above). Processing unduplicated named groups should now be as fast at
442 numerical groups, and processing duplicated groups should be faster than
443 before.
444
445 20. Two patches to the CMake build system, by Alexander Barkov:
446
447 (1) Replace the "source" command by "." in CMakeLists.txt because
448 "source" is a bash-ism.
449
450 (2) Add missing HAVE_STDINT_H and HAVE_INTTYPES_H to config-cmake.h.in;
451 without these the CMake build does not work on Solaris.
452
453 21. Perl has changed its handling of \8 and \9. If there is no previously
454 encountered capturing group of those numbers, they are treated as the
455 literal characters 8 and 9 instead of a binary zero followed by the
456 literals. PCRE now does the same.
457
458 22. Following Perl, added \o{} to specify codepoints in octal, making it
459 possible to specify values greater than 0777 and also making them
460 unambiguous.
461
462 23. Perl now gives an error for missing closing braces after \x{... instead of
463 treating the string as literal. PCRE now does the same.
464
465 24. RunTest used to grumble if an inappropriate test was selected explicitly,
466 but just skip it when running all tests. This make it awkward to run ranges
467 of tests when one of them was inappropriate. Now it just skips any
468 inappropriate tests, as it always did when running all tests.
469
470 25. If PCRE_AUTO_CALLOUT and PCRE_UCP were set for a pattern that contained
471 character types such as \d or \w, too many callouts were inserted, and the
472 data that they returned was rubbish.
473
474 26. In UCP mode, \s was not matching two of the characters that Perl matches,
475 namely NEL (U+0085) and MONGOLIAN VOWEL SEPARATOR (U+180E), though they
476 were matched by \h. The code has now been refactored so that the lists of
477 the horizontal and vertical whitespace characters used for \h and \v (which
478 are defined only in one place) are now also used for \s.
479
480 27. Add JIT support for the 64 bit TileGX architecture.
481 Patch by Jiong Wang (Tilera Corporation).
482
483 28. Possessive quantifiers for classes (both explicit and automatically
484 generated) now use special opcodes instead of wrapping in ONCE brackets.
485
486 29. Whereas an item such as A{4}+ ignored the possessivenes of the quantifier
487 (because it's meaningless), this was not happening when PCRE_CASELESS was
488 set. Not wrong, but inefficient.
489
490 30. Updated perltest.pl to add /u (force Unicode mode) when /W (use Unicode
491 properties for \w, \d, etc) is present in a test regex. Otherwise if the
492 test contains no characters greater than 255, Perl doesn't realise it
493 should be using Unicode semantics.
494
495 31. Upgraded the handling of the POSIX classes [:graph:], [:print:], and
496 [:punct:] when PCRE_UCP is set so as to include the same characters as Perl
497 does in Unicode mode.
498
499 32. Added the "forbid" facility to pcretest so that putting tests into the
500 wrong test files can sometimes be quickly detected.
501
502 33. There is now a limit (default 250) on the depth of nesting of parentheses.
503 This limit is imposed to control the amount of system stack used at compile
504 time. It can be changed at build time by --with-parens-nest-limit=xxx or
505 the equivalent in CMake.
506
507 34. Character classes such as [A-\d] or [a-[:digit:]] now cause compile-time
508 errors. Perl warns for these when in warning mode, but PCRE has no facility
509 for giving warnings.
510
511 35. Change 34 for 8.13 allowed quantifiers on assertions, because Perl does.
512 However, this was not working for (?!) because it is optimized to (*FAIL),
513 for which PCRE does not allow quantifiers. The optimization is now disabled
514 when a quantifier follows (?!). I can't see any use for this, but it makes
515 things uniform.
516
517 36. Perl no longer allows group names to start with digits, so I have made this
518 change also in PCRE. It simplifies the code a bit.
519
520 37. In extended mode, Perl ignores spaces before a + that indicates a
521 possessive quantifier. PCRE allowed a space before the quantifier, but not
522 before the possessive +. It now does.
523
524 38. The use of \K (reset reported match start) within a repeated possessive
525 group such as (a\Kb)*+ was not working.
526
527 40. Document that the same character tables must be used at compile time and
528 run time, and that the facility to pass tables to pcre_exec() and
529 pcre_dfa_exec() is for use only with saved/restored patterns.
530
531 41. Applied Jeff Trawick's patch CMakeLists.txt, which "provides two new
532 features for Builds with MSVC:
533
534 1. Support pcre.rc and/or pcreposix.rc (as is already done for MinGW
535 builds). The .rc files can be used to set FileDescription and many other
536 attributes.
537
538 2. Add an option (-DINSTALL_MSVC_PDB) to enable installation of .pdb files.
539 This allows higher-level build scripts which want .pdb files to avoid
540 hard-coding the exact files needed."
541
542 42. Added support for [[:<:]] and [[:>:]] as used in the BSD POSIX library to
543 mean "start of word" and "end of word", respectively, as a transition aid.
544
545 43. A minimizing repeat of a class containing codepoints greater than 255 in
546 non-UTF 16-bit or 32-bit modes caused an internal error when PCRE was
547 compiled to use the heap for recursion.
548
549 44. Got rid of some compiler warnings for unused variables when UTF but not UCP
550 is configured.
551
552
553 Version 8.33 28-May-2013
554 ------------------------
555
556 1. Added 'U' to some constants that are compared to unsigned integers, to
557 avoid compiler signed/unsigned warnings. Added (int) casts to unsigned
558 variables that are added to signed variables, to ensure the result is
559 signed and can be negated.
560
561 2. Applied patch by Daniel Richard G for quashing MSVC warnings to the
562 CMake config files.
563
564 3. Revise the creation of config.h.generic so that all boolean macros are
565 #undefined, whereas non-boolean macros are #ifndef/#endif-ed. This makes
566 overriding via -D on the command line possible.
567
568 4. Changing the definition of the variable "op" in pcre_exec.c from pcre_uchar
569 to unsigned int is reported to make a quite noticeable speed difference in
570 a specific Windows environment. Testing on Linux did also appear to show
571 some benefit (and it is clearly not harmful). Also fixed the definition of
572 Xop which should be unsigned.
573
574 5. Related to (4), changing the definition of the intermediate variable cc
575 in repeated character loops from pcre_uchar to pcre_uint32 also gave speed
576 improvements.
577
578 6. Fix forward search in JIT when link size is 3 or greater. Also removed some
579 unnecessary spaces.
580
581 7. Adjust autogen.sh and configure.ac to lose warnings given by automake 1.12
582 and later.
583
584 8. Fix two buffer over read issues in 16 and 32 bit modes. Affects JIT only.
585
586 9. Optimizing fast_forward_start_bits in JIT.
587
588 10. Adding support for callouts in JIT, and fixing some issues revealed
589 during this work. Namely:
590
591 (a) Unoptimized capturing brackets incorrectly reset on backtrack.
592
593 (b) Minimum length was not checked before the matching is started.
594
595 11. The value of capture_last that is passed to callouts was incorrect in some
596 cases when there was a capture on one path that was subsequently abandoned
597 after a backtrack. Also, the capture_last value is now reset after a
598 recursion, since all captures are also reset in this case.
599
600 12. The interpreter no longer returns the "too many substrings" error in the
601 case when an overflowing capture is in a branch that is subsequently
602 abandoned after a backtrack.
603
604 13. In the pathological case when an offset vector of size 2 is used, pcretest
605 now prints out the matched string after a yield of 0 or 1.
606
607 14. Inlining subpatterns in recursions, when certain conditions are fulfilled.
608 Only supported by the JIT compiler at the moment.
609
610 15. JIT compiler now supports 32 bit Macs thanks to Lawrence Velazquez.
611
612 16. Partial matches now set offsets[2] to the "bumpalong" value, that is, the
613 offset of the starting point of the matching process, provided the offsets
614 vector is large enough.
615
616 17. The \A escape now records a lookbehind value of 1, though its execution
617 does not actually inspect the previous character. This is to ensure that,
618 in partial multi-segment matching, at least one character from the old
619 segment is retained when a new segment is processed. Otherwise, if there
620 are no lookbehinds in the pattern, \A might match incorrectly at the start
621 of a new segment.
622
623 18. Added some #ifdef __VMS code into pcretest.c to help VMS implementations.
624
625 19. Redefined some pcre_uchar variables in pcre_exec.c as pcre_uint32; this
626 gives some modest performance improvement in 8-bit mode.
627
628 20. Added the PCRE-specific property \p{Xuc} for matching characters that can
629 be expressed in certain programming languages using Universal Character
630 Names.
631
632 21. Unicode validation has been updated in the light of Unicode Corrigendum #9,
633 which points out that "non characters" are not "characters that may not
634 appear in Unicode strings" but rather "characters that are reserved for
635 internal use and have only local meaning".
636
637 22. When a pattern was compiled with automatic callouts (PCRE_AUTO_CALLOUT) and
638 there was a conditional group that depended on an assertion, if the
639 assertion was false, the callout that immediately followed the alternation
640 in the condition was skipped when pcre_exec() was used for matching.
641
642 23. Allow an explicit callout to be inserted before an assertion that is the
643 condition for a conditional group, for compatibility with automatic
644 callouts, which always insert a callout at this point.
645
646 24. In 8.31, (*COMMIT) was confined to within a recursive subpattern. Perl also
647 confines (*SKIP) and (*PRUNE) in the same way, and this has now been done.
648
649 25. (*PRUNE) is now supported by the JIT compiler.
650
651 26. Fix infinite loop when /(?<=(*SKIP)ac)a/ is matched against aa.
652
653 27. Fix the case where there are two or more SKIPs with arguments that may be
654 ignored.
655
656 28. (*SKIP) is now supported by the JIT compiler.
657
658 29. (*THEN) is now supported by the JIT compiler.
659
660 30. Update RunTest with additional test selector options.
661
662 31. The way PCRE handles backtracking verbs has been changed in two ways.
663
664 (1) Previously, in something like (*COMMIT)(*SKIP), COMMIT would override
665 SKIP. Now, PCRE acts on whichever backtracking verb is reached first by
666 backtracking. In some cases this makes it more Perl-compatible, but Perl's
667 rather obscure rules do not always do the same thing.
668
669 (2) Previously, backtracking verbs were confined within assertions. This is
670 no longer the case for positive assertions, except for (*ACCEPT). Again,
671 this sometimes improves Perl compatibility, and sometimes does not.
672
673 32. A number of tests that were in test 2 because Perl did things differently
674 have been moved to test 1, because either Perl or PCRE has changed, and
675 these tests are now compatible.
676
677 32. Backtracking control verbs are now handled in the same way in JIT and
678 interpreter.
679
680 33. An opening parenthesis in a MARK/PRUNE/SKIP/THEN name in a pattern that
681 contained a forward subroutine reference caused a compile error.
682
683 34. Auto-detect and optimize limited repetitions in JIT.
684
685 35. Implement PCRE_NEVER_UTF to lock out the use of UTF, in particular,
686 blocking (*UTF) etc.
687
688 36. In the interpreter, maximizing pattern repetitions for characters and
689 character types now use tail recursion, which reduces stack usage.
690
691 37. The value of the max lookbehind was not correctly preserved if a compiled
692 and saved regex was reloaded on a host of different endianness.
693
694 38. Implemented (*LIMIT_MATCH) and (*LIMIT_RECURSION). As part of the extension
695 of the compiled pattern block, expand the flags field from 16 to 32 bits
696 because it was almost full.
697
698 39. Try madvise first before posix_madvise.
699
700 40. Change 7 for PCRE 7.9 made it impossible for pcregrep to find empty lines
701 with a pattern such as ^$. It has taken 4 years for anybody to notice! The
702 original change locked out all matches of empty strings. This has been
703 changed so that one match of an empty string per line is recognized.
704 Subsequent searches on the same line (for colouring or for --only-matching,
705 for example) do not recognize empty strings.
706
707 41. Applied a user patch to fix a number of spelling mistakes in comments.
708
709 42. Data lines longer than 65536 caused pcretest to crash.
710
711 43. Clarified the data type for length and startoffset arguments for pcre_exec
712 and pcre_dfa_exec in the function-specific man pages, where they were
713 explicitly stated to be in bytes, never having been updated. I also added
714 some clarification to the pcreapi man page.
715
716 44. A call to pcre_dfa_exec() with an output vector size less than 2 caused
717 a segmentation fault.
718
719
720 Version 8.32 30-November-2012
721 -----------------------------
722
723 1. Improved JIT compiler optimizations for first character search and single
724 character iterators.
725
726 2. Supporting IBM XL C compilers for PPC architectures in the JIT compiler.
727 Patch by Daniel Richard G.
728
729 3. Single character iterator optimizations in the JIT compiler.
730
731 4. Improved JIT compiler optimizations for character ranges.
732
733 5. Rename the "leave" variable names to "quit" to improve WinCE compatibility.
734 Reported by Giuseppe D'Angelo.
735
736 6. The PCRE_STARTLINE bit, indicating that a match can occur only at the start
737 of a line, was being set incorrectly in cases where .* appeared inside
738 atomic brackets at the start of a pattern, or where there was a subsequent
739 *PRUNE or *SKIP.
740
741 7. Improved instruction cache flush for POWER/PowerPC.
742 Patch by Daniel Richard G.
743
744 8. Fixed a number of issues in pcregrep, making it more compatible with GNU
745 grep:
746
747 (a) There is now no limit to the number of patterns to be matched.
748
749 (b) An error is given if a pattern is too long.
750
751 (c) Multiple uses of --exclude, --exclude-dir, --include, and --include-dir
752 are now supported.
753
754 (d) --exclude-from and --include-from (multiple use) have been added.
755
756 (e) Exclusions and inclusions now apply to all files and directories, not
757 just to those obtained from scanning a directory recursively.
758
759 (f) Multiple uses of -f and --file-list are now supported.
760
761 (g) In a Windows environment, the default for -d has been changed from
762 "read" (the GNU grep default) to "skip", because otherwise the presence
763 of a directory in the file list provokes an error.
764
765 (h) The documentation has been revised and clarified in places.
766
767 9. Improve the matching speed of capturing brackets.
768
769 10. Changed the meaning of \X so that it now matches a Unicode extended
770 grapheme cluster.
771
772 11. Patch by Daniel Richard G to the autoconf files to add a macro for sorting
773 out POSIX threads when JIT support is configured.
774
775 12. Added support for PCRE_STUDY_EXTRA_NEEDED.
776
777 13. In the POSIX wrapper regcomp() function, setting re_nsub field in the preg
778 structure could go wrong in environments where size_t is not the same size
779 as int.
780
781 14. Applied user-supplied patch to pcrecpp.cc to allow PCRE_NO_UTF8_CHECK to be
782 set.
783
784 15. The EBCDIC support had decayed; later updates to the code had included
785 explicit references to (e.g.) \x0a instead of CHAR_LF. There has been a
786 general tidy up of EBCDIC-related issues, and the documentation was also
787 not quite right. There is now a test that can be run on ASCII systems to
788 check some of the EBCDIC-related things (but is it not a full test).
789
790 16. The new PCRE_STUDY_EXTRA_NEEDED option is now used by pcregrep, resulting
791 in a small tidy to the code.
792
793 17. Fix JIT tests when UTF is disabled and both 8 and 16 bit mode are enabled.
794
795 18. If the --only-matching (-o) option in pcregrep is specified multiple
796 times, each one causes appropriate output. For example, -o1 -o2 outputs the
797 substrings matched by the 1st and 2nd capturing parentheses. A separating
798 string can be specified by --om-separator (default empty).
799
800 19. Improving the first n character searches.
801
802 20. Turn case lists for horizontal and vertical white space into macros so that
803 they are defined only once.
804
805 21. This set of changes together give more compatible Unicode case-folding
806 behaviour for characters that have more than one other case when UCP
807 support is available.
808
809 (a) The Unicode property table now has offsets into a new table of sets of
810 three or more characters that are case-equivalent. The MultiStage2.py
811 script that generates these tables (the pcre_ucd.c file) now scans
812 CaseFolding.txt instead of UnicodeData.txt for character case
813 information.
814
815 (b) The code for adding characters or ranges of characters to a character
816 class has been abstracted into a generalized function that also handles
817 case-independence. In UTF-mode with UCP support, this uses the new data
818 to handle characters with more than one other case.
819
820 (c) A bug that is fixed as a result of (b) is that codepoints less than 256
821 whose other case is greater than 256 are now correctly matched
822 caselessly. Previously, the high codepoint matched the low one, but not
823 vice versa.
824
825 (d) The processing of \h, \H, \v, and \ in character classes now makes use
826 of the new class addition function, using character lists defined as
827 macros alongside the case definitions of 20 above.
828
829 (e) Caseless back references now work with characters that have more than
830 one other case.
831
832 (f) General caseless matching of characters with more than one other case
833 is supported.
834
835 22. Unicode character properties were updated from Unicode 6.2.0
836
837 23. Improved CMake support under Windows. Patch by Daniel Richard G.
838
839 24. Add support for 32-bit character strings, and UTF-32
840
841 25. Major JIT compiler update (code refactoring and bugfixing).
842 Experimental Sparc 32 support is added.
843
844 26. Applied a modified version of Daniel Richard G's patch to create
845 pcre.h.generic and config.h.generic by "make" instead of in the
846 PrepareRelease script.
847
848 27. Added a definition for CHAR_NULL (helpful for the z/OS port), and use it in
849 pcre_compile.c when checking for a zero character.
850
851 28. Introducing a native interface for JIT. Through this interface, the compiled
852 machine code can be directly executed. The purpose of this interface is to
853 provide fast pattern matching, so several sanity checks are not performed.
854 However, feature tests are still performed. The new interface provides
855 1.4x speedup compared to the old one.
856
857 29. If pcre_exec() or pcre_dfa_exec() was called with a negative value for
858 the subject string length, the error given was PCRE_ERROR_BADOFFSET, which
859 was confusing. There is now a new error PCRE_ERROR_BADLENGTH for this case.
860
861 30. In 8-bit UTF-8 mode, pcretest failed to give an error for data codepoints
862 greater than 0x7fffffff (which cannot be represented in UTF-8, even under
863 the "old" RFC 2279). Instead, it ended up passing a negative length to
864 pcre_exec().
865
866 31. Add support for GCC's visibility feature to hide internal functions.
867
868 32. Running "pcretest -C pcre8" or "pcretest -C pcre16" gave a spurious error
869 "unknown -C option" after outputting 0 or 1.
870
871 33. There is now support for generating a code coverage report for the test
872 suite in environments where gcc is the compiler and lcov is installed. This
873 is mainly for the benefit of the developers.
874
875 34. If PCRE is built with --enable-valgrind, certain memory regions are marked
876 unaddressable using valgrind annotations, allowing valgrind to detect
877 invalid memory accesses. This is mainly for the benefit of the developers.
878
879 25. (*UTF) can now be used to start a pattern in any of the three libraries.
880
881 26. Give configure error if --enable-cpp but no C++ compiler found.
882
883
884 Version 8.31 06-July-2012
885 -------------------------
886
887 1. Fixing a wrong JIT test case and some compiler warnings.
888
889 2. Removed a bashism from the RunTest script.
890
891 3. Add a cast to pcre_exec.c to fix the warning "unary minus operator applied
892 to unsigned type, result still unsigned" that was given by an MS compiler
893 on encountering the code "-sizeof(xxx)".
894
895 4. Partial matching support is added to the JIT compiler.
896
897 5. Fixed several bugs concerned with partial matching of items that consist
898 of more than one character:
899
900 (a) /^(..)\1/ did not partially match "aba" because checking references was
901 done on an "all or nothing" basis. This also applied to repeated
902 references.
903
904 (b) \R did not give a hard partial match if \r was found at the end of the
905 subject.
906
907 (c) \X did not give a hard partial match after matching one or more
908 characters at the end of the subject.
909
910 (d) When newline was set to CRLF, a pattern such as /a$/ did not recognize
911 a partial match for the string "\r".
912
913 (e) When newline was set to CRLF, the metacharacter "." did not recognize
914 a partial match for a CR character at the end of the subject string.
915
916 6. If JIT is requested using /S++ or -s++ (instead of just /S+ or -s+) when
917 running pcretest, the text "(JIT)" added to the output whenever JIT is
918 actually used to run the match.
919
920 7. Individual JIT compile options can be set in pcretest by following -s+[+]
921 or /S+[+] with a digit between 1 and 7.
922
923 8. OP_NOT now supports any UTF character not just single-byte ones.
924
925 9. (*MARK) control verb is now supported by the JIT compiler.
926
927 10. The command "./RunTest list" lists the available tests without actually
928 running any of them. (Because I keep forgetting what they all are.)
929
930 11. Add PCRE_INFO_MAXLOOKBEHIND.
931
932 12. Applied a (slightly modified) user-supplied patch that improves performance
933 when the heap is used for recursion (compiled with --disable-stack-for-
934 recursion). Instead of malloc and free for each heap frame each time a
935 logical recursion happens, frames are retained on a chain and re-used where
936 possible. This sometimes gives as much as 30% improvement.
937
938 13. As documented, (*COMMIT) is now confined to within a recursive subpattern
939 call.
940
941 14. As documented, (*COMMIT) is now confined to within a positive assertion.
942
943 15. It is now possible to link pcretest with libedit as an alternative to
944 libreadline.
945
946 16. (*COMMIT) control verb is now supported by the JIT compiler.
947
948 17. The Unicode data tables have been updated to Unicode 6.1.0.
949
950 18. Added --file-list option to pcregrep.
951
952 19. Added binary file support to pcregrep, including the -a, --binary-files,
953 -I, and --text options.
954
955 20. The madvise function is renamed for posix_madvise for QNX compatibility
956 reasons. Fixed by Giuseppe D'Angelo.
957
958 21. Fixed a bug for backward assertions with REVERSE 0 in the JIT compiler.
959
960 22. Changed the option for creating symbolic links for 16-bit man pages from
961 -s to -sf so that re-installing does not cause issues.
962
963 23. Support PCRE_NO_START_OPTIMIZE in JIT as (*MARK) support requires it.
964
965 24. Fixed a very old bug in pcretest that caused errors with restarted DFA
966 matches in certain environments (the workspace was not being correctly
967 retained). Also added to pcre_dfa_exec() a simple plausibility check on
968 some of the workspace data at the beginning of a restart.
969
970 25. \s*\R was auto-possessifying the \s* when it should not, whereas \S*\R
971 was not doing so when it should - probably a typo introduced by SVN 528
972 (change 8.10/14).
973
974 26. When PCRE_UCP was not set, \w+\x{c4} was incorrectly auto-possessifying the
975 \w+ when the character tables indicated that \x{c4} was a word character.
976 There were several related cases, all because the tests for doing a table
977 lookup were testing for characters less than 127 instead of 255.
978
979 27. If a pattern contains capturing parentheses that are not used in a match,
980 their slots in the ovector are set to -1. For those that are higher than
981 any matched groups, this happens at the end of processing. In the case when
982 there were back references that the ovector was too small to contain
983 (causing temporary malloc'd memory to be used during matching), and the
984 highest capturing number was not used, memory off the end of the ovector
985 was incorrectly being set to -1. (It was using the size of the temporary
986 memory instead of the true size.)
987
988 28. To catch bugs like 27 using valgrind, when pcretest is asked to specify an
989 ovector size, it uses memory at the end of the block that it has got.
990
991 29. Check for an overlong MARK name and give an error at compile time. The
992 limit is 255 for the 8-bit library and 65535 for the 16-bit library.
993
994 30. JIT compiler update.
995
996 31. JIT is now supported on jailbroken iOS devices. Thanks for Ruiger
997 Rill for the patch.
998
999 32. Put spaces around SLJIT_PRINT_D in the JIT compiler. Required by CXX11.
1000
1001 33. Variable renamings in the PCRE-JIT compiler. No functionality change.
1002
1003 34. Fixed typos in pcregrep: in two places there was SUPPORT_LIBZ2 instead of
1004 SUPPORT_LIBBZ2. This caused a build problem when bzip2 but not gzip (zlib)
1005 was enabled.
1006
1007 35. Improve JIT code generation for greedy plus quantifier.
1008
1009 36. When /((?:a?)*)*c/ or /((?>a?)*)*c/ was matched against "aac", it set group
1010 1 to "aa" instead of to an empty string. The bug affected repeated groups
1011 that could potentially match an empty string.
1012
1013 37. Optimizing single character iterators in JIT.
1014
1015 38. Wide characters specified with \uxxxx in JavaScript mode are now subject to
1016 the same checks as \x{...} characters in non-JavaScript mode. Specifically,
1017 codepoints that are too big for the mode are faulted, and in a UTF mode,
1018 disallowed codepoints are also faulted.
1019
1020 39. If PCRE was compiled with UTF support, in three places in the DFA
1021 matcher there was code that should only have been obeyed in UTF mode, but
1022 was being obeyed unconditionally. In 8-bit mode this could cause incorrect
1023 processing when bytes with values greater than 127 were present. In 16-bit
1024 mode the bug would be provoked by values in the range 0xfc00 to 0xdc00. In
1025 both cases the values are those that cannot be the first data item in a UTF
1026 character. The three items that might have provoked this were recursions,
1027 possessively repeated groups, and atomic groups.
1028
1029 40. Ensure that libpcre is explicitly listed in the link commands for pcretest
1030 and pcregrep, because some OS require shared objects to be explicitly
1031 passed to ld, causing the link step to fail if they are not.
1032
1033 41. There were two incorrect #ifdefs in pcre_study.c, meaning that, in 16-bit
1034 mode, patterns that started with \h* or \R* might be incorrectly matched.
1035
1036
1037 Version 8.30 04-February-2012
1038 -----------------------------
1039
1040 1. Renamed "isnumber" as "is_a_number" because in some Mac environments this
1041 name is defined in ctype.h.
1042
1043 2. Fixed a bug in fixed-length calculation for lookbehinds that would show up
1044 only in quite long subpatterns.
1045
1046 3. Removed the function pcre_info(), which has been obsolete and deprecated
1047 since it was replaced by pcre_fullinfo() in February 2000.
1048
1049 4. For a non-anchored pattern, if (*SKIP) was given with a name that did not
1050 match a (*MARK), and the match failed at the start of the subject, a
1051 reference to memory before the start of the subject could occur. This bug
1052 was introduced by fix 17 of release 8.21.
1053
1054 5. A reference to an unset group with zero minimum repetition was giving
1055 totally wrong answers (in non-JavaScript-compatibility mode). For example,
1056 /(another)?(\1?)test/ matched against "hello world test". This bug was
1057 introduced in release 8.13.
1058
1059 6. Add support for 16-bit character strings (a large amount of work involving
1060 many changes and refactorings).
1061
1062 7. RunGrepTest failed on msys because \r\n was replaced by whitespace when the
1063 command "pattern=`printf 'xxx\r\njkl'`" was run. The pattern is now taken
1064 from a file.
1065
1066 8. Ovector size of 2 is also supported by JIT based pcre_exec (the ovector size
1067 rounding is not applied in this particular case).
1068
1069 9. The invalid Unicode surrogate codepoints U+D800 to U+DFFF are now rejected
1070 if they appear, or are escaped, in patterns.
1071
1072 10. Get rid of a number of -Wunused-but-set-variable warnings.
1073
1074 11. The pattern /(?=(*:x))(q|)/ matches an empty string, and returns the mark
1075 "x". The similar pattern /(?=(*:x))((*:y)q|)/ did not return a mark at all.
1076 Oddly, Perl behaves the same way. PCRE has been fixed so that this pattern
1077 also returns the mark "x". This bug applied to capturing parentheses,
1078 non-capturing parentheses, and atomic parentheses. It also applied to some
1079 assertions.
1080
1081 12. Stephen Kelly's patch to CMakeLists.txt allows it to parse the version
1082 information out of configure.ac instead of relying on pcre.h.generic, which
1083 is not stored in the repository.
1084
1085 13. Applied Dmitry V. Levin's patch for a more portable method for linking with
1086 -lreadline.
1087
1088 14. ZH added PCRE_CONFIG_JITTARGET; added its output to pcretest -C.
1089
1090 15. Applied Graycode's patch to put the top-level frame on the stack rather
1091 than the heap when not using the stack for recursion. This gives a
1092 performance improvement in many cases when recursion is not deep.
1093
1094 16. Experimental code added to "pcretest -C" to output the stack frame size.
1095
1096
1097 Version 8.21 12-Dec-2011
1098 ------------------------
1099
1100 1. Updating the JIT compiler.
1101
1102 2. JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
1103 are added as well.
1104
1105 3. Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
1106 PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
1107 calling _pcre_jit_exec. Some extra comments are added.
1108
1109 4. (*MARK) settings inside atomic groups that do not contain any capturing
1110 parentheses, for example, (?>a(*:m)), were not being passed out. This bug
1111 was introduced by change 18 for 8.20.
1112
1113 5. Supporting of \x, \U and \u in JavaScript compatibility mode based on the
1114 ECMA-262 standard.
1115
1116 6. Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
1117 erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
1118 This bug was probably introduced by change 9 of 8.13.
1119
1120 7. While fixing 6 above, I noticed that a number of other items were being
1121 incorrectly rejected as "not fixed length". This arose partly because newer
1122 opcodes had not been added to the fixed-length checking code. I have (a)
1123 corrected the bug and added tests for these items, and (b) arranged for an
1124 error to occur if an unknown opcode is encountered while checking for fixed
1125 length instead of just assuming "not fixed length". The items that were
1126 rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
1127 (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
1128 repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
1129
1130 8. A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
1131 being incorrectly compiled and would have given unpredicatble results.
1132
1133 9. A possessively repeated subpattern with minimum repeat count greater than
1134 one behaved incorrectly. For example, (A){2,}+ behaved as if it was
1135 (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
1136 the first (A) could occur when it should not.
1137
1138 10. Add a cast and remove a redundant test from the code.
1139
1140 11. JIT should use pcre_malloc/pcre_free for allocation.
1141
1142 12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
1143 best practice nowadays, and helps with cross-compiling. (If the exec_prefix
1144 is anything other than /usr, -L is still shown).
1145
1146 13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
1147
1148 14. Perl does not support \N without a following name in a [] class; PCRE now
1149 also gives an error.
1150
1151 15. If a forward reference was repeated with an upper limit of around 2000,
1152 it caused the error "internal error: overran compiling workspace". The
1153 maximum number of forward references (including repeats) was limited by the
1154 internal workspace, and dependent on the LINK_SIZE. The code has been
1155 rewritten so that the workspace expands (via pcre_malloc) if necessary, and
1156 the default depends on LINK_SIZE. There is a new upper limit (for safety)
1157 of around 200,000 forward references. While doing this, I also speeded up
1158 the filling in of repeated forward references.
1159
1160 16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
1161 incorrectly expecting the subject to contain another "a" after the start.
1162
1163 17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
1164 in the match, the SKIP should be ignored. This was not happening; instead
1165 the SKIP was being treated as NOMATCH. For patterns such as
1166 /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
1167 tested.
1168
1169 18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
1170 now much more compatible with Perl, in particular in cases where the result
1171 is a non-match for a non-anchored pattern. For example, if
1172 /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
1173 "m", where previously it did not return a name. A side effect of this
1174 change is that for partial matches, the last encountered mark name is
1175 returned, as for non matches. A number of tests that were previously not
1176 Perl-compatible have been moved into the Perl-compatible test files. The
1177 refactoring has had the pleasing side effect of removing one argument from
1178 the match() function, thus reducing its stack requirements.
1179
1180 19. If the /S+ option was used in pcretest to study a pattern using JIT,
1181 subsequent uses of /S (without +) incorrectly behaved like /S+.
1182
1183 21. Retrieve executable code size support for the JIT compiler and fixing
1184 some warnings.
1185
1186 22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
1187 not work when the shorter character appeared right at the end of the
1188 subject string.
1189
1190 23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
1191 systems.
1192
1193 24. Added PCRE_INFO_JITSIZE to pass on the value from (21) above, and also
1194 output it when the /M option is used in pcretest.
1195
1196 25. The CheckMan script was not being included in the distribution. Also, added
1197 an explicit "perl" to run Perl scripts from the PrepareRelease script
1198 because this is reportedly needed in Windows.
1199
1200 26. If study data was being save in a file and studying had not found a set of
1201 "starts with" bytes for the pattern, the data written to the file (though
1202 never used) was taken from uninitialized memory and so caused valgrind to
1203 complain.
1204
1205 27. Updated RunTest.bat as provided by Sheri Pierce.
1206
1207 28. Fixed a possible uninitialized memory bug in pcre_jit_compile.c.
1208
1209 29. Computation of memory usage for the table of capturing group names was
1210 giving an unnecessarily large value.
1211
1212
1213 Version 8.20 21-Oct-2011
1214 ------------------------
1215
1216 1. Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
1217 a POSIX class. After further experiments with Perl, which convinced me that
1218 Perl has bugs and confusions, a closing square bracket is no longer allowed
1219 in a POSIX name. This bug also affected patterns with classes that started
1220 with full stops.
1221
1222 2. If a pattern such as /(a)b|ac/ is matched against "ac", there is no
1223 captured substring, but while checking the failing first alternative,
1224 substring 1 is temporarily captured. If the output vector supplied to
1225 pcre_exec() was not big enough for this capture, the yield of the function
1226 was still zero ("insufficient space for captured substrings"). This cannot
1227 be totally fixed without adding another stack variable, which seems a lot
1228 of expense for a edge case. However, I have improved the situation in cases
1229 such as /(a)(b)x|abc/ matched against "abc", where the return code
1230 indicates that fewer than the maximum number of slots in the ovector have
1231 been set.
1232
1233 3. Related to (2) above: when there are more back references in a pattern than
1234 slots in the output vector, pcre_exec() uses temporary memory during
1235 matching, and copies in the captures as far as possible afterwards. It was
1236 using the entire output vector, but this conflicts with the specification
1237 that only 2/3 is used for passing back captured substrings. Now it uses
1238 only the first 2/3, for compatibility. This is, of course, another edge
1239 case.
1240
1241 4. Zoltan Herczeg's just-in-time compiler support has been integrated into the
1242 main code base, and can be used by building with --enable-jit. When this is
1243 done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
1244 runtime --no-jit option is given.
1245
1246 5. When the number of matches in a pcre_dfa_exec() run exactly filled the
1247 ovector, the return from the function was zero, implying that there were
1248 other matches that did not fit. The correct "exactly full" value is now
1249 returned.
1250
1251 6. If a subpattern that was called recursively or as a subroutine contained
1252 (*PRUNE) or any other control that caused it to give a non-standard return,
1253 invalid errors such as "Error -26 (nested recursion at the same subject
1254 position)" or even infinite loops could occur.
1255
1256 7. If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
1257 computing the minimum length on reaching *ACCEPT, and so ended up with the
1258 wrong value of 1 rather than 0. Further investigation indicates that
1259 computing a minimum subject length in the presence of *ACCEPT is difficult
1260 (think back references, subroutine calls), and so I have changed the code
1261 so that no minimum is registered for a pattern that contains *ACCEPT.
1262
1263 8. If (*THEN) was present in the first (true) branch of a conditional group,
1264 it was not handled as intended. [But see 16 below.]
1265
1266 9. Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
1267 Sheri Pierce.
1268
1269 10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
1270 the first byte in a match must be "a".
1271
1272 11. Change 17 for 8.13 increased the recursion depth for patterns like
1273 /a(?:.)*?a/ drastically. I've improved things by remembering whether a
1274 pattern contains any instances of (*THEN). If it does not, the old
1275 optimizations are restored. It would be nice to do this on a per-group
1276 basis, but at the moment that is not feasible.
1277
1278 12. In some environments, the output of pcretest -C is CRLF terminated. This
1279 broke RunTest's code that checks for the link size. A single white space
1280 character after the value is now allowed for.
1281
1282 13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
1283 For "fr", it uses the Windows-specific input and output files.
1284
1285 14. If (*THEN) appeared in a group that was called recursively or as a
1286 subroutine, it did not work as intended. [But see next item.]
1287
1288 15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
1289 pattern fragments (but not containing any | characters). If A and B are
1290 matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
1291 was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
1292 D. In other words, Perl considers parentheses that do not contain any |
1293 characters to be part of a surrounding alternative, whereas PCRE was
1294 treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
1295 differently. PCRE now behaves in the same way as Perl, except in the case
1296 of subroutine/recursion calls such as (?1) which have in any case always
1297 been different (but PCRE had them first :-).
1298
1299 16. Related to 15 above: Perl does not treat the | in a conditional group as
1300 creating alternatives. Such a group is treated in the same way as an
1301 ordinary group without any | characters when processing (*THEN). PCRE has
1302 been changed to match Perl's behaviour.
1303
1304 17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
1305 RunGrepTest script failed.
1306
1307 18. Change 22 for version 13 caused atomic groups to use more stack. This is
1308 inevitable for groups that contain captures, but it can lead to a lot of
1309 stack use in large patterns. The old behaviour has been restored for atomic
1310 groups that do not contain any capturing parentheses.
1311
1312 19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
1313 suppress the check for a minimum subject length at run time. (If it was
1314 given to pcre_exec() or pcre_dfa_exec() it did work.)
1315
1316 20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
1317 fail to work when decoding hex characters in data strings in EBCDIC
1318 environments.
1319
1320 21. It appears that in at least one Mac OS environment, the isxdigit() function
1321 is implemented as a macro that evaluates to its argument more than once,
1322 contravening the C 90 Standard (I haven't checked a later standard). There
1323 was an instance in pcretest which caused it to go wrong when processing
1324 \x{...} escapes in subject strings. The has been rewritten to avoid using
1325 things like p++ in the argument of isxdigit().
1326
1327
1328 Version 8.13 16-Aug-2011
1329 ------------------------
1330
1331 1. The Unicode data tables have been updated to Unicode 6.0.0.
1332
1333 2. Two minor typos in pcre_internal.h have been fixed.
1334
1335 3. Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
1336 pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
1337 in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
1338
1339 4. There were a number of related bugs in the code for matching backrefences
1340 caselessly in UTF-8 mode when codes for the characters concerned were
1341 different numbers of bytes. For example, U+023A and U+2C65 are an upper
1342 and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
1343 (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
1344 code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
1345 2-byte code at the end of the subject (it thought there wasn't enough data
1346 left).
1347
1348 5. Comprehensive information about what went wrong is now returned by
1349 pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
1350 as the output vector has at least 2 elements. The offset of the start of
1351 the failing character and a reason code are placed in the vector.
1352
1353 6. When the UTF-8 string check fails for pcre_compile(), the offset that is
1354 now returned is for the first byte of the failing character, instead of the
1355 last byte inspected. This is an incompatible change, but I hope it is small
1356 enough not to be a problem. It makes the returned offset consistent with
1357 pcre_exec() and pcre_dfa_exec().
1358
1359 7. pcretest now gives a text phrase as well as the error number when
1360 pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
1361 failure, the offset and reason code are output.
1362
1363 8. When \R was used with a maximizing quantifier it failed to skip backwards
1364 over a \r\n pair if the subsequent match failed. Instead, it just skipped
1365 back over a single character (\n). This seems wrong (because it treated the
1366 two characters as a single entity when going forwards), conflicts with the
1367 documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
1368 behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
1369 has been changed.
1370
1371 9. Some internal refactoring has changed the processing so that the handling
1372 of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
1373 time (the PCRE_DOTALL option was changed this way some time ago: version
1374 7.7 change 16). This has made it possible to abolish the OP_OPT op code,
1375 which was always a bit of a fudge. It also means that there is one less
1376 argument for the match() function, which reduces its stack requirements
1377 slightly. This change also fixes an incompatibility with Perl: the pattern
1378 (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
1379
1380 10. More internal refactoring has drastically reduced the number of recursive
1381 calls to match() for possessively repeated groups such as (abc)++ when
1382 using pcre_exec().
1383
1384 11. While implementing 10, a number of bugs in the handling of groups were
1385 discovered and fixed:
1386
1387 (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
1388 (a|)*(?1) gave a compile-time internal error.
1389 ((a|)+)+ did not notice that the outer group could match an empty string.
1390 (^a|^)+ was not marked as anchored.
1391 (.*a|.*)+ was not marked as matching at start or after a newline.
1392
1393 12. Yet more internal refactoring has removed another argument from the match()
1394 function. Special calls to this function are now indicated by setting a
1395 value in a variable in the "match data" data block.
1396
1397 13. Be more explicit in pcre_study() instead of relying on "default" for
1398 opcodes that mean there is no starting character; this means that when new
1399 ones are added and accidentally left out of pcre_study(), testing should
1400 pick them up.
1401
1402 14. The -s option of pcretest has been documented for ages as being an old
1403 synonym of -m (show memory usage). I have changed it to mean "force study
1404 for every regex", that is, assume /S for every regex. This is similar to -i
1405 and -d etc. It's slightly incompatible, but I'm hoping nobody is still
1406 using it. It makes it easier to run collections of tests with and without
1407 study enabled, and thereby test pcre_study() more easily. All the standard
1408 tests are now run with and without -s (but some patterns can be marked as
1409 "never study" - see 20 below).
1410
1411 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
1412 restoration of the capturing data to the outer values was not happening
1413 correctly.
1414
1415 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
1416 empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
1417 pattern had matched an empty string, and so incorrectly returned a no
1418 match.
1419
1420 17. There was optimizing code for the last branch of non-capturing parentheses,
1421 and also for the obeyed branch of a conditional subexpression, which used
1422 tail recursion to cut down on stack usage. Unfortunately, now that there is
1423 the possibility of (*THEN) occurring in these branches, tail recursion is
1424 no longer possible because the return has to be checked for (*THEN). These
1425 two optimizations have therefore been removed. [But see 8.20/11 above.]
1426
1427 18. If a pattern containing \R was studied, it was assumed that \R always
1428 matched two bytes, thus causing the minimum subject length to be
1429 incorrectly computed because \R can also match just one byte.
1430
1431 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
1432 was incorrectly computed.
1433
1434 20. If /S is present twice on a test pattern in pcretest input, it now
1435 *disables* studying, thereby overriding the use of -s on the command line
1436 (see 14 above). This is necessary for one or two tests to keep the output
1437 identical in both cases.
1438
1439 21. When (*ACCEPT) was used in an assertion that matched an empty string and
1440 PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
1441
1442 22. When an atomic group that contained a capturing parenthesis was
1443 successfully matched, but the branch in which it appeared failed, the
1444 capturing was not being forgotten if a higher numbered group was later
1445 captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
1446 group 1 to "a", when in fact it should be unset. This applied to multi-
1447 branched capturing and non-capturing groups, repeated or not, and also to
1448 positive assertions (capturing in negative assertions does not happen
1449 in PCRE) and also to nested atomic groups.
1450
1451 23. Add the ++ qualifier feature to pcretest, to show the remainder of the
1452 subject after a captured substring, to make it easier to tell which of a
1453 number of identical substrings has been captured.
1454
1455 24. The way atomic groups are processed by pcre_exec() has been changed so that
1456 if they are repeated, backtracking one repetition now resets captured
1457 values correctly. For example, if ((?>(a+)b)+aabab) is matched against
1458 "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
1459 "aaa". Previously, it would have been "a". As part of this code
1460 refactoring, the way recursive calls are handled has also been changed.
1461
1462 25. If an assertion condition captured any substrings, they were not passed
1463 back unless some other capturing happened later. For example, if
1464 (?(?=(a))a) was matched against "a", no capturing was returned.
1465
1466 26. When studying a pattern that contained subroutine calls or assertions,
1467 the code for finding the minimum length of a possible match was handling
1468 direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
1469 group 1 called group 2 while simultaneously a separate group 2 called group
1470 1). A stack overflow occurred in this case. I have fixed this by limiting
1471 the recursion depth to 10.
1472
1473 27. Updated RunTest.bat in the distribution to the version supplied by Tom
1474 Fortmann. This supports explicit test numbers on the command line, and has
1475 argument validation and error reporting.
1476
1477 28. An instance of \X with an unlimited repeat could fail if at any point the
1478 first character it looked at was a mark character.
1479
1480 29. Some minor code refactoring concerning Unicode properties and scripts
1481 should reduce the stack requirement of match() slightly.
1482
1483 30. Added the '=' option to pcretest to check the setting of unused capturing
1484 slots at the end of the pattern, which are documented as being -1, but are
1485 not included in the return count.
1486
1487 31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
1488 compiled something random. Now it gives a compile-time error (as does
1489 Perl).
1490
1491 32. A *MARK encountered during the processing of a positive assertion is now
1492 recorded and passed back (compatible with Perl).
1493
1494 33. If --only-matching or --colour was set on a pcregrep call whose pattern
1495 had alternative anchored branches, the search for a second match in a line
1496 was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
1497 matched the line "0102" twice. The same bug affected patterns that started
1498 with a backwards assertion. For example /\b01|\b02/ also matched "0102"
1499 twice.
1500
1501 34. Previously, PCRE did not allow quantification of assertions. However, Perl
1502 does, and because of capturing effects, quantifying parenthesized
1503 assertions may at times be useful. Quantifiers are now allowed for
1504 parenthesized assertions.
1505
1506 35. A minor code tidy in pcre_compile() when checking options for \R usage.
1507
1508 36. \g was being checked for fancy things in a character class, when it should
1509 just be a literal "g".
1510
1511 37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
1512 appearance of a nested POSIX class supersedes an apparent external class.
1513 For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
1514 unescaped square brackets may also appear as part of class names. For
1515 example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
1516 more like Perl. (But see 8.20/1 above.)
1517
1518 38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
1519 was because it thought it was \N{name}, which is not supported).
1520
1521 39. Add minix to OS list not supporting the -S option in pcretest.
1522
1523 40. PCRE tries to detect cases of infinite recursion at compile time, but it
1524 cannot analyze patterns in sufficient detail to catch mutual recursions
1525 such as ((?1))((?2)). There is now a runtime test that gives an error if a
1526 subgroup is called recursively as a subpattern for a second time at the
1527 same position in the subject string. In previous releases this might have
1528 been caught by the recursion limit, or it might have run out of stack.
1529
1530 41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
1531 happen only once. PCRE was, however incorrectly giving a compile time error
1532 "recursive call could loop indefinitely" because it cannot analyze the
1533 pattern in sufficient detail. The compile time test no longer happens when
1534 PCRE is compiling a conditional subpattern, but actual runaway loops are
1535 now caught at runtime (see 40 above).
1536
1537 42. It seems that Perl allows any characters other than a closing parenthesis
1538 to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
1539 has been changed to be the same.
1540
1541 43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
1542 as not to get warnings when autogen.sh is called. Also changed
1543 AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
1544
1545 44. To help people who use pcregrep to scan files containing exceedingly long
1546 lines, the following changes have been made:
1547
1548 (a) The default value of the buffer size parameter has been increased from
1549 8K to 20K. (The actual buffer used is three times this size.)
1550
1551 (b) The default can be changed by ./configure --with-pcregrep-bufsize when
1552 PCRE is built.
1553
1554 (c) A --buffer-size=n option has been added to pcregrep, to allow the size
1555 to be set at run time.
1556
1557 (d) Numerical values in pcregrep options can be followed by K or M, for
1558 example --buffer-size=50K.
1559
1560 (e) If a line being scanned overflows pcregrep's buffer, an error is now
1561 given and the return code is set to 2.
1562
1563 45. Add a pointer to the latest mark to the callout data block.
1564
1565 46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
1566 partial match of an empty string instead of no match. This was specific to
1567 the use of ".".
1568
1569 47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
1570 complete match instead of a partial match. This bug was dependent on both
1571 the PCRE_UTF8 and PCRE_DOTALL options being set.
1572
1573 48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
1574 starting byte set, because \b was not being ignored.
1575
1576
1577 Version 8.12 15-Jan-2011
1578 ------------------------
1579
1580 1. Fixed some typos in the markup of the man pages, and wrote a script that
1581 checks for such things as part of the documentation building process.
1582
1583 2. On a big-endian 64-bit system, pcregrep did not correctly process the
1584 --match-limit and --recursion-limit options (added for 8.11). In
1585 particular, this made one of the standard tests fail. (The integer value
1586 went into the wrong half of a long int.)
1587
1588 3. If the --colour option was given to pcregrep with -v (invert match), it
1589 did strange things, either producing crazy output, or crashing. It should,
1590 of course, ignore a request for colour when reporting lines that do not
1591 match.
1592
1593 4. Another pcregrep bug caused similar problems if --colour was specified with
1594 -M (multiline) and the pattern match finished with a line ending.
1595
1596 5. In pcregrep, when a pattern that ended with a literal newline sequence was
1597 matched in multiline mode, the following line was shown as part of the
1598 match. This seems wrong, so I have changed it.
1599
1600 6. Another pcregrep bug in multiline mode, when --colour was specified, caused
1601 the check for further matches in the same line (so they could be coloured)
1602 to overrun the end of the current line. If another match was found, it was
1603 incorrectly shown (and then shown again when found in the next line).
1604
1605 7. If pcregrep was compiled under Windows, there was a reference to the
1606 function pcregrep_exit() before it was defined. I am assuming this was
1607 the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
1608 reported by a user. I've moved the definition above the reference.
1609
1610
1611 Version 8.11 10-Dec-2010
1612 ------------------------
1613
1614 1. (*THEN) was not working properly if there were untried alternatives prior
1615 to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
1616 backtracked to try for "b" instead of moving to the next alternative branch
1617 at the same level (in this case, to look for "c"). The Perl documentation
1618 is clear that when (*THEN) is backtracked onto, it goes to the "next
1619 alternative in the innermost enclosing group".
1620
1621 2. (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
1622 such as (A(*COMMIT)B(*THEN)C|D) any failure after matching A should
1623 result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
1624 (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
1625 (*THEN).
1626
1627 3. If \s appeared in a character class, it removed the VT character from
1628 the class, even if it had been included by some previous item, for example
1629 in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
1630 of \s, but is part of the POSIX "space" class.)
1631
1632 4. A partial match never returns an empty string (because you can always
1633 match an empty string at the end of the subject); however the checking for
1634 an empty string was starting at the "start of match" point. This has been
1635 changed to the "earliest inspected character" point, because the returned
1636 data for a partial match starts at this character. This means that, for
1637 example, /(?<=abc)def/ gives a partial match for the subject "abc"
1638 (previously it gave "no match").
1639
1640 5. Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
1641 of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
1642 previously a full match would be given. However, setting PCRE_PARTIAL_HARD
1643 has an implication that the given string is incomplete (because a partial
1644 match is preferred over a full match). For this reason, these items now
1645 give a partial match in this situation. [Aside: previously, the one case
1646 /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
1647 match rather than a full match, which was wrong by the old rules, but is
1648 now correct.]
1649
1650 6. There was a bug in the handling of #-introduced comments, recognized when
1651 PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
1652 If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
1653 UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
1654 scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
1655 but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
1656 places in pcre_compile().
1657
1658 7. Related to (6) above, when pcre_compile() was skipping #-introduced
1659 comments when looking ahead for named forward references to subpatterns,
1660 the only newline sequence it recognized was NL. It now handles newlines
1661 according to the set newline convention.
1662
1663 8. SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
1664 former, but used strtoul(), whereas pcretest avoided strtoul() but did not
1665 cater for a lack of strerror(). These oversights have been fixed.
1666
1667 9. Added --match-limit and --recursion-limit to pcregrep.
1668
1669 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
1670
1671 11. When the -o option was used, pcregrep was setting a return code of 1, even
1672 when matches were found, and --line-buffered was not being honoured.
1673
1674 12. Added an optional parentheses number to the -o and --only-matching options
1675 of pcregrep.
1676
1677 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
1678 can match an empty string. The code to do it in pcretest and pcredemo
1679 needed fixing:
1680
1681 (a) When the newline convention was "crlf", pcretest got it wrong, skipping
1682 only one byte after an empty string match just before CRLF (this case
1683 just got forgotten; "any" and "anycrlf" were OK).
1684
1685 (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
1686 mode when an empty string match preceded an ASCII character followed by
1687 a non-ASCII character. (The code for advancing by one character rather
1688 than one byte was nonsense.)
1689
1690 (c) The pcredemo.c sample program did not have any code at all to handle
1691 the cases when CRLF is a valid newline sequence.
1692
1693 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
1694 as a starting offset was within the subject string. There is now a new
1695 error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
1696 negative or greater than the length of the string. In order to test this,
1697 pcretest is extended to allow the setting of negative starting offsets.
1698
1699 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
1700 starting offset points to the beginning of a UTF-8 character was
1701 unnecessarily clumsy. I tidied it up.
1702
1703 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
1704 bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
1705
1706 17. Nobody had reported that the --include_dir option, which was added in
1707 release 7.7 should have been called --include-dir (hyphen, not underscore)
1708 for compatibility with GNU grep. I have changed it to --include-dir, but
1709 left --include_dir as an undocumented synonym, and the same for
1710 --exclude-dir, though that is not available in GNU grep, at least as of
1711 release 2.5.4.
1712
1713 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
1714 characters from a string of bytes) have been redefined so as not to use
1715 loops, in order to improve performance in some environments. At the same
1716 time, I abstracted some of the common code into auxiliary macros to save
1717 repetition (this should not affect the compiled code).
1718
1719 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
1720 compile-time error is now given if \c is not followed by an ASCII
1721 character, that is, a byte less than 128. (In EBCDIC mode, the code is
1722 different, and any byte value is allowed.)
1723
1724 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
1725 START_OPTIMIZE option, which is now allowed at compile time - but just
1726 passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
1727 to pcregrep and other applications that have no direct access to PCRE
1728 options. The new /Y option in pcretest sets this option when calling
1729 pcre_compile().
1730
1731 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
1732 back references. Groups containing recursive back references were forced to
1733 be atomic by that change, but in the case of named groups, the amount of
1734 memory required was incorrectly computed, leading to "Failed: internal
1735 error: code overflow". This has been fixed.
1736
1737 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
1738 pcretest.c, to avoid build problems in some Borland environments.
1739
1740
1741 Version 8.10 25-Jun-2010
1742 ------------------------
1743
1744 1. Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
1745 THEN.
1746
1747 2. (*ACCEPT) was not working when inside an atomic group.
1748
1749 3. Inside a character class, \B is treated as a literal by default, but
1750 faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
1751 causes the error). The code is unchanged, but I tidied the documentation.
1752
1753 4. Inside a character class, PCRE always treated \R and \X as literals,
1754 whereas Perl faults them if its -w option is set. I have changed PCRE so
1755 that it faults them when PCRE_EXTRA is set.
1756
1757 5. Added support for \N, which always matches any character other than
1758 newline. (It is the same as "." when PCRE_DOTALL is not set.)
1759
1760 6. When compiling pcregrep with newer versions of gcc which may have
1761 FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
1762 declared with attribute warn_unused_result" were given. Just casting the
1763 result to (void) does not stop the warnings; a more elaborate fudge is
1764 needed. I've used a macro to implement this.
1765
1766 7. Minor change to pcretest.c to avoid a compiler warning.
1767
1768 8. Added four artifical Unicode properties to help with an option to make
1769 \s etc use properties (see next item). The new properties are: Xan
1770 (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
1771
1772 9. Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
1773 use Unicode properties. (*UCP) at the start of a pattern can be used to set
1774 this option. Modified pcretest to add /W to test this facility. Added
1775 REG_UCP to make it available via the POSIX interface.
1776
1777 10. Added --line-buffered to pcregrep.
1778
1779 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
1780 studied, and the match started with a letter with a code point greater than
1781 127 whose first byte was different to the first byte of the other case of
1782 the letter, the other case of this starting letter was not recognized
1783 (#976).
1784
1785 12. If a pattern that was studied started with a repeated Unicode property
1786 test, for example, \p{Nd}+, there was the theoretical possibility of
1787 setting up an incorrect bitmap of starting bytes, but fortunately it could
1788 not have actually happened in practice until change 8 above was made (it
1789 added property types that matched character-matching opcodes).
1790
1791 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
1792 possible starting bytes for non-anchored patterns.
1793
1794 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
1795 \R, and also a number of cases that involve Unicode properties, both
1796 explicit and implicit when PCRE_UCP is set.
1797
1798 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
1799 input, it could crash or give wrong results if characters with values
1800 greater than 0xc0 were present in the subject string. (Detail: it assumed
1801 UTF-8 input when processing these items.)
1802
1803 16. Added a lot of (int) casts to avoid compiler warnings in systems where
1804 size_t is 64-bit (#991).
1805
1806 17. Added a check for running out of memory when PCRE is compiled with
1807 --disable-stack-for-recursion (#990).
1808
1809 18. If the last data line in a file for pcretest does not have a newline on
1810 the end, a newline was missing in the output.
1811
1812 19. The default pcre_chartables.c file recognizes only ASCII characters (values
1813 less than 128) in its various bitmaps. However, there is a facility for
1814 generating tables according to the current locale when PCRE is compiled. It
1815 turns out that in some environments, 0x85 and 0xa0, which are Unicode space
1816 characters, are recognized by isspace() and therefore were getting set in
1817 these tables, and indeed these tables seem to approximate to ISO 8859. This
1818 caused a problem in UTF-8 mode when pcre_study() was used to create a list
1819 of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
1820 which of course cannot start UTF-8 characters. I have changed the code so
1821 that only real ASCII characters (less than 128) and the correct starting
1822 bytes for UTF-8 encodings are set for characters greater than 127 when in
1823 UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
1824 altogether.)
1825
1826 20. Added the /T option to pcretest so as to be able to run tests with non-
1827 standard character tables, thus making it possible to include the tests
1828 used for 19 above in the standard set of tests.
1829
1830 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
1831 reference to a subpattern the other side of a comment that contains an
1832 opening parenthesis caused either an internal compiling error, or a
1833 reference to the wrong subpattern.
1834
1835
1836 Version 8.02 19-Mar-2010
1837 ------------------------
1838
1839 1. The Unicode data tables have been updated to Unicode 5.2.0.
1840
1841 2. Added the option --libs-cpp to pcre-config, but only when C++ support is
1842 configured.
1843
1844 3. Updated the licensing terms in the pcregexp.pas file, as agreed with the
1845 original author of that file, following a query about its status.
1846
1847 4. On systems that do not have stdint.h (e.g. Solaris), check for and include
1848 inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
1849
1850 5. A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
1851 quantifier applied to a forward-referencing subroutine call, could compile
1852 incorrect code or give the error "internal error: previously-checked
1853 referenced subpattern not found".
1854
1855 6. Both MS Visual Studio and Symbian OS have problems with initializing
1856 variables to point to external functions. For these systems, therefore,
1857 pcre_malloc etc. are now initialized to local functions that call the
1858 relevant global functions.
1859
1860 7. There were two entries missing in the vectors called coptable and poptable
1861 in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
1862 I've fixed the data, and added a kludgy way of testing at compile time that
1863 the lengths are correct (equal to the number of opcodes).
1864
1865 8. Following on from 7, I added a similar kludge to check the length of the
1866 eint vector in pcreposix.c.
1867
1868 9. Error texts for pcre_compile() are held as one long string to avoid too
1869 much relocation at load time. To find a text, the string is searched,
1870 counting zeros. There was no check for running off the end of the string,
1871 which could happen if a new error number was added without updating the
1872 string.
1873
1874 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
1875
1876 11. \K was not working if it appeared in an atomic group or in a group that
1877 was called as a "subroutine", or in an assertion. Perl 5.11 documents that
1878 \K is "not well defined" if used in an assertion. PCRE now accepts it if
1879 the assertion is positive, but not if it is negative.
1880
1881 12. Change 11 fortuitously reduced the size of the stack frame used in the
1882 "match()" function of pcre_exec.c by one pointer. Forthcoming
1883 implementation of support for (*MARK) will need an extra pointer on the
1884 stack; I have reserved it now, so that the stack frame size does not
1885 decrease.
1886
1887 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
1888 item in branch that calls a recursion is a subroutine call - as in the
1889 second branch in the above example - was incorrectly given the compile-
1890 time error "recursive call could loop indefinitely" because pcre_compile()
1891 was not correctly checking the subroutine for matching a non-empty string.
1892
1893 14. The checks for overrunning compiling workspace could trigger after an
1894 overrun had occurred. This is a "should never occur" error, but it can be
1895 triggered by pathological patterns such as hundreds of nested parentheses.
1896 The checks now trigger 100 bytes before the end of the workspace.
1897
1898 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
1899
1900
1901 Version 8.01 19-Jan-2010
1902 ------------------------
1903
1904 1. If a pattern contained a conditional subpattern with only one branch (in
1905 particular, this includes all (*DEFINE) patterns), a call to pcre_study()
1906 computed the wrong minimum data length (which is of course zero for such
1907 subpatterns). This could cause incorrect "no match" results.
1908
1909 2. For patterns such as (?i)a(?-i)b|c where an option setting at the start of
1910 the pattern is reset in the first branch, pcre_compile() failed with
1911 "internal error: code overflow at offset...". This happened only when
1912 the reset was to the original external option setting. (An optimization
1913 abstracts leading options settings into an external setting, which was the
1914 cause of this.)
1915
1916 3. A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
1917 of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
1918 assertion pattern did not match (meaning that the assertion was true), it
1919 was incorrectly treated as false if the SKIP had been reached during the
1920 matching. This also applied to assertions used as conditions.
1921
1922 4. If an item that is not supported by pcre_dfa_exec() was encountered in an
1923 assertion subpattern, including such a pattern used as a condition,
1924 unpredictable results occurred, instead of the error return
1925 PCRE_ERROR_DFA_UITEM.
1926
1927 5. The C++ GlobalReplace function was not working like Perl for the special
1928 situation when an empty string is matched. It now does the fancy magic
1929 stuff that is necessary.
1930
1931 6. In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
1932 removed. (These were left over from very, very early versions of PCRE.)
1933
1934 7. Some cosmetic changes to the code to make life easier when compiling it
1935 as part of something else:
1936
1937 (a) Change DEBUG to PCRE_DEBUG.
1938
1939 (b) In pcre_compile(), rename the member of the "branch_chain" structure
1940 called "current" as "current_branch", to prevent a collision with the
1941 Linux macro when compiled as a kernel module.
1942
1943 (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
1944 prevent a collision with the Linux macro when compiled as a kernel
1945 module.
1946
1947 8. In pcre_compile() there are some checks for integer overflows that used to
1948 cast potentially large values to (double). This has been changed to that
1949 when building, a check for int64_t is made, and if it is found, it is used
1950 instead, thus avoiding the use of floating point arithmetic. (There is no
1951 other use of FP in PCRE.) If int64_t is not found, the fallback is to
1952 double.
1953
1954 9. Added two casts to avoid signed/unsigned warnings from VS Studio Express
1955 2005 (difference between two addresses compared to an unsigned value).
1956
1957 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
1958 custom one, because of the following reported problem in Windows:
1959
1960 - libbz2 uses the Pascal calling convention (WINAPI) for the functions
1961 under Win32.
1962 - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
1963 therefore missing the function definition.
1964 - The compiler thus generates a "C" signature for the test function.
1965 - The linker fails to find the "C" function.
1966 - PCRE fails to configure if asked to do so against libbz2.
1967
1968 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
1969 messages were output:
1970
1971 Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
1972 rerunning libtoolize, to keep the correct libtool macros in-tree.
1973 Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
1974
1975 I have done both of these things.
1976
1977 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
1978 most of the time, it *can* run out if it is given a pattern that contains a
1979 runaway infinite recursion. I updated the discussion in the pcrestack man
1980 page.
1981
1982 13. Now that we have gone to the x.xx style of version numbers, the minor
1983 version may start with zero. Using 08 or 09 is a bad idea because users
1984 might check the value of PCRE_MINOR in their code, and 08 or 09 may be
1985 interpreted as invalid octal numbers. I've updated the previous comment in
1986 configure.ac, and also added a check that gives an error if 08 or 09 are
1987 used.
1988
1989 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
1990 causing partial matching to fail when the end of the subject matched \W
1991 in a UTF-8 pattern where \W was quantified with a minimum of 3.
1992
1993 15. There were some discrepancies between the declarations in pcre_internal.h
1994 of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
1995 their definitions. The declarations used "const uschar *" and the
1996 definitions used USPTR. Even though USPTR is normally defined as "const
1997 unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
1998 reported that: "This difference in casting confuses some C++ compilers, for
1999 example, SunCC recognizes above declarations as different functions and
2000 generates broken code for hbpcre." I have changed the declarations to use
2001 USPTR.
2002
2003 16. GNU libtool is named differently on some systems. The autogen.sh script now
2004 tries several variants such as glibtoolize (MacOSX) and libtoolize1x
2005 (FreeBSD).
2006
2007 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
2008 (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
2009 comment: "Figure out how to create a longlong from a string: strtoll and
2010 equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
2011 instance, but it only takes 2 args instead of 3!"
2012
2013 18. A subtle bug concerned with back references has been fixed by a change of
2014 specification, with a corresponding code fix. A pattern such as
2015 ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
2016 refers, was giving matches when it shouldn't. For example, xa=xaaa would
2017 match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
2018 same bug. Such groups have to be quantified to be useful, or contained
2019 inside another quantified group. (If there's no repetition, the reference
2020 can never match.) The problem arises because, having left the group and
2021 moved on to the rest of the pattern, a later failure that backtracks into
2022 the group uses the captured value from the final iteration of the group
2023 rather than the correct earlier one. I have fixed this in PCRE by forcing
2024 any group that contains a reference to itself to be an atomic group; that
2025 is, there cannot be any backtracking into it once it has completed. This is
2026 similar to recursive and subroutine calls.
2027
2028
2029 Version 8.00 19-Oct-09
2030 ----------------------
2031
2032 1. The table for translating pcre_compile() error codes into POSIX error codes
2033 was out-of-date, and there was no check on the pcre_compile() error code
2034 being within the table. This could lead to an OK return being given in
2035 error.
2036
2037 2. Changed the call to open a subject file in pcregrep from fopen(pathname,
2038 "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
2039 in a Windows environment.
2040
2041 3. The pcregrep --count option prints the count for each file even when it is
2042 zero, as does GNU grep. However, pcregrep was also printing all files when
2043 --files-with-matches was added. Now, when both options are given, it prints
2044 counts only for those files that have at least one match. (GNU grep just
2045 prints the file name in this circumstance, but including the count seems
2046 more useful - otherwise, why use --count?) Also ensured that the
2047 combination -clh just lists non-zero counts, with no names.
2048
2049 4. The long form of the pcregrep -F option was incorrectly implemented as
2050 --fixed_strings instead of --fixed-strings. This is an incompatible change,
2051 but it seems right to fix it, and I didn't think it was worth preserving
2052 the old behaviour.
2053
2054 5. The command line items --regex=pattern and --regexp=pattern were not
2055 recognized by pcregrep, which required --regex pattern or --regexp pattern
2056 (with a space rather than an '='). The man page documented the '=' forms,
2057 which are compatible with GNU grep; these now work.
2058
2059 6. No libpcreposix.pc file was created for pkg-config; there was just
2060 libpcre.pc and libpcrecpp.pc. The omission has been rectified.
2061
2062 7. Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
2063 when UCP support is not needed, by modifying the Python script that
2064 generates it from Unicode data files. This should not matter if the module
2065 is correctly used as a library, but I received one complaint about 50K of
2066 unwanted data. My guess is that the person linked everything into his
2067 program rather than using a library. Anyway, it does no harm.
2068
2069 8. A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
2070 was a minimum greater than 1 for a wide character in a possessive
2071 repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
2072 which had an unlimited repeat of a nested, fixed maximum repeat of a wide
2073 character. Chaos in the form of incorrect output or a compiling loop could
2074 result.
2075
2076 9. The restrictions on what a pattern can contain when partial matching is
2077 requested for pcre_exec() have been removed. All patterns can now be
2078 partially matched by this function. In addition, if there are at least two
2079 slots in the offset vector, the offset of the earliest inspected character
2080 for the match and the offset of the end of the subject are set in them when
2081 PCRE_ERROR_PARTIAL is returned.
2082
2083 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
2084 synonymous with PCRE_PARTIAL, for backwards compatibility, and
2085 PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
2086 and may be more useful for multi-segment matching.
2087
2088 11. Partial matching with pcre_exec() is now more intuitive. A partial match
2089 used to be given if ever the end of the subject was reached; now it is
2090 given only if matching could not proceed because another character was
2091 needed. This makes a difference in some odd cases such as Z(*FAIL) with the
2092 string "Z", which now yields "no match" instead of "partial match". In the
2093 case of pcre_dfa_exec(), "no match" is given if every matching path for the
2094 final character ended with (*FAIL).
2095
2096 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
2097 if the pattern had a "must contain" character that was already found in the
2098 earlier partial match, unless partial matching was again requested. For
2099 example, with the pattern /dog.(body)?/, the "must contain" character is
2100 "g". If the first part-match was for the string "dog", restarting with
2101 "sbody" failed. This bug has been fixed.
2102
2103 13. The string returned by pcre_dfa_exec() after a partial match has been
2104 changed so that it starts at the first inspected character rather than the
2105 first character of the match. This makes a difference only if the pattern
2106 starts with a lookbehind assertion or \b or \B (\K is not supported by
2107 pcre_dfa_exec()). It's an incompatible change, but it makes the two
2108 matching functions compatible, and I think it's the right thing to do.
2109
2110 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
2111 so that the demonstration program is easily available in environments where
2112 PCRE has not been installed from source.
2113
2114 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
2115 libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
2116 library.
2117
2118 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
2119 It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
2120 is not the first non-POSIX option to be added. Clearly some people find
2121 these options useful.
2122
2123 17. If a caller to the POSIX matching function regexec() passes a non-zero
2124 value for nmatch with a NULL value for pmatch, the value of
2125 nmatch is forced to zero.
2126
2127 18. RunGrepTest did not have a test for the availability of the -u option of
2128 the diff command, as RunTest does. It now checks in the same way as
2129 RunTest, and also checks for the -b option.
2130
2131 19. If an odd number of negated classes containing just a single character
2132 interposed, within parentheses, between a forward reference to a named
2133 subpattern and the definition of the subpattern, compilation crashed with
2134 an internal error, complaining that it could not find the referenced
2135 subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
2136 [The bug was that it was starting one character too far in when skipping
2137 over the character class, thus treating the ] as data rather than
2138 terminating the class. This meant it could skip too much.]
2139
2140 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
2141 /g option in pcretest when the pattern contains \K, which makes it possible
2142 to have an empty string match not at the start, even when the pattern is
2143 anchored. Updated pcretest and pcredemo to use this option.
2144
2145 21. If the maximum number of capturing subpatterns in a recursion was greater
2146 than the maximum at the outer level, the higher number was returned, but
2147 with unset values at the outer level. The correct (outer level) value is
2148 now given.
2149
2150 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
2151 PCRE did not set those parentheses (unlike Perl). I have now found a way to
2152 make it do so. The string so far is captured, making this feature
2153 compatible with Perl.
2154
2155 23. The tests have been re-organized, adding tests 11 and 12, to make it
2156 possible to check the Perl 5.10 features against Perl 5.10.
2157
2158 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
2159 pattern matches a fixed length string. PCRE did not allow this; now it
2160 does. Neither allows recursion.
2161
2162 25. I finally figured out how to implement a request to provide the minimum
2163 length of subject string that was needed in order to match a given pattern.
2164 (It was back references and recursion that I had previously got hung up
2165 on.) This code has now been added to pcre_study(); it finds a lower bound
2166 to the length of subject needed. It is not necessarily the greatest lower
2167 bound, but using it to avoid searching strings that are too short does give
2168 some useful speed-ups. The value is available to calling programs via
2169 pcre_fullinfo().
2170
2171 26. While implementing 25, I discovered to my embarrassment that pcretest had
2172 not been passing the result of pcre_study() to pcre_dfa_exec(), so the
2173 study optimizations had never been tested with that matching function.
2174 Oops. What is worse, even when it was passed study data, there was a bug in
2175 pcre_dfa_exec() that meant it never actually used it. Double oops. There
2176 were also very few tests of studied patterns with pcre_dfa_exec().
2177
2178 27. If (?| is used to create subpatterns with duplicate numbers, they are now
2179 allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
2180 on the other side of the coin, they are no longer allowed to have different
2181 names, because these cannot be distinguished in PCRE, and this has caused
2182 confusion. (This is a difference from Perl.)
2183
2184 28. When duplicate subpattern names are present (necessarily with different
2185 numbers, as required by 27 above), and a test is made by name in a
2186 conditional pattern, either for a subpattern having been matched, or for
2187 recursion in such a pattern, all the associated numbered subpatterns are
2188 tested, and the overall condition is true if the condition is true for any
2189 one of them. This is the way Perl works, and is also more like the way
2190 testing by number works.
2191
2192
2193 Version 7.9 11-Apr-09
2194 ---------------------
2195
2196 1. When building with support for bzlib/zlib (pcregrep) and/or readline
2197 (pcretest), all targets were linked against these libraries. This included
2198 libpcre, libpcreposix, and libpcrecpp, even though they do not use these
2199 libraries. This caused unwanted dependencies to be created. This problem
2200 has been fixed, and now only pcregrep is linked with bzlib/zlib and only
2201 pcretest is linked with readline.
2202
2203 2. The "typedef int BOOL" in pcre_internal.h that was included inside the
2204 "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
2205 moved outside it again, because FALSE and TRUE are already defined in AIX,
2206 but BOOL is not.
2207
2208 3. The pcre_config() function was treating the PCRE_MATCH_LIMIT and
2209 PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
2210
2211 4. The pcregrep documentation said spaces were inserted as well as colons (or
2212 hyphens) following file names and line numbers when outputting matching
2213 lines. This is not true; no spaces are inserted. I have also clarified the
2214 wording for the --colour (or --color) option.
2215
2216 5. In pcregrep, when --colour was used with -o, the list of matching strings
2217 was not coloured; this is different to GNU grep, so I have changed it to be
2218 the same.
2219
2220 6. When --colo(u)r was used in pcregrep, only the first matching substring in
2221 each matching line was coloured. Now it goes on to look for further matches
2222 of any of the test patterns, which is the same behaviour as GNU grep.
2223
2224 7. A pattern that could match an empty string could cause pcregrep to loop; it
2225 doesn't make sense to accept an empty string match in pcregrep, so I have
2226 locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
2227 seems to be how GNU grep behaves. [But see later change 40 for release
2228 8.33.]
2229
2230 8. The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
2231 start or after a newline", because the conditional assertion was not being
2232 correctly handled. The rule now is that both the assertion and what follows
2233 in the first alternative must satisfy the test.
2234
2235 9. If auto-callout was enabled in a pattern with a conditional group whose
2236 condition was an assertion, PCRE could crash during matching, both with
2237 pcre_exec() and pcre_dfa_exec().
2238
2239 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
2240 used for matching.
2241
2242 11. Unicode property support in character classes was not working for
2243 characters (bytes) greater than 127 when not in UTF-8 mode.
2244
2245 12. Added the -M command line option to pcretest.
2246
2247 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
2248
2249 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
2250
2251 16. Added comments and documentation about mis-use of no_arg in the C++
2252 wrapper.
2253
2254 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
2255 from Martin Jerabek that uses macro names for all relevant character and
2256 string constants.
2257
2258 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
2259 SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
2260 SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
2261 these, but not everybody uses configure.
2262
2263 19. A conditional group that had only one branch was not being correctly
2264 recognized as an item that could match an empty string. This meant that an
2265 enclosing group might also not be so recognized, causing infinite looping
2266 (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
2267 with the subject "ab", where knowledge that the repeated group can match
2268 nothing is needed in order to break the loop.
2269
2270 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
2271 exec(), but without supplying a callout function, matching went wrong.
2272
2273 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
2274 leak if the size of the offset vector was greater than 30. When the vector
2275 is smaller, the saved offsets during recursion go onto a local stack
2276 vector, but for larger vectors malloc() is used. It was failing to free
2277 when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
2278 error, in fact).
2279
2280 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
2281 heapframe that is used only when UTF-8 support is enabled. This caused no
2282 problem, but was untidy.
2283
2284 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
2285 CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
2286 included within another project.
2287
2288 24. Steven Van Ingelgem's patches to add more options to the CMake support,
2289 slightly modified by me:
2290
2291 (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
2292 not building pcregrep.
2293
2294 (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
2295 if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
2296
2297 25. Forward references, both numeric and by name, in patterns that made use of
2298 duplicate group numbers, could behave incorrectly or give incorrect errors,
2299 because when scanning forward to find the reference group, PCRE was not
2300 taking into account the duplicate group numbers. A pattern such as
2301 ^X(?3)(a)(?|(b)|(q))(Y) is an example.
2302
2303 26. Changed a few more instances of "const unsigned char *" to USPTR, making
2304 the feature of a custom pointer more persuasive (as requested by a user).
2305
2306 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
2307 pcretest.c, inside #ifndefs, because it seems they are sometimes already
2308 pre-defined.
2309
2310 28. Added support for (*UTF8) at the start of a pattern.
2311
2312 29. Arrange for flags added by the "release type" setting in CMake to be shown
2313 in the configuration summary.
2314
2315
2316 Version 7.8 05-Sep-08
2317 ---------------------
2318
2319 1. Replaced UCP searching code with optimized version as implemented for Ad
2320 Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
2321 stage table and inline lookup instead of a function, giving speed ups of 2
2322 to 5 times on some simple patterns that I tested. Permission was given to
2323 distribute the MultiStage2.py script that generates the tables (it's not in
2324 the tarball, but is in the Subversion repository).
2325
2326 2. Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
2327 scripts.
2328
2329 3. Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
2330 a group with a zero qualifier. The result of the study could be incorrect,
2331 or the function might crash, depending on the pattern.
2332
2333 4. Caseless matching was not working for non-ASCII characters in back
2334 references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
2335 It now works when Unicode Property Support is available.
2336
2337 5. In pcretest, an escape such as \x{de} in the data was always generating
2338 a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
2339 non-UTF-8 mode. If the value is greater than 255, it gives a warning about
2340 truncation.
2341
2342 6. Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
2343
2344 7. Added two (int) casts to pcregrep when printing the difference of two
2345 pointers, in case they are 64-bit values.
2346
2347 8. Added comments about Mac OS X stack usage to the pcrestack man page and to
2348 test 2 if it fails.
2349
2350 9. Added PCRE_CALL_CONVENTION just before the names of all exported functions,
2351 and a #define of that name to empty if it is not externally set. This is to
2352 allow users of MSVC to set it if necessary.
2353
2354 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
2355 the convenience functions in the pcre_get.c source file.
2356
2357 11. An option change at the start of a pattern that had top-level alternatives
2358 could cause overwriting and/or a crash. This command provoked a crash in
2359 some environments:
2360
2361 printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
2362
2363 This potential security problem was recorded as CVE-2008-2371.
2364
2365 12. For a pattern where the match had to start at the beginning or immediately
2366 after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
2367 pcre_dfa_exec() could read past the end of the passed subject if there was
2368 no match. To help with detecting such bugs (e.g. with valgrind), I modified
2369 pcretest so that it places the subject at the end of its malloc-ed buffer.
2370
2371 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
2372 exec() might read past the end of the data buffer in UTF-8 mode.
2373
2374 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
2375 the data contained the byte 0x85 as part of a UTF-8 character within its
2376 first line. This applied both to normal and DFA matching.
2377
2378 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
2379 /^[^d]*?$/8 failed to match "abc".
2380
2381 16. Added a missing copyright notice to pcrecpp_internal.h.
2382
2383 17. Make it more clear in the documentation that values returned from
2384 pcre_exec() in ovector are byte offsets, not character counts.
2385
2386 18. Tidied a few places to stop certain compilers from issuing warnings.
2387
2388 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
2389 supplied by Stefan Weber. I made a further small update for 7.8 because
2390 there is a change of source arrangements: the pcre_searchfuncs.c module is
2391 replaced by pcre_ucd.c.
2392
2393
2394 Version 7.7 07-May-08
2395 ---------------------
2396
2397 1. Applied Craig's patch to sort out a long long problem: "If we can't convert
2398 a string to a long long, pretend we don't even have a long long." This is
2399 done by checking for the strtoq, strtoll, and _strtoi64 functions.
2400
2401 2. Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
2402 pre-7.6 versions, which defined a global no_arg variable instead of putting
2403 it in the RE class. (See also #8 below.)
2404
2405 3. Remove a line of dead code, identified by coverity and reported by Nuno
2406 Lopes.
2407
2408 4. Fixed two related pcregrep bugs involving -r with --include or --exclude:
2409
2410 (1) The include/exclude patterns were being applied to the whole pathnames
2411 of files, instead of just to the final components.
2412
2413 (2) If there was more than one level of directory, the subdirectories were
2414 skipped unless they satisfied the include/exclude conditions. This is
2415 inconsistent with GNU grep (and could even be seen as contrary to the
2416 pcregrep specification - which I improved to make it absolutely clear).
2417 The action now is always to scan all levels of directory, and just
2418 apply the include/exclude patterns to regular files.
2419
2420 5. Added the --include_dir and --exclude_dir patterns to pcregrep, and used
2421 --exclude_dir in the tests to avoid scanning .svn directories.
2422
2423 6. Applied Craig's patch to the QuoteMeta function so that it escapes the
2424 NUL character as backslash + 0 rather than backslash + NUL, because PCRE
2425 doesn't support NULs in patterns.
2426
2427 7. Added some missing "const"s to declarations of static tables in
2428 pcre_compile.c and pcre_dfa_exec.c.
2429
2430 8. Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
2431 caused by fix #2 above. (Subsequently also a second patch to fix the
2432 first patch. And a third patch - this was a messy problem.)
2433
2434 9. Applied Craig's patch to remove the use of push_back().
2435
2436 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
2437 matching function regexec().
2438
2439 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
2440 which, however, unlike Perl's \g{...}, are subroutine calls, not back
2441 references. PCRE supports relative numbers with this syntax (I don't think
2442 Oniguruma does).
2443
2444 12. Previously, a group with a zero repeat such as (...){0} was completely
2445 omitted from the compiled regex. However, this means that if the group
2446 was called as a subroutine from elsewhere in the pattern, things went wrong
2447 (an internal error was given). Such groups are now left in the compiled
2448 pattern, with a new opcode that causes them to be skipped at execution
2449 time.
2450
2451 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
2452 to the way PCRE behaves:
2453
2454 (a) A lone ] character is dis-allowed (Perl treats it as data).
2455
2456 (b) A back reference to an unmatched subpattern matches an empty string
2457 (Perl fails the current match path).
2458
2459 (c) A data ] in a character class must be notated as \] because if the
2460 first data character in a class is ], it defines an empty class. (In
2461 Perl it is not possible to have an empty class.) The empty class []
2462 never matches; it forces failure and is equivalent to (*FAIL) or (?!).
2463 The negative empty class [^] matches any one character, independently
2464 of the DOTALL setting.
2465
2466 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
2467 non-existent subpattern following a character class starting with ']' and
2468 containing () gave an internal compiling error instead of "reference to
2469 non-existent subpattern". Fortunately, when the pattern did exist, the
2470 compiled code was correct. (When scanning forwards to check for the
2471 existence of the subpattern, it was treating the data ']' as terminating
2472 the class, so got the count wrong. When actually compiling, the reference
2473 was subsequently set up correctly.)
2474
2475 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
2476 it was being rejected as not supported by pcre_dfa_exec(), even though
2477 other assertions are supported. I have made pcre_dfa_exec() support
2478 (*FAIL).
2479
2480 16. The implementation of 13c above involved the invention of a new opcode,
2481 OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
2482 cannot be changed at match time, I realized I could make a small
2483 improvement to matching performance by compiling OP_ALLANY instead of
2484 OP_ANY for "." when DOTALL was set, and then removing the runtime tests
2485 on the OP_ANY path.
2486
2487 17. Compiling pcretest on Windows with readline support failed without the
2488 following two fixes: (1) Make the unistd.h include conditional on
2489 HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
2490
2491 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
2492 ncurses library to be included for pcretest when ReadLine support is
2493 requested, but also to allow for it to be overridden. This patch came from
2494 Daniel Bergström.
2495
2496 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
2497 as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
2498 any errors with the current Unicode tables. Thanks to Peter Kankowski for
2499 spotting this.
2500
2501
2502 Version 7.6 28-Jan-08
2503 ---------------------
2504
2505 1. A character class containing a very large number of characters with
2506 codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
2507 overflow.
2508
2509 2. Patch to cut out the "long long" test in pcrecpp_unittest when
2510 HAVE_LONG_LONG is not defined.
2511
2512 3. Applied Christian Ehrlicher's patch to update the CMake build files to
2513 bring them up to date and include new features. This patch includes:
2514
2515 - Fixed PH's badly added libz and libbz2 support.
2516 - Fixed a problem with static linking.
2517 - Added pcredemo. [But later removed - see 7 below.]
2518 - Fixed dftables problem and added an option.
2519 - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
2520 HAVE_LONG_LONG.
2521 - Added readline support for pcretest.
2522 - Added an listing of the option settings after cmake has run.
2523
2524 4. A user submitted a patch to Makefile that makes it easy to create
2525 "pcre.dll" under mingw when using Configure/Make. I added stuff to
2526 Makefile.am that cause it to include this special target, without
2527 affecting anything else. Note that the same mingw target plus all
2528 the other distribution libraries and programs are now supported
2529 when configuring with CMake (see 6 below) instead of with
2530 Configure/Make.
2531
2532 5. Applied Craig's patch that moves no_arg into the RE class in the C++ code.
2533 This is an attempt to solve the reported problem "pcrecpp::no_arg is not
2534 exported in the Windows port". It has not yet been confirmed that the patch
2535 solves the problem, but it does no harm.
2536
2537 6. Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
2538 NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
2539 with CMake, and also correct the comment about stack recursion.
2540
2541 7. Remove the automatic building of pcredemo from the ./configure system and
2542 from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
2543 of a program that users should build themselves after PCRE is installed, so
2544 building it automatically is not really right. What is more, it gave
2545 trouble in some build environments.
2546
2547 8. Further tidies to CMakeLists.txt from Sheri and Christian.
2548
2549
2550 Version 7.5 10-Jan-08
2551 ---------------------
2552
2553 1. Applied a patch from Craig: "This patch makes it possible to 'ignore'
2554 values in parens when parsing an RE using the C++ wrapper."
2555
2556 2. Negative specials like \S did not work in character classes in UTF-8 mode.
2557 Characters greater than 255 were excluded from the class instead of being
2558 included.
2559
2560 3. The same bug as (2) above applied to negated POSIX classes such as
2561 [:^space:].
2562
2563 4. PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
2564 defined or documented. It seems to have been a typo for PCRE_STATIC, so
2565 I have changed it.
2566
2567 5. The construct (?&) was not diagnosed as a syntax error (it referenced the
2568 first named subpattern) and a construct such as (?&a) would reference the
2569 first named subpattern whose name started with "a" (in other words, the
2570 length check was missing). Both these problems are fixed. "Subpattern name
2571 expected" is now given for (?&) (a zero-length name), and this patch also
2572 makes it give the same error for \k'' (previously it complained that that
2573 was a reference to a non-existent subpattern).
2574
2575 6. The erroneous patterns (?+-a) and (?-+a) give different error messages;
2576 this is right because (?- can be followed by option settings as well as by
2577 digits. I have, however, made the messages clearer.
2578
2579 7. Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
2580 than the number used in the conditional) now cause a compile-time error.
2581 This is actually not compatible with Perl, which accepts such patterns, but
2582 treats the conditional as always being FALSE (as PCRE used to), but it
2583 seems to me that giving a diagnostic is better.
2584
2585 8. Change "alphameric" to the more common word "alphanumeric" in comments
2586 and messages.
2587
2588 9. Fix two occurrences of "backslash" in comments that should have been
2589 "backspace".
2590
2591 10. Remove two redundant lines of code that can never be obeyed (their function
2592 was moved elsewhere).
2593
2594 11. The program that makes PCRE's Unicode character property table had a bug
2595 which caused it to generate incorrect table entries for sequences of
2596 characters that have the same character type, but are in different scripts.
2597 It amalgamated them into a single range, with the script of the first of
2598 them. In other words, some characters were in the wrong script. There were
2599 thirteen such cases, affecting characters in the following ranges:
2600
2601 U+002b0 - U+002c1
2602 U+0060c - U+0060d
2603 U+0061e - U+00612
2604 U+0064b - U+0065e
2605 U+0074d - U+0076d
2606 U+01800 - U+01805
2607 U+01d00 - U+01d77
2608 U+01d9b - U+01dbf
2609 U+0200b - U+0200f
2610 U+030fc - U+030fe
2611 U+03260 - U+0327f
2612 U+0fb46 - U+0fbb1
2613 U+10450 - U+1049d
2614
2615 12. The -o option (show only the matching part of a line) for pcregrep was not
2616 compatible with GNU grep in that, if there was more than one match in a
2617 line, it showed only the first of them. It now behaves in the same way as
2618 GNU grep.
2619
2620 13. If the -o and -v options were combined for pcregrep, it printed a blank
2621 line for every non-matching line. GNU grep prints nothing, and pcregrep now
2622 does the same. The return code can be used to tell if there were any
2623 non-matching lines.
2624
2625 14. Added --file-offsets and --line-offsets to pcregrep.
2626
2627 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
2628 infinitely looping recursion. The bug was that positive lookaheads were not
2629 being skipped when checking for a possible empty match (negative lookaheads
2630 and both kinds of lookbehind were skipped).
2631
2632 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
2633 inclusion of <windows.h> to before rather than after the definition of
2634 INVALID_FILE_ATTRIBUTES (patch from David Byron).
2635
2636 17. Specifying a possessive quantifier with a specific limit for a Unicode
2637 character property caused pcre_compile() to compile bad code, which led at
2638 runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
2639 are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
2640 caused the error; without that there was no problem.
2641
2642 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
2643
2644 19. Added --enable-pcretest-libreadline.
2645
2646 20. In pcrecpp.cc, the variable 'count' was incremented twice in
2647 RE::GlobalReplace(). As a result, the number of replacements returned was
2648 double what it should be. I removed one of the increments, but Craig sent a
2649 later patch that removed the other one (the right fix) and added unit tests
2650 that check the return values (which was not done before).
2651
2652 21. Several CMake things:
2653
2654 (1) Arranged that, when cmake is used on Unix, the libraries end up with
2655 the names libpcre and libpcreposix, not just pcre and pcreposix.
2656
2657 (2) The above change means that pcretest and pcregrep are now correctly
2658 linked with the newly-built libraries, not previously installed ones.
2659
2660 (3) Added PCRE_SUPPORT_LIBREADLINE, PCRE_SUPPORT_LIBZ, PCRE_SUPPORT_LIBBZ2.
2661
2662 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
2663 crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
2664 UTF-8 newline character). The key issue is that the pattern starts .*;
2665 this means that the match must be either at the beginning, or after a
2666 newline. The bug was in the code for advancing after a failed match and
2667 checking that the new position followed a newline. It was not taking
2668 account of UTF-8 characters correctly.
2669
2670 23. PCRE was behaving differently from Perl in the way it recognized POSIX
2671 character classes. PCRE was not treating the sequence [:...:] as a
2672 character class unless the ... were all letters. Perl, however, seems to
2673 allow any characters between [: and :], though of course it rejects as
2674 unknown any "names" that contain non-letters, because all the known class
2675 names consist only of letters. Thus, Perl gives an error for [[:1234:]],
2676 for example, whereas PCRE did not - it did not recognize a POSIX character
2677 class. This seemed a bit dangerous, so the code has been changed to be
2678 closer to Perl. The behaviour is not identical to Perl, because PCRE will
2679 diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
2680 treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
2681 Perl does, and where it didn't before.
2682
2683 24. Rewrite so as to remove the single use of %n from pcregrep because in some
2684 Windows environments %n is disabled by default.
2685
2686
2687 Version 7.4 21-Sep-07
2688 ---------------------
2689
2690 1. Change 7.3/28 was implemented for classes by looking at the bitmap. This
2691 means that a class such as [\s] counted as "explicit reference to CR or
2692 LF". That isn't really right - the whole point of the change was to try to
2693 help when there was an actual mention of one of the two characters. So now
2694 the change happens only if \r or \n (or a literal CR or LF) character is
2695 encountered.
2696
2697 2. The 32-bit options word was also used for 6 internal flags, but the numbers
2698 of both had grown to the point where there were only 3 bits left.
2699 Fortunately, there was spare space in the data structure, and so I have
2700 moved the internal flags into a new 16-bit field to free up more option
2701 bits.
2702
2703 3. The appearance of (?J) at the start of a pattern set the DUPNAMES option,
2704 but did not set the internal JCHANGED flag - either of these is enough to
2705 control the way the "get" function works - but the PCRE_INFO_JCHANGED
2706 facility is supposed to tell if (?J) was ever used, so now (?J) at the
2707 start sets both bits.
2708
2709 4. Added options (at build time, compile time, exec time) to change \R from
2710 matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
2711
2712 5. doc/pcresyntax.html was missing from the distribution.
2713
2714 6. Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
2715 compatibility, even though it is no longer used.
2716
2717 7. Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
2718 strtoull to pcrecpp.cc to select the available functions in WIN32 when the
2719 windows.h file is present (where different names are used). [This was
2720 reversed later after testing - see 16 below.]
2721
2722 8. Changed all #include <config.h> to #include "config.h". There were also
2723 some further <pcre.h> cases that I changed to "pcre.h".
2724
2725 9. When pcregrep was used with the --colour option, it missed the line ending
2726 sequence off the lines that it output.
2727
2728 10. It was pointed out to me that arrays of string pointers cause lots of
2729 relocations when a shared library is dynamically loaded. A technique of
2730 using a single long string with a table of offsets can drastically reduce
2731 these. I have refactored PCRE in four places to do this. The result is
2732 dramatic:
2733
2734 Originally: 290
2735 After changing UCP table: 187
2736 After changing error message table: 43
2737 After changing table of "verbs" 36
2738 After changing table of Posix names 22
2739
2740 Thanks to the folks working on Gregex for glib for this insight.
2741
2742 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
2743 unicode-properties was also set.
2744
2745 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
2746
2747 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
2748 checked only for CRLF.
2749
2750 14. Added casts to pcretest.c to avoid compiler warnings.
2751
2752 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
2753
2754 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
2755 and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
2756 entirely. This removes changes made in 7 above.
2757
2758 17. The CMake files have been updated, and there is now more information about
2759 building with CMake in the NON-UNIX-USE document.
2760
2761
2762 Version 7.3 28-Aug-07
2763 ---------------------
2764
2765 1. In the rejigging of the build system that eventually resulted in 7.1, the
2766 line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
2767 brackets there is not right, since it causes compilers to look for an
2768 installed pcre.h, not the version that is in the source that is being
2769 compiled (which of course may be different). I have changed it back to:
2770
2771 #include "pcre.h"
2772
2773 I have a vague recollection that the change was concerned with compiling in
2774 different directories, but in the new build system, that is taken care of
2775 by the VPATH setting the Makefile.
2776
2777 2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
2778 when the subject happened to end in the byte 0x85 (e.g. if the last
2779 character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
2780 characters but of course it shouldn't be taken as a newline when it is part
2781 of another character. The bug was that, for an unlimited repeat of . in
2782 not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
2783 characters when looking for a newline.
2784
2785 3. A small performance improvement in the DOTALL UTF-8 mode .* case.
2786
2787 4. Debugging: adjusted the names of opcodes for different kinds of parentheses
2788 in debug output.
2789
2790 5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
2791 long printing in the pcrecpp unittest when running under MinGW.
2792
2793 6. ESC_K was left out of the EBCDIC table.
2794
2795 7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
2796 parentheses; I made it 1000, which seemed large enough. Unfortunately, the
2797 limit also applies to "virtual nesting" when a pattern is recursive, and in
2798 this case 1000 isn't so big. I have been able to remove this limit at the
2799 expense of backing off one optimization in certain circumstances. Normally,
2800 when pcre_exec() would call its internal match() function recursively and
2801 immediately return the result unconditionally, it uses a "tail recursion"
2802 feature to save stack. However, when a subpattern that can match an empty
2803 string has an unlimited repetition quantifier, it no longer makes this
2804 optimization. That gives it a stack frame in which to save the data for
2805 checking that an empty string has been matched. Previously this was taken
2806 from the 1000-entry workspace that had been reserved. So now there is no
2807 explicit limit, but more stack is used.
2808
2809 8. Applied Daniel's patches to solve problems with the import/export magic
2810 syntax that is required for Windows, and which was going wrong for the
2811 pcreposix and pcrecpp parts of the library. These were overlooked when this
2812 problem was solved for the main library.
2813
2814 9. There were some crude static tests to avoid integer overflow when computing
2815 the size of patterns that contain repeated groups with explicit upper
2816 limits. As the maximum quantifier is 65535, the maximum group length was
2817 set at 30,000 so that the product of these two numbers did not overflow a
2818 32-bit integer. However, it turns out that people want to use groups that
2819 are longer than 30,000 bytes (though not repeat them that many times).
2820 Change 7.0/17 (the refactoring of the way the pattern size is computed) has
2821 made it possible to implement the integer overflow checks in a much more
2822 dynamic way, which I have now done. The artificial limitation on group
2823 length has been removed - we now have only the limit on the total length of
2824 the compiled pattern, which depends on the LINK_SIZE setting.
2825
2826 10. Fixed a bug in the documentation for get/copy named substring when
2827 duplicate names are permitted. If none of the named substrings are set, the
2828 functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
2829 empty string.
2830
2831 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
2832 instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
2833 because the ] is interpreted as the first data character and the
2834 terminating ] is not found. PCRE has been made compatible with Perl in this
2835 regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
2836 cause memory overwriting.
2837
2838 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
2839 string has been matched (to stop an infinite loop). It was not recognizing
2840 a conditional subpattern that could match an empty string if that
2841 subpattern was within another subpattern. For example, it looped when
2842 trying to match (((?(1)X|))*) but it was OK with ((?(1)X|)*) where the
2843 condition was not nested. This bug has been fixed.
2844
2845 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
2846 past the start of the subject in the presence of bytes with the top bit
2847 set, for example "\x8aBCD".
2848
2849 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
2850 (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
2851
2852 14. Optimized (?!) to (*FAIL).
2853
2854 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
2855 This restricts code points to be within the range 0 to 0x10FFFF, excluding
2856 the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
2857 full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
2858 does: it's just the validity check that is more restrictive.
2859
2860 16. Inserted checks for integer overflows during escape sequence (backslash)
2861 processing, and also fixed erroneous offset values for syntax errors during
2862 backslash processing.
2863
2864 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
2865 for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
2866
2867 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
2868 caused an overrun.
2869
2870 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
2871 something other than just ASCII characters) inside a group that had an
2872 unlimited repeat caused a loop at compile time (while checking to see
2873 whether the group could match an empty string).
2874
2875 20. Debugging a pattern containing \p or \P could cause a crash. For example,
2876 [\P{Any}] did so. (Error in the code for printing property names.)
2877
2878 21. An orphan \E inside a character class could cause a crash.
2879
2880 22. A repeated capturing bracket such as (A)? could cause a wild memory
2881 reference during compilation.
2882
2883 23. There are several functions in pcre_compile() that scan along a compiled
2884 expression for various reasons (e.g. to see if it's fixed length for look
2885 behind). There were bugs in these functions when a repeated \p or \P was
2886 present in the pattern. These operators have additional parameters compared
2887 with \d, etc, and these were not being taken into account when moving along
2888 the compiled data. Specifically:
2889
2890 (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
2891 length.
2892
2893 (b) An item such as \pL+ within a repeated group could cause crashes or
2894 loops.
2895
2896 (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
2897 "reference to non-existent subpattern" error.
2898
2899 (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
2900
2901 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
2902 characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
2903
2904 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
2905
2906 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
2907 character were causing crashes (broken optimization).
2908
2909 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
2910 \p or \P) caused a compile-time loop.
2911
2912 28. More problems have arisen in unanchored patterns when CRLF is a valid line
2913 break. For example, the unstudied pattern [\r\n]A does not match the string
2914 "\r\nA" because change 7.0/46 below moves the current point on by two
2915 characters after failing to match at the start. However, the pattern \nA
2916 *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
2917 the same is true. There doesn't seem any very clean way out of this, but
2918 what I have chosen to do makes the common cases work: PCRE now takes note
2919 of whether there can be an explicit match for \r or \n anywhere in the
2920 pattern, and if so, 7.0/46 no longer applies. As part of this change,
2921 there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
2922 pattern has explicit CR or LF references.
2923
2924 29. Added (*CR) etc for changing newline setting at start of pattern.
2925
2926
2927 Version 7.2 19-Jun-07
2928 ---------------------
2929
2930 1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
2931 which is apparently normally available under Windows.
2932
2933 2. Re-jig the pcregrep tests with different newline settings in an attempt
2934 to make them independent of the local environment's newline setting.
2935
2936 3. Add code to configure.ac to remove -g from the CFLAGS default settings.
2937
2938 4. Some of the "internals" tests were previously cut out when the link size
2939 was not 2, because the output contained actual offsets. The recent new
2940 "Z" feature of pcretest means that these can be cut out, making the tests
2941 usable with all link sizes.
2942
2943 5. Implemented Stan Switzer's goto replacement for longjmp() when not using
2944 stack recursion. This gives a massive performance boost under BSD, but just
2945 a small improvement under Linux. However, it saves one field in the frame
2946 in all cases.
2947
2948 6. Added more features from the forthcoming Perl 5.10:
2949
2950 (a) (?-n) (where n is a string of digits) is a relative subroutine or
2951 recursion call. It refers to the nth most recently opened parentheses.
2952
2953 (b) (?+n) is also a relative subroutine call; it refers to the nth next
2954 to be opened parentheses.
2955
2956 (c) Conditions that refer to capturing parentheses can be specified
2957 relatively, for example, (?(-2)... or (?(+3)...
2958
2959 (d) \K resets the start of the current match so that everything before
2960 is not part of it.
2961
2962 (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
2963
2964 (f) \g{name} is another synonym - part of Perl 5.10's unification of
2965 reference syntax.
2966
2967 (g) (?| introduces a group in which the numbering of parentheses in each
2968 alternative starts with the same number.
2969
2970 (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
2971
2972 7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
2973 PCRE_INFO_JCHANGED.
2974
2975 8. A pattern such as (.*(.)?)* caused pcre_exec() to fail by either not
2976 terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
2977 for detecting groups that can match an empty string.
2978
2979 9. A pattern with a very large number of alternatives (more than several
2980 hundred) was running out of internal workspace during the pre-compile
2981 phase, where pcre_compile() figures out how much memory will be needed. A
2982 bit of new cunning has reduced the workspace needed for groups with
2983 alternatives. The 1000-alternative test pattern now uses 12 bytes of
2984 workspace instead of running out of the 4096 that are available.
2985
2986 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
2987
2988 11. Applied patch from Google to remove an optimization that didn't quite work.
2989 The report of the bug said:
2990
2991 pcrecpp::RE("a*").FullMatch("aaa") matches, while
2992 pcrecpp::RE("a*?").FullMatch("aaa") does not, and
2993 pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
2994
2995 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
2996 it matched the wrong number of bytes.
2997
2998
2999 Version 7.1 24-Apr-07
3000 ---------------------
3001
3002 1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
3003 that is more "standard", making use of automake and other Autotools. There
3004 is some re-arrangement of the files and adjustment of comments consequent
3005 on this.
3006
3007 2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
3008 for recursive directory scanning broke on some systems because the files
3009 are not scanned in any specific order and on different systems the order
3010 was different. A call to "sort" has been inserted into RunGrepTest for the
3011 approprate test as a short-term fix. In the longer term there may be an
3012 alternative.
3013
3014 3. I had an email from Eric Raymond about problems translating some of PCRE's
3015 man pages to HTML (despite the fact that I distribute HTML pages, some
3016 people do their own conversions for various reasons). The problems
3017 concerned the use of low-level troff macros .br and .in. I have therefore
3018 removed all such uses from the man pages (some were redundant, some could
3019 be replaced by .nf/.fi pairs). The 132html script that I use to generate
3020 HTML has been updated to handle .nf/.fi and to complain if it encounters
3021 .br or .in.
3022
3023 4. Updated comments in configure.ac that get placed in config.h.in and also
3024 arranged for config.h to be included in the distribution, with the name
3025 config.h.generic, for the benefit of those who have to compile without
3026 Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
3027
3028 5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
3029 Weber: (1) pcre_internal.h was missing some function renames; (2) updated
3030 makevp.bat for the current PCRE, using the additional files
3031 makevp_c.txt, makevp_l.txt, and pcregexp.pas.
3032
3033 6. A Windows user reported a minor discrepancy with test 2, which turned out
3034 to be caused by a trailing space on an input line that had got lost in his
3035 copy. The trailing space was an accident, so I've just removed it.
3036
3037 7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
3038 that is needed.
3039
3040 8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
3041 as "const" (a) because they are and (b) because it helps the PHP
3042 maintainers who have recently made a script to detect big data structures
3043 in the php code that should be moved to the .rodata section. I remembered
3044 to update Builducptable as well, so it won't revert if ucptable.h is ever
3045 re-created.
3046
3047 9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
3048 pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
3049 order to be able to cut out the UTF-8 tables in the latter when UTF-8
3050 support is not required. This saves 1.5-2K of code, which is important in
3051 some applications.
3052
3053 Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
3054 so as not to refer to the tables, even though these functions will never be
3055 called when UTF-8 support is disabled. Otherwise there are problems with a
3056 shared library.
3057
3058 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
3059
3060 (a) It was defining its arguments as char * instead of void *.
3061
3062 (b) It was assuming that all moves were upwards in memory; this was true
3063 a long time ago when I wrote it, but is no longer the case.
3064
3065 The emulated memove() is provided for those environments that have neither
3066 memmove() nor bcopy(). I didn't think anyone used it these days, but that
3067 is clearly not the case, as these two bugs were recently reported.
3068
3069 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
3070 and Detrail to create the HTML documentation, the .txt form of the man
3071 pages, and it removes trailing spaces from listed files. It also creates
3072 pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
3073 case, it wraps all the #defines with #ifndefs. This script should be run
3074 before "make dist".
3075
3076 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
3077 with Unicode property support.
3078
3079 (a) For a maximizing quantifier, if the two different cases of the
3080 character were of different lengths in their UTF-8 codings (there are
3081 some cases like this - I found 11), and the matching function had to
3082 back up over a mixture of the two cases, it incorrectly assumed they
3083 were both the same length.
3084
3085 (b) When PCRE was configured to use the heap rather than the stack for
3086 recursion during matching, it was not correctly preserving the data for
3087 the other case of a UTF-8 character when checking ahead for a match
3088 while processing a minimizing repeat. If the check also involved
3089 matching a wide character, but failed, corruption could cause an
3090 erroneous result when trying to check for a repeat of the original
3091 character.
3092
3093 13. Some tidying changes to the testing mechanism:
3094
3095 (a) The RunTest script now detects the internal link size and whether there
3096 is UTF-8 and UCP support by running ./pcretest -C instead of relying on
3097 values substituted by "configure". (The RunGrepTest script already did
3098 this for UTF-8.) The configure.ac script no longer substitutes the
3099 relevant variables.
3100
3101 (b) The debugging options /B and /D in pcretest show the compiled bytecode
3102 with length and offset values. This means that the output is different
3103 for different internal link sizes. Test 2 is skipped for link sizes
3104 other than 2 because of this, bypassing the problem. Unfortunately,
3105 there was also a test in test 3 (the locale tests) that used /B and
3106 failed for link sizes other than 2. Rather than cut the whole test out,
3107 I have added a new /Z option to pcretest that replaces the length and
3108 offset values with spaces. This is now used to make test 3 independent
3109 of link size. (Test 2 will be tidied up later.)
3110
3111 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
3112 segmentation fault instead of returning the appropriate error message.
3113
3114 15. In multiline mode when the newline sequence was set to "any", the pattern
3115 ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
3116 This doesn't seem right; it now treats the CRLF combination as the line
3117 ending, and so does not match in that case. It's only a pattern such as ^$
3118 that would hit this one: something like ^ABC$ would have failed after \r
3119 and then tried again after \r\n.
3120
3121 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
3122 in an attempt to make files that differ only in their line terminators
3123 compare equal. This works on Linux.
3124
3125 17. Under certain error circumstances pcregrep might try to free random memory
3126 as it exited. This is now fixed, thanks to valgrind.
3127
3128 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
3129 "abc\r\n\r\n", it found an unwanted second match after the second \r. This
3130 was because its rules for how to advance for /g after matching an empty
3131 string at the end of a line did not allow for this case. They now check for
3132 it specially.
3133
3134 20. pcretest is supposed to handle patterns and data of any length, by
3135 extending its buffers when necessary. It was getting this wrong when the
3136 buffer for a data line had to be extended.
3137
3138 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
3139 CRLF as a newline sequence.
3140
3141 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
3142 out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
3143 I have nevertheless tidied it up.
3144
3145 23. Added some casts to kill warnings from HP-UX ia64 compiler.
3146
3147 24. Added a man page for pcre-config.
3148
3149
3150 Version 7.0 19-Dec-06
3151 ---------------------
3152
3153 1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
3154 moving to gcc 4.1.1.
3155
3156 2. The -S option for pcretest uses setrlimit(); I had omitted to #include
3157 sys/time.h, which is documented as needed for this function. It doesn't
3158 seem to matter on Linux, but it showed up on some releases of OS X.
3159
3160 3. It seems that there are systems where bytes whose values are greater than
3161 127 match isprint() in the "C" locale. The "C" locale should be the
3162 default when a C program starts up. In most systems, only ASCII printing
3163 characters match isprint(). This difference caused the output from pcretest
3164 to vary, making some of the tests fail. I have changed pcretest so that:
3165
3166 (a) When it is outputting text in the compiled version of a pattern, bytes
3167 other than 32-126 are always shown as hex escapes.
3168
3169 (b) When it is outputting text that is a matched part of a subject string,
3170 it does the same, unless a different locale has been set for the match
3171 (using the /L modifier). In this case, it uses isprint() to decide.
3172
3173 4. Fixed a major bug that caused incorrect computation of the amount of memory
3174 required for a compiled pattern when options that changed within the
3175 pattern affected the logic of the preliminary scan that determines the
3176 length. The relevant options are -x, and -i in UTF-8 mode. The result was
3177 that the computed length was too small. The symptoms of this bug were
3178 either the PCRE error "internal error: code overflow" from pcre_compile(),
3179 or a glibc crash with a message such as "pcretest: free(): invalid next
3180 size (fast)". Examples of patterns that provoked this bug (shown in
3181 pcretest format) are:
3182
3183 /(?-x: )/x
3184 /(?x)(?-x: \s*#\s*)/
3185 /((?i)[\x{c0}])/8
3186 /(?i:[\x{c0}])/8
3187
3188 HOWEVER: Change 17 below makes this fix obsolete as the memory computation
3189 is now done differently.
3190
3191 5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
3192 wrapper classes; (b) implement a new function in the C++ scanner that is
3193 more efficient than the old way of doing things because it avoids levels of
3194 recursion in the regex matching; (c) add a paragraph to the documentation
3195 for the FullMatch() function.
3196
3197 6. The escape sequence \n was being treated as whatever was defined as
3198 "newline". Not only was this contrary to the documentation, which states
3199 that \n is character 10 (hex 0A), but it also went horribly wrong when
3200 "newline" was defined as CRLF. This has been fixed.
3201
3202 7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
3203 was being set to -1 for the "end of line" case (supposedly a value that no
3204 character can have). Though this value is never used (the check for end of
3205 line is "zero bytes in current character"), it caused compiler complaints.
3206 I've changed it to 0xffffffff.
3207
3208 8. In pcre_version.c, the version string was being built by a sequence of
3209 C macros that, in the event of PCRE_PRERELEASE being defined as an empty
3210 string (as it is for production releases) called a macro with an empty
3211 argument. The C standard says the result of this is undefined. The gcc
3212 compiler treats it as an empty string (which was what was wanted) but it is
3213 reported that Visual C gives an error. The source has been hacked around to
3214 avoid this problem.
3215
3216 9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
3217 builds of pcretest, and changed the call to _setmode() to use _O_BINARY
3218 instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
3219 of them did).
3220
3221 10. Originally, pcretest opened its input and output without "b"; then I was
3222 told that "b" was needed in some environments, so it was added for release
3223 5.0 to both the input and output. (It makes no difference on Unix-like
3224 systems.) Later I was told that it is wrong for the input on Windows. I've
3225 now abstracted the modes into two macros, to make it easier to fiddle with
3226 them, and removed "b" from the input mode under Windows.
3227
3228 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
3229
3230 12. Added -help and --help to pcretest as an official way of being reminded
3231 of the options.
3232
3233 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
3234 and pcrecpp.cc because they annoy compilers at high warning levels.
3235
3236 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
3237
3238 15. Fixed an occurrence of == in configure.ac that should have been = (shell
3239 scripts are not C programs :-) and which was not noticed because it works
3240 on Linux.
3241
3242 16. pcretest is supposed to handle any length of pattern and data line (as one
3243 line or as a continued sequence of lines) by extending its input buffer if
3244 necessary. This feature was broken for very long pattern lines, leading to
3245 a string of junk being passed to pcre_compile() if the pattern was longer
3246 than about 50K.
3247
3248 17. I have done a major re-factoring of the way pcre_compile() computes the
3249 amount of memory needed for a compiled pattern. Previously, there was code
3250 that made a preliminary scan of the pattern in order to do this. That was
3251 OK when PCRE was new, but as the facilities have expanded, it has become
3252 harder and harder to keep it in step with the real compile phase, and there
3253 have been a number of bugs (see for example, 4 above). I have now found a
3254 cunning way of running the real compile function in a "fake" mode that
3255 enables it to compute how much memory it would need, while actually only
3256 ever using a few hundred bytes of working memory and without too many
3257 tests of the mode. This should make future maintenance and development
3258 easier. A side effect of this work is that the limit of 200 on the nesting
3259 depth of parentheses has been removed (though this was never a serious
3260 limitation, I suspect). However, there is a downside: pcre_compile() now
3261 runs more slowly than before (30% or more, depending on the pattern). I
3262 hope this isn't a big issue. There is no effect on runtime performance.
3263
3264 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
3265 newline (only possible for the last line of a file) and it was a
3266 pattern that set a locale (followed by /Lsomething), pcretest crashed.
3267
3268 19. Added additional timing features to pcretest. (1) The -tm option now times
3269 matching only, not compiling. (2) Both -t and -tm can be followed, as a
3270 separate command line item, by a number that specifies the number of
3271 repeats to use when timing. The default is 50000; this gives better
3272 precision, but takes uncomfortably long for very large patterns.
3273
3274 20. Extended pcre_study() to be more clever in cases where a branch of a
3275 subpattern has no definite first character. For example, (a*|b*)[cd] would
3276 previously give no result from pcre_study(). Now it recognizes that the
3277 first character must be a, b, c, or d.
3278
3279 21. There was an incorrect error "recursive call could loop indefinitely" if
3280 a subpattern (or the entire pattern) that was being tested for matching an
3281 empty string contained only one non-empty item after a nested subpattern.
3282 For example, the pattern (?>\x{100}*)\d(?R) provoked this error
3283 incorrectly, because the \d was being skipped in the check.
3284
3285 22. The pcretest program now has a new pattern option /B and a command line
3286 option -b, which is equivalent to adding /B to every pattern. This causes
3287 it to show the compiled bytecode, without the additional information that
3288 -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
3289 is the same as /B/I).
3290
3291 23. A new optimization is now able automatically to treat some sequences such
3292 as a*b as a*+b. More specifically, if something simple (such as a character
3293 or a simple class like \d) has an unlimited quantifier, and is followed by
3294 something that cannot possibly match the quantified thing, the quantifier
3295 is automatically "possessified".
3296
3297 24. A recursive reference to a subpattern whose number was greater than 39
3298 went wrong under certain circumstances in UTF-8 mode. This bug could also
3299 have affected the operation of pcre_study().
3300
3301 25. Realized that a little bit of performance could be had by replacing
3302 (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
3303
3304 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
3305
3306 27. Possessive quantifiers such as a++ were previously implemented by turning
3307 them into atomic groups such as ($>a+). Now they have their own opcodes,
3308 which improves performance. This includes the automatically created ones
3309 from 23 above.
3310
3311 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
3312 lookahead was broken if it was not anchored. PCRE was mistakenly expecting
3313 the first matched character to be a colon. This applied both to named and
3314 numbered groups.
3315
3316 29. The ucpinternal.h header file was missing its idempotency #ifdef.
3317
3318 30. I was sent a "project" file called libpcre.a.dev which I understand makes
3319 building PCRE on Windows easier, so I have included it in the distribution.
3320
3321 31. There is now a check in pcretest against a ridiculously large number being
3322 returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
3323 loop, the loop is abandoned.
3324
3325 32. Forward references to subpatterns in conditions such as (?(2)...) where
3326 subpattern 2 is defined later cause pcre_compile() to search forwards in
3327 the pattern for the relevant set of parentheses. This search went wrong
3328 when there were unescaped parentheses in a character class, parentheses
3329 escaped with \Q...\E, or parentheses in a #-comment in /x mode.
3330
3331 33. "Subroutine" calls and backreferences were previously restricted to
3332 referencing subpatterns earlier in the regex. This restriction has now
3333 been removed.
3334
3335 34. Added a number of extra features that are going to be in Perl 5.10. On the
3336 whole, these are just syntactic alternatives for features that PCRE had
3337 previously implemented using the Python syntax or my own invention. The
3338 other formats are all retained for compatibility.
3339
3340 (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
3341 as (?P<name>...). The new forms, as well as being in Perl 5.10, are
3342 also .NET compatible.
3343
3344 (b) A recursion or subroutine call to a named group can now be defined as
3345 (?&name) as well as (?P>name).
3346
3347 (c) A backreference to a named group can now be defined as \k<name> or
3348 \k'name' as well as (?P=name). The new forms, as well as being in Perl
3349 5.10, are also .NET compatible.
3350
3351 (d) A conditional reference to a named group can now use the syntax
3352 (?(<name>) or (?('name') as well as (?(name).
3353
3354 (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
3355 groups (named and numbered) that are never evaluated inline, but can be
3356 called as "subroutines" from elsewhere. In effect, the DEFINE condition
3357 is always false. There may be only one alternative in such a group.
3358
3359 (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
3360 as the simple (?(R). The condition is true only if the most recent
3361 recursion is that of the given number or name. It does not search out
3362 through the entire recursion stack.
3363
3364 (g) The escape \gN or \g{N} has been added, where N is a positive or
3365 negative number, specifying an absolute or relative reference.
3366
3367 35. Tidied to get rid of some further signed/unsigned compiler warnings and
3368 some "unreachable code" warnings.
3369
3370 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
3371 things, this adds five new scripts.
3372
3373 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
3374 There were also incompatibilities regarding the handling of \Q..\E inside
3375 character classes, for example with patterns like [\Qa\E-\Qz\E] where the
3376 hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
3377
3378 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
3379 matches an empty string, and forcibly breaks the loop. There were bugs in
3380 this code in non-simple cases. For a pattern such as ^(a()*)* matched
3381 against aaaa the result was just "a" rather than "aaaa", for example. Two
3382 separate and independent bugs (that affected different cases) have been
3383 fixed.
3384
3385 39. Refactored the code to abolish the use of different opcodes for small
3386 capturing bracket numbers. This is a tidy that I avoided doing when I
3387 removed the limit on the number of capturing brackets for 3.5 back in 2001.
3388 The new approach is not only tidier, it makes it possible to reduce the
3389 memory needed to fix the previous bug (38).
3390
3391 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
3392 sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
3393 processing dot, circumflex, or dollar metacharacters, or #-comments in /x
3394 mode.
3395
3396 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
3397 report.
3398
3399 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
3400 copy construction and assignment in the C++ wrapper.
3401
3402 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
3403 couple of bugs that could have given wrong results in the "--newline=crlf"
3404 case.
3405
3406 44. Added a number of casts and did some reorganization of signed/unsigned int
3407 variables following suggestions from Dair Grant. Also renamed the variable
3408 "this" as "item" because it is a C++ keyword.
3409
3410 45. Arranged for dftables to add
3411
3412 #include "pcre_internal.h"
3413
3414 to pcre_chartables.c because without it, gcc 4.x may remove the array
3415 definition from the final binary if PCRE is built into a static library and
3416 dead code stripping is activated.
3417
3418 46. For an unanchored pattern, if a match attempt fails at the start of a
3419 newline sequence, and the newline setting is CRLF or ANY, and the next two
3420 characters are CRLF, advance by two characters instead of one.
3421
3422
3423 Version 6.7 04-Jul-06
3424 ---------------------
3425
3426 1. In order to handle tests when input lines are enormously long, pcretest has
3427 been re-factored so that it automatically extends its buffers when
3428 necessary. The code is crude, but this _is_ just a test program. The
3429 default size has been increased from 32K to 50K.
3430
3431 2. The code in pcre_study() was using the value of the re argument before
3432 testing it for NULL. (Of course, in any sensible call of the function, it
3433 won't be NULL.)
3434
3435 3. The memmove() emulation function in pcre_internal.h, which is used on
3436 systems that lack both memmove() and bcopy() - that is, hardly ever -
3437 was missing a "static" storage class specifier.
3438
3439 4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
3440 containing an extended class (one that cannot be represented by a bitmap
3441 because it contains high-valued characters or Unicode property items, e.g.
3442 [\pZ]). Almost always one would set UTF-8 mode when processing such a
3443 pattern, but PCRE should not loop if you do not (it no longer does).
3444 [Detail: two cases were found: (a) a repeated subpattern containing an
3445 extended class; (b) a recursive reference to a subpattern that followed a
3446 previous extended class. It wasn't skipping over the extended class
3447 correctly when UTF-8 mode was not set.]
3448
3449 5. A negated single-character class was not being recognized as fixed-length
3450 in lookbehind assertions such as (?<=[^f]), leading to an incorrect
3451 compile error "lookbehind assertion is not fixed length".
3452
3453 6. The RunPerlTest auxiliary script was showing an unexpected difference
3454 between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
3455 write a Perl script that can interpret lines of an input file either as
3456 byte characters or as UTF-8, which is what "perltest" was being required to
3457 do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
3458 can't do is switch easily at run time between having the "use utf8;" pragma
3459 or not. In the end, I fudged it by using the RunPerlTest script to insert
3460 "use utf8;" explicitly for the UTF-8 tests.
3461
3462 7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
3463 the end of the subject string, contrary to the documentation and to what
3464 Perl does. This was true of both matching functions. Now it matches only at
3465 the start of the subject and immediately after *internal* newlines.
3466
3467 8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
3468 a pointer to an int instead of a pointer to an unsigned long int. This
3469 caused problems on 64-bit systems.
3470
3471 9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
3472 instance of the 'standard' template library not being so standard".
3473
3474 10. There was no check on the number of named subpatterns nor the maximum
3475 length of a subpattern name. The product of these values is used to compute
3476 the size of the memory block for a compiled pattern. By supplying a very
3477 long subpattern name and a large number of named subpatterns, the size
3478 computation could be caused to overflow. This is now prevented by limiting
3479 the length of names to 32 characters, and the number of named subpatterns
3480 to 10,000.
3481
3482 11. Subpatterns that are repeated with specific counts have to be replicated in
3483 the compiled pattern. The size of memory for this was computed from the
3484 length of the subpattern and the repeat count. The latter is limited to
3485 65535, but there was no limit on the former, meaning that integer overflow
3486 could in principle occur. The compiled length of a repeated subpattern is
3487 now limited to 30,000 bytes in order to prevent this.
3488
3489 12. Added the optional facility to have named substrings with the same name.
3490
3491 13. Added the ability to use a named substring as a condition, using the
3492 Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
3493 are numbers (not recommended). Forward references are permitted.
3494
3495 14. Added forward references in named backreferences (if you see what I mean).
3496
3497 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
3498 pattern could run off the end of the subject. For example, the pattern
3499 "(?s)(.{1,5})"8 did this with the subject "ab".
3500
3501 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
3502 PCRE_CASELESS was set when matching characters that were quantified with ?
3503 or *.
3504
3505 17. A character class other than a single negated character that had a minimum
3506 but no maximum quantifier - for example [ab]{6,} - was not handled
3507 correctly by pce_dfa_exec(). It would match only one character.
3508
3509 18. A valid (though odd) pattern that looked like a POSIX character
3510 class but used an invalid character after [ (for example [[,abc,]]) caused
3511 pcre_compile() to give the error "Failed: internal error: code overflow" or
3512 in some cases to crash with a glibc free() error. This could even happen if
3513 the pattern terminated after [[ but there just happened to be a sequence of
3514 letters, a binary zero, and a closing ] in the memory that followed.
3515
3516 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
3517 over the years. Originally (before any Unicode support), just the bottom 8
3518 bits were taken. Thus, for example, \500 really meant \100. Nowadays the
3519 output from "man perlunicode" includes this:
3520
3521 The regular expression compiler produces polymorphic opcodes. That
3522 is, the pattern adapts to the data and automatically switches to
3523 the Unicode character scheme when presented with Unicode data--or
3524 instead uses a traditional byte scheme when presented with byte
3525 data.
3526
3527 Sadly, a wide octal escape does not cause a switch, and in a string with
3528 no other multibyte characters, these octal escapes are treated as before.
3529 Thus, in Perl, the pattern /\500/ actually matches \100 but the pattern
3530 /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
3531 Unicode string.
3532
3533 I have not perpetrated such confusion in PCRE. Up till now, it took just
3534 the bottom 8 bits, as in old Perl. I have now made octal escapes with
3535 values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
3536 translate to the appropriate multibyte character.
3537
3538 29. Applied some refactoring to reduce the number of warnings from Microsoft
3539 and Borland compilers. This has included removing the fudge introduced
3540 seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
3541 a warning about an unused variable.
3542
3543 21. PCRE has not included VT (character 0x0b) in the set of whitespace
3544 characters since release 4.0, because Perl (from release 5.004) does not.
3545 [Or at least, is documented not to: some releases seem to be in conflict
3546 with the documentation.] However, when a pattern was studied with
3547 pcre_study() and all its branches started with \s, PCRE still included VT
3548 as a possible starting character. Of course, this did no harm; it just
3549 caused an unnecessary match attempt.
3550
3551 22. Removed a now-redundant internal flag bit that recorded the fact that case
3552 dependency changed within the pattern. This was once needed for "required
3553 byte" processing, but is no longer used. This recovers a now-scarce options
3554 bit. Also moved the least significant internal flag bit to the most-
3555 significant bit of the word, which was not previously used (hangover from
3556 the days when it was an int rather than a uint) to free up another bit for
3557 the future.
3558
3559 23. Added support for CRLF line endings as well as CR and LF. As well as the
3560 default being selectable at build time, it can now be changed at runtime
3561 via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
3562 specify that it is scanning data with non-default line endings.
3563
3564 24. Changed the definition of CXXLINK to make it agree with the definition of
3565 LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
3566
3567 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
3568 recursions. This makes a big different to stack usage for some patterns.
3569
3570 26. If a subpattern containing a named recursion or subroutine reference such
3571 as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
3572 the space required for the compiled pattern went wrong and gave too small a
3573 value. Depending on the environment, this could lead to "Failed: internal
3574 error: code overflow at offset 49" or "glibc detected double free or
3575 corruption" errors.
3576
3577 27. Applied patches from Google (a) to support the new newline modes and (b) to
3578 advance over multibyte UTF-8 characters in GlobalReplace.
3579
3580 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
3581 difference for some implementation of PCRE in some Windows version.
3582
3583 29. Added some extra testing facilities to pcretest:
3584
3585 \q<number> in a data line sets the "match limit" value
3586 \Q<number> in a data line sets the "match recursion limt" value
3587 -S <number> sets the stack size, where <number> is in megabytes
3588
3589 The -S option isn't available for Windows.
3590
3591
3592 Version 6.6 06-Feb-06
3593 ---------------------
3594
3595 1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
3596 in pcreposix.h. I have copied the definition from pcre.h.
3597
3598 2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
3599 because pcre.h is no longer a built file.
3600
3601 3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
3602 not normally included in the compiled code.
3603
3604
3605 Version 6.5 01-Feb-06
3606 ---------------------
3607
3608 1. When using the partial match feature with pcre_dfa_exec(), it was not
3609 anchoring the second and subsequent partial matches at the new starting
3610 point. This could lead to incorrect results. For example, with the pattern
3611 /1234/, partially matching against "123" and then "a4" gave a match.
3612
3613 2. Changes to pcregrep:
3614
3615 (a) All non-match returns from pcre_exec() were being treated as failures
3616 to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
3617 error message is output. Some extra information is given for the
3618 PCRE_ERROR_MATCHLIMIT and PCRE_ERROR_RECURSIONLIMIT errors, which are
3619 probably the only errors that are likely to be caused by users (by
3620 specifying a regex that has nested indefinite repeats, for instance).
3621 If there are more than 20 of these errors, pcregrep is abandoned.
3622
3623 (b) A binary zero was treated as data while matching, but terminated the
3624 output line if it was written out. This has been fixed: binary zeroes
3625 are now no different to any other data bytes.
3626
3627 (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
3628 used to set a locale for matching. The --locale=xxxx long option has
3629 been added (no short equivalent) to specify a locale explicitly on the
3630 pcregrep command, overriding the environment variables.
3631
3632 (d) When -B was used with -n, some line numbers in the output were one less
3633 than they should have been.
3634
3635 (e) Added the -o (--only-matching) option.
3636
3637 (f) If -A or -C was used with -c (count only), some lines of context were
3638 accidentally printed for the final match.
3639
3640 (g) Added the -H (--with-filename) option.
3641
3642 (h) The combination of options -rh failed to suppress file names for files
3643 that were found from directory arguments.
3644
3645 (i) Added the -D (--devices) and -d (--directories) options.
3646
3647 (j) Added the -F (--fixed-strings) option.
3648
3649 (k) Allow "-" to be used as a file name for -f as well as for a data file.
3650
3651 (l) Added the --colo(u)r option.
3652
3653 (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
3654 is not present by default.
3655
3656 3. A nasty bug was discovered in the handling of recursive patterns, that is,
3657 items such as (?R) or (?1), when the recursion could match a number of
3658 alternatives. If it matched one of the alternatives, but subsequently,
3659 outside the recursion, there was a failure, the code tried to back up into
3660 the recursion. However, because of the way PCRE is implemented, this is not
3661 possible, and the result was an incorrect result from the match.
3662
3663 In order to prevent this happening, the specification of recursion has
3664 been changed so that all such subpatterns are automatically treated as
3665 atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
3666
3667 4. I had overlooked the fact that, in some locales, there are characters for
3668 which isalpha() is true but neither isupper() nor islower() are true. In
3669 the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
3670 and ordfeminine) are like this. This affected the treatment of \w and \W
3671 when they appeared in character classes, but not when they appeared outside
3672 a character class. The bit map for "word" characters is now created
3673 separately from the results of isalnum() instead of just taking it from the
3674 upper, lower, and digit maps. (Plus the underscore character, of course.)
3675
3676 5. The above bug also affected the handling of POSIX character classes such as
3677 [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
3678 permanent tables. Instead, the bit maps for such a class were previously
3679 created as the appropriate unions of the upper, lower, and digit bitmaps.
3680 Now they are created by subtraction from the [[:word:]] class, which has
3681 its own bitmap.
3682
3683 6. The [[:blank:]] character class matches horizontal, but not vertical space.
3684 It is created by subtracting the vertical space characters (\x09, \x0a,
3685 \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
3686 subtraction was done in the overall bitmap for a character class, meaning
3687 that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
3688 be recognized. This bug has been fixed.
3689
3690 7. Patches from the folks at Google:
3691
3692 (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
3693 real life, but is still worth protecting against".
3694
3695 (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
3696 regular expressions".
3697
3698 (c) pcre_scanner.cc: avoid use of std::count() because not all systems
3699 have it.
3700
3701 (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
3702 "configure" and the latter not, in order to fix a problem somebody had
3703 with compiling the Arg class on HP-UX.
3704
3705 (e) Improve the error-handling of the C++ wrapper a little bit.
3706
3707 (f) New tests for checking recursion limiting.
3708
3709 8. The pcre_memmove() function, which is used only if the environment does not
3710 have a standard memmove() function (and is therefore rarely compiled),
3711 contained two bugs: (a) use of int instead of size_t, and (b) it was not
3712 returning a result (though PCRE never actually uses the result).
3713
3714 9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
3715 large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
3716 returned instead of calling malloc() with an overflowing number that would
3717 most likely cause subsequent chaos.
3718
3719 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
3720
3721 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
3722 with this option is matched, the nmatch and pmatch options of regexec() are
3723 ignored.
3724
3725 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
3726 provided in case anyone wants to the the POSIX interface with UTF-8
3727 strings.
3728
3729 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
3730 C++ linking (needed for some HP-UX environments).
3731
3732 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
3733 (unused parameter) and in the pcre_printint() function (omitted "default"
3734 switch label when the default is to do nothing).
3735
3736 15. Added some code to make it possible, when PCRE is compiled as a C++
3737 library, to replace subject pointers for pcre_exec() with a smart pointer
3738 class, thus making it possible to process discontinuous strings.
3739
3740 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
3741 much the same function. They were added by different people who were trying
3742 to make PCRE easy to compile on non-Unix systems. It has been suggested
3743 that PCRE_EXPORT be abolished now that there is more automatic apparatus
3744 for compiling on Windows systems. I have therefore replaced it with
3745 PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
3746 defaults to "extern" for C or "extern C" for C++, which works fine on
3747 Unix-like systems. It is now possible to override the value of PCRE_DATA_
3748 SCOPE with something explicit in config.h. In addition:
3749
3750 (a) pcreposix.h still had just "extern" instead of either of these macros;
3751 I have replaced it with PCRE_DATA_SCOPE.
3752
3753 (b) Functions such as _pcre_xclass(), which are internal to the library,
3754 but external in the C sense, all had PCRE_EXPORT in their definitions.
3755 This is apparently wrong for the Windows case, so I have removed it.
3756 (It makes no difference on Unix-like systems.)
3757
3758 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
3759 of recursive calls to match(). This is different to MATCH_LIMIT because
3760 that limits the total number of calls to match(), not all of which increase
3761 the depth of recursion. Limiting the recursion depth limits the amount of
3762 stack (or heap if NO_RECURSE is set) that is used. The default can be set
3763 when PCRE is compiled, and changed at run time. A patch from Google adds
3764 this functionality to the C++ interface.
3765
3766 18. Changes to the handling of Unicode character properties:
3767
3768 (a) Updated the table to Unicode 4.1.0.
3769
3770 (b) Recognize characters that are not in the table as "Cn" (undefined).
3771
3772 (c) I revised the way the table is implemented to a much improved format
3773 which includes recognition of ranges. It now supports the ranges that
3774 are defined in UnicodeData.txt, and it also amalgamates other
3775 characters into ranges. This has reduced the number of entries in the
3776 table from around 16,000 to around 3,000, thus reducing its size
3777 considerably. I realized I did not need to use a tree structure after
3778 all - a binary chop search is just as efficient. Having reduced the
3779 number of entries, I extended their size from 6 bytes to 8 bytes to
3780 allow for more data.
3781
3782 (d) Added support for Unicode script names via properties such as \p{Han}.
3783
3784 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
3785 matching that character.
3786
3787 20. When matching a repeated Unicode property with a minimum greater than zero,
3788 (for example \pL{2,}), PCRE could look past the end of the subject if it
3789 reached it while seeking the minimum number of characters. This could
3790 happen only if some of the characters were more than one byte long, because
3791 there is a check for at least the minimum number of bytes.
3792
3793 21. Refactored the implementation of \p and \P so as to be more general, to
3794 allow for more different types of property in future. This has changed the
3795 compiled form incompatibly. Anybody with saved compiled patterns that use
3796 \p or \P will have to recompile them.
3797
3798 22. Added "Any" and "L&" to the supported property types.
3799
3800 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
3801 but give a compile time error if the value is greater than 0xff.
3802
3803 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
3804 accidentally not being installed or uninstalled.
3805
3806 25. The pcre.h file was built from pcre.h.in, but the only changes that were
3807 made were to insert the current release number. This seemed silly, because
3808 it made things harder for people building PCRE on systems that don't run
3809 "configure". I have turned pcre.h into a distributed file, no longer built
3810 by "configure", with the version identification directly included. There is
3811 no longer a pcre.h.in file.
3812
3813 However, this change necessitated a change to the pcre-config script as
3814 well. It is built from pcre-config.in, and one of the substitutions was the
3815 release number. I have updated configure.ac so that ./configure now finds
3816 the release number by grepping pcre.h.
3817
3818 26. Added the ability to run the tests under valgrind.
3819
3820
3821 Version 6.4 05-Sep-05
3822 ---------------------
3823
3824 1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
3825 "--" to be printed when multiple files were scanned, even when none of the
3826 -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
3827 consider it to be a bug, and have restored the previous behaviour.
3828
3829 2. A couple of code tidies to get rid of compiler warnings.
3830
3831 3. The pcretest program used to cheat by referring to symbols in the library
3832 whose names begin with _pcre_. These are internal symbols that are not
3833 really supposed to be visible externally, and in some environments it is
3834 possible to suppress them. The cheating is now confined to including
3835 certain files from the library's source, which is a bit cleaner.
3836
3837 4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
3838 file's purpose clearer.
3839
3840 5. Reorganized pcre_ucp_findchar().
3841
3842
3843 Version 6.3 15-Aug-05
3844 ---------------------
3845
3846 1. The file libpcre.pc.in did not have general read permission in the tarball.
3847
3848 2. There were some problems when building without C++ support:
3849
3850 (a) If C++ support was not built, "make install" and "make test" still
3851 tried to test it.
3852
3853 (b) There were problems when the value of CXX was explicitly set. Some
3854 changes have been made to try to fix these, and ...
3855
3856 (c) --disable-cpp can now be used to explicitly disable C++ support.
3857
3858 (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
3859 backslash in a target when C++ was disabled. This confuses some
3860 versions of "make", apparently. Using an intermediate variable solves
3861 this. (Same for CPP_LOBJ.)
3862
3863 3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
3864 (non-Windows) now includes $(CFLAGS) because these flags are sometimes
3865 necessary on certain architectures.
3866
3867 4. Added a setting of -export-symbols-regex to the link command to remove
3868 those symbols that are exported in the C sense, but actually are local
3869 within the library, and not documented. Their names all begin with
3870 "_pcre_". This is not a perfect job, because (a) we have to except some
3871 symbols that pcretest ("illegally") uses, and (b) the facility isn't always
3872 available (and never for static libraries). I have made a note to try to
3873 find a way round (a) in the future.
3874
3875
3876 Version 6.2 01-Aug-05
3877 ---------------------
3878
3879 1. There was no test for integer overflow of quantifier values. A construction
3880 such as {1111111111111111} would give undefined results. What is worse, if
3881 a minimum quantifier for a parenthesized subpattern overflowed and became
3882 negative, the calculation of the memory size went wrong. This could have
3883 led to memory overwriting.
3884
3885 2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
3886
3887 3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
3888 operating environments where this matters.
3889
3890 4. Applied Giuseppe Maxia's patch to add additional features for controlling
3891 PCRE options from within the C++ wrapper.
3892
3893 5. Named capturing subpatterns were not being correctly counted when a pattern
3894 was compiled. This caused two problems: (a) If there were more than 100
3895 such subpatterns, the calculation of the memory needed for the whole
3896 compiled pattern went wrong, leading to an overflow error. (b) Numerical
3897 back references of the form \12, where the number was greater than 9, were
3898 not recognized as back references, even though there were sufficient
3899 previous subpatterns.
3900
3901 6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
3902 versions of gcc, e.g. 2.95.4.
3903
3904
3905 Version 6.1 21-Jun-05
3906 ---------------------
3907
3908 1. There was one reference to the variable "posix" in pcretest.c that was not
3909 surrounded by "#if !defined NOPOSIX".
3910
3911 2. Make it possible to compile pcretest without DFA support, UTF8 support, or
3912 the cross-check on the old pcre_info() function, for the benefit of the
3913 cut-down version of PCRE that is currently imported into Exim.
3914
3915 3. A (silly) pattern starting with (?i)(?-i) caused an internal space
3916 allocation error. I've done the easy fix, which wastes 2 bytes for sensible
3917 patterns that start (?i) but I don't think that matters. The use of (?i) is
3918 just an example; this all applies to the other options as well.
3919
3920 4. Since libtool seems to echo the compile commands it is issuing, the output
3921 from "make" can be reduced a bit by putting "@" in front of each libtool
3922 compile command.
3923
3924 5. Patch from the folks at Google for configure.in to be a bit more thorough
3925 in checking for a suitable C++ installation before trying to compile the
3926 C++ stuff. This should fix a reported problem when a compiler was present,
3927 but no suitable headers.
3928
3929 6. The man pages all had just "PCRE" as their title. I have changed them to
3930 be the relevant file name. I have also arranged that these names are
3931 retained in the file doc/pcre.txt, which is a concatenation in text format
3932 of all the man pages except the little individual ones for each function.
3933
3934 7. The NON-UNIX-USE file had not been updated for the different set of source
3935 files that come with release 6. I also added a few comments about the C++
3936 wrapper.
3937
3938
3939 Version 6.0 07-Jun-05
3940 ---------------------
3941
3942 1. Some minor internal re-organization to help with my DFA experiments.
3943
3944 2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
3945 didn't matter for the library itself when fully configured, but did matter
3946 when compiling without UCP support, or within Exim, where the ucp files are
3947 not imported.
3948
3949 3. Refactoring of the library code to split up the various functions into
3950 different source modules. The addition of the new DFA matching code (see
3951 below) to a single monolithic source would have made it really too
3952 unwieldy, quite apart from causing all the code to be include in a
3953 statically linked application, when only some functions are used. This is
3954 relevant even without the DFA addition now that patterns can be compiled in
3955 one application and matched in another.
3956
3957 The downside of splitting up is that there have to be some external
3958 functions and data tables that are used internally in different modules of
3959 the library but which are not part of the API. These have all had their
3960 names changed to start with "_pcre_" so that they are unlikely to clash
3961 with other external names.
3962
3963 4. Added an alternate matching function, pcre_dfa_exec(), which matches using
3964 a different (DFA) algorithm. Although it is slower than the original
3965 function, it does have some advantages for certain types of matching
3966 problem.
3967
3968 5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
3969 including restarting after a partial match.
3970
3971 6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
3972 defined when compiling for Windows was sent to me. I have put it into the
3973 code, though I have no means of testing or verifying it.
3974
3975 7. Added the pcre_refcount() auxiliary function.
3976
3977 8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
3978 match before or at the first newline in the subject string. In pcretest,
3979 the /f option on a pattern can be used to set this.
3980
3981 9. A repeated \w when used in UTF-8 mode with characters greater than 256
3982 would behave wrongly. This has been present in PCRE since release 4.0.
3983
3984 10. A number of changes to the pcregrep command:
3985
3986 (a) Refactored how -x works; insert ^(...)$ instead of setting
3987 PCRE_ANCHORED and checking the length, in preparation for adding
3988 something similar for -w.
3989
3990 (b) Added the -w (match as a word) option.
3991
3992 (c) Refactored the way lines are read and buffered so as to have more
3993 than one at a time available.
3994
3995 (d) Implemented a pcregrep test script.
3996
3997 (e) Added the -M (multiline match) option. This allows patterns to match
3998 over several lines of the subject. The buffering ensures that at least
3999 8K, or the rest of the document (whichever is the shorter) is available
4000 for matching (and similarly the previous 8K for lookbehind assertions).
4001
4002 (f) Changed the --help output so that it now says
4003
4004 -w, --word-regex(p)
4005
4006 instead of two lines, one with "regex" and the other with "regexp"
4007 because that confused at least one person since the short forms are the
4008 same. (This required a bit of code, as the output is generated
4009 automatically from a table. It wasn't just a text change.)
4010
4011 (g) -- can be used to terminate pcregrep options if the next thing isn't an
4012 option but starts with a hyphen. Could be a pattern or a path name
4013 starting with a hyphen, for instance.
4014
4015 (h) "-" can be given as a file name to represent stdin.
4016
4017 (i) When file names are being printed, "(standard input)" is used for
4018 the standard input, for compatibility with GNU grep. Previously
4019 "<stdin>" was used.
4020
4021 (j) The option --label=xxx can be used to supply a name to be used for
4022 stdin when file names are being printed. There is no short form.
4023
4024 (k) Re-factored the options decoding logic because we are going to add
4025 two more options that take data. Such options can now be given in four
4026 different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
4027
4028 (l) Added the -A, -B, and -C options for requesting that lines of context
4029 around matches be printed.
4030
4031 (m) Added the -L option to print the names of files that do not contain
4032 any matching lines, that is, the complement of -l.
4033
4034 (n) The return code is 2 if any file cannot be opened, but pcregrep does
4035 continue to scan other files.
4036
4037 (o) The -s option was incorrectly implemented. For compatibility with other
4038 greps, it now suppresses the error message for a non-existent or non-
4039 accessible file (but not the return code). There is a new option called
4040 -q that suppresses the output of matching lines, which was what -s was
4041 previously doing.
4042
4043 (p) Added --include and --exclude options to specify files for inclusion
4044 and exclusion when recursing.
4045
4046 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
4047 Hopefully, it now does.
4048
4049 12. Missing cast in pcre_study().
4050
4051 13. Added an "uninstall" target to the makefile.
4052
4053 14. Replaced "extern" in the function prototypes in Makefile.in with
4054 "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
4055 world, but is set differently for Windows.
4056
4057 15. Added a second compiling function called pcre_compile2(). The only
4058 difference is that it has an extra argument, which is a pointer to an
4059 integer error code. When there is a compile-time failure, this is set
4060 non-zero, in addition to the error test pointer being set to point to an
4061 error message. The new argument may be NULL if no error number is required
4062 (but then you may as well call pcre_compile(), which is now just a
4063 wrapper). This facility is provided because some applications need a
4064 numeric error indication, but it has also enabled me to tidy up the way
4065 compile-time errors are handled in the POSIX wrapper.
4066
4067 16. Added VPATH=.libs to the makefile; this should help when building with one
4068 prefix path and installing with another. (Or so I'm told by someone who
4069 knows more about this stuff than I do.)
4070
4071 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
4072 passes PCRE_DOTALL to the pcre_compile() function, making the "." character
4073 match everything, including newlines. This is not POSIX-compatible, but
4074 somebody wanted the feature. From pcretest it can be activated by using
4075 both the P and the s flags.
4076
4077 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
4078
4079 19. libpcre.pc was being incorrectly installed as executable.
4080
4081 20. A couple of places in pcretest check for end-of-line by looking for '\n';
4082 it now also looks for '\r' so that it will work unmodified on Windows.
4083
4084 21. Added Google's contributed C++ wrapper to the distribution.
4085
4086 22. Added some untidy missing memory free() calls in pcretest, to keep
4087 Electric Fence happy when testing.
4088
4089
4090
4091 Version 5.0 13-Sep-04
4092 ---------------------
4093
4094 1. Internal change: literal characters are no longer packed up into items
4095 containing multiple characters in a single byte-string. Each character
4096 is now matched using a separate opcode. However, there may be more than one
4097 byte in the character in UTF-8 mode.
4098
4099 2. The pcre_callout_block structure has two new fields: pattern_position and
4100 next_item_length. These contain the offset in the pattern to the next match
4101 item, and its length, respectively.
4102
4103 3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
4104 insertion of callouts before each pattern item. Added the /C option to
4105 pcretest to make use of this.
4106
4107 4. On the advice of a Windows user, the lines
4108
4109 #if defined(_WIN32) || defined(WIN32)
4110 _setmode( _fileno( stdout ), 0x8000 );
4111 #endif /* defined(_WIN32) || defined(WIN32) */
4112
4113 have been added to the source of pcretest. This apparently does useful
4114 magic in relation to line terminators.
4115
4116 5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
4117 for the benefit of those environments where the "b" makes a difference.
4118
4119 6. The icc compiler has the same options as gcc, but "configure" doesn't seem
4120 to know about it. I have put a hack into configure.in that adds in code
4121 to set GCC=yes if CC=icc. This seems to end up at a point in the
4122 generated configure script that is early enough to affect the setting of
4123 compiler options, which is what is needed, but I have no means of testing
4124 whether it really works. (The user who reported this had patched the
4125 generated configure script, which of course I cannot do.)
4126
4127 LATER: After change 22 below (new libtool files), the configure script
4128 seems to know about icc (and also ecc). Therefore, I have commented out
4129 this hack in configure.in.
4130
4131 7. Added support for pkg-config (2 patches were sent in).
4132
4133 8. Negated POSIX character classes that used a combination of internal tables
4134 were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
4135 [[:^ascii]]. Typically, they would match almost any characters. The other
4136 POSIX classes were not broken in this way.
4137
4138 9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
4139 to find the match, as PCRE was deluded into thinking that the match had to
4140 start at the start point or following a newline. The same bug applied to
4141 patterns with negative forward assertions or any backward assertions
4142 preceding ".*" at the start, unless the pattern required a fixed first
4143 character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
4144
4145 10. In UTF-8 mode, when moving forwards in the subject after a failed match
4146 starting at the last subject character, bytes beyond the end of the subject
4147 string were read.
4148
4149 11. Renamed the variable "class" as "classbits" to make life easier for C++
4150 users. (Previously there was a macro definition, but it apparently wasn't
4151 enough.)
4152
4153 12. Added the new field "tables" to the extra data so that tables can be passed
4154 in at exec time, or the internal tables can be re-selected. This allows
4155 a compiled regex to be saved and re-used at a later time by a different
4156 program that might have everything at different addresses.
4157
4158 13. Modified the pcre-config script so that, when run on Solaris, it shows a
4159 -R library as well as a -L library.
4160
4161 14. The debugging options of pcretest (-d on the command line or D on a
4162 pattern) showed incorrect output for anything following an extended class
4163 that contained multibyte characters and which was followed by a quantifier.
4164
4165 15. Added optional support for general category Unicode character properties
4166 via the \p, \P, and \X escapes. Unicode property support implies UTF-8
4167 support. It adds about 90K to the size of the library. The meanings of the
4168 inbuilt class escapes such as \d and \s have NOT been changed.
4169
4170 16. Updated pcredemo.c to include calls to free() to release the memory for the
4171 compiled pattern.
4172
4173 17. The generated file chartables.c was being created in the source directory
4174 instead of in the building directory. This caused the build to fail if the
4175 source directory was different from the building directory, and was
4176 read-only.
4177
4178 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
4179 file. No doubt somebody will tell me if they don't make sense... Also added
4180 Dan Mooney's comments about building on OpenVMS.
4181
4182 19. Added support for partial matching via the PCRE_PARTIAL option for
4183 pcre_exec() and the \P data escape in pcretest.
4184
4185 20. Extended pcretest with 3 new pattern features:
4186
4187 (i) A pattern option of the form ">rest-of-line" causes pcretest to
4188 write the compiled pattern to the file whose name is "rest-of-line".
4189 This is a straight binary dump of the data, with the saved pointer to
4190 the character tables forced to be NULL. The study data, if any, is
4191 written too. After writing, pcretest reads a new pattern.
4192
4193 (ii) If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
4194 compiled pattern from the given file. There must not be any
4195 occurrences of "<" in the file name (pretty unlikely); if there are,
4196 pcretest will instead treat the initial "<" as a pattern delimiter.
4197 After reading in the pattern, pcretest goes on to read data lines as
4198 usual.
4199
4200 (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
4201 and 16-bit fields in a compiled pattern, to simulate a pattern that
4202 was compiled on a host of opposite endianness.
4203
4204 21. The pcre-exec() function can now cope with patterns that were compiled on
4205 hosts of opposite endianness, with this restriction:
4206
4207 As for any compiled expression that is saved and used later, the tables
4208 pointer field cannot be preserved; the extra_data field in the arguments
4209 to pcre_exec() should be used to pass in a tables address if a value
4210 other than the default internal tables were used at compile time.
4211
4212 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
4213 now diagnosed as an error. Previously, most of the time, a negative number
4214 would have been treated as zero, but if in addition "ovector" was passed as
4215 NULL, a crash could occur.
4216
4217 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
4218 new versions from the libtool 1.5 distribution (the last one is a copy of
4219 a file called libtool.m4). This seems to have fixed the need to patch
4220 "configure" to support Darwin 1.3 (which I used to do). However, I still
4221 had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
4222 workstation).
4223
4224 24. Changed the PCRE licence to be the more standard "BSD" licence.
4225
4226
4227 Version 4.5 01-Dec-03
4228 ---------------------
4229
4230 1. There has been some re-arrangement of the code for the match() function so
4231 that it can be compiled in a version that does not call itself recursively.
4232 Instead, it keeps those local variables that need separate instances for
4233 each "recursion" in a frame on the heap, and gets/frees frames whenever it
4234 needs to "recurse". Keeping track of where control must go is done by means
4235 of setjmp/longjmp. The whole thing is implemented by a set of macros that
4236 hide most of the details from the main code, and operates only if
4237 NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
4238 "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
4239 operating.
4240
4241 To make it easier for callers to provide specially tailored get/free
4242 functions for this usage, two new functions, pcre_stack_malloc, and
4243 pcre_stack_free, are used. They are always called in strict stacking order,
4244 and the size of block requested is always the same.
4245
4246 The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
4247 PCRE has been compiled to use the stack or the heap for recursion. The
4248 -C option of pcretest uses this to show which version is compiled.
4249
4250 A new data escape \S, is added to pcretest; it causes the amounts of store
4251 obtained and freed by both kinds of malloc/free at match time to be added
4252 to the output.
4253
4254 2. Changed the locale test to use "fr_FR" instead of "fr" because that's
4255 what's available on my current Linux desktop machine.
4256
4257 3. When matching a UTF-8 string, the test for a valid string at the start has
4258 been extended. If start_offset is not zero, PCRE now checks that it points
4259 to a byte that is the start of a UTF-8 character. If not, it returns
4260 PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
4261 this is necessary because there may be backward assertions in the pattern.
4262 When matching the same subject several times, it may save resources to use
4263 PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
4264
4265 4. The code for checking the validity of UTF-8 strings has been tightened so
4266 that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
4267 containing "overlong sequences".
4268
4269 5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
4270 I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
4271 should have been "&", but it just so happened that all the cases this let
4272 through by mistake were picked up later in the function.
4273
4274 6. I had used a variable called "isblank" - this is a C99 function, causing
4275 some compilers to warn. To avoid this, I renamed it (as "blankclass").
4276
4277 7. Cosmetic: (a) only output another newline at the end of pcretest if it is
4278 prompting; (b) run "./pcretest /dev/null" at the start of the test script
4279 so the version is shown; (c) stop "make test" echoing "./RunTest".
4280
4281 8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
4282
4283 9. The prototype for memmove() for systems that don't have it was using
4284 size_t, but the inclusion of the header that defines size_t was later. I've
4285 moved the #includes for the C headers earlier to avoid this.
4286
4287 10. Added some adjustments to the code to make it easier to compiler on certain
4288 special systems:
4289
4290 (a) Some "const" qualifiers were missing.
4291 (b) Added the macro EXPORT before all exported functions; by default this
4292 is defined to be empty.
4293 (c) Changed the dftables auxiliary program (that builds chartables.c) so
4294 that it reads its output file name as an argument instead of writing
4295 to the standard output and assuming this can be redirected.
4296
4297 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
4298 class containing characters with values greater than 255, PCRE compilation
4299 went into a loop.
4300
4301 12. A recursive reference to a subpattern that was within another subpattern
4302 that had a minimum quantifier of zero caused PCRE to crash. For example,
4303 (x(y(?2))z)? provoked this bug with a subject that got as far as the
4304 recursion. If the recursively-called subpattern itself had a zero repeat,
4305 that was OK.
4306
4307 13. In pcretest, the buffer for reading a data line was set at 30K, but the
4308 buffer into which it was copied (for escape processing) was still set at
4309 1024, so long lines caused crashes.
4310
4311 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
4312 "internal error: code overflow...". This applied to any character class
4313 that was followed by a possessive quantifier.
4314
4315 15. Modified the Makefile to add libpcre.la as a prerequisite for
4316 libpcreposix.la because I was told this is needed for a parallel build to
4317 work.
4318
4319 16. If a pattern that contained .* following optional items at the start was
4320 studied, the wrong optimizing data was generated, leading to matching
4321 errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
4322 matching string must start with a or b or c. The correct conclusion for
4323 this pattern is that a match can start with any character.
4324
4325
4326 Version 4.4 13-Aug-03
4327 ---------------------
4328
4329 1. In UTF-8 mode, a character class containing characters with values between
4330 127 and 255 was not handled correctly if the compiled pattern was studied.
4331 In fixing this, I have also improved the studying algorithm for such
4332 classes (slightly).
4333
4334 2. Three internal functions had redundant arguments passed to them. Removal
4335 might give a very teeny performance improvement.
4336
4337 3. Documentation bug: the value of the capture_top field in a callout is *one
4338 more than* the number of the hightest numbered captured substring.
4339
4340 4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
4341 in incorrectly linking with a previously installed version. They now link
4342 explicitly with libpcre.la.
4343
4344 5. configure.in no longer needs to recognize Cygwin specially.
4345
4346 6. A problem in pcre.in for Windows platforms is fixed.
4347
4348 7. If a pattern was successfully studied, and the -d (or /D) flag was given to
4349 pcretest, it used to include the size of the study block as part of its
4350 output. Unfortunately, the structure contains a field that has a different
4351 size on different hardware architectures. This meant that the tests that
4352 showed this size failed. As the block is currently always of a fixed size,
4353 this information isn't actually particularly useful in pcretest output, so
4354 I have just removed it.
4355
4356 8. Three pre-processor statements accidentally did not start in column 1.
4357 Sadly, there are *still* compilers around that complain, even though
4358 standard C has not required this for well over a decade. Sigh.
4359
4360 9. In pcretest, the code for checking callouts passed small integers in the
4361 callout_data field, which is a void * field. However, some picky compilers
4362 complained about the casts involved for this on 64-bit systems. Now
4363 pcretest passes the address of the small integer instead, which should get
4364 rid of the warnings.
4365
4366 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
4367 both compile and run time, and gives an error if an invalid UTF-8 sequence
4368 is found. There is a option for disabling this check in cases where the
4369 string is known to be correct and/or the maximum performance is wanted.
4370
4371 11. In response to a bug report, I changed one line in Makefile.in from
4372
4373 -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
4374 to
4375 -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
4376
4377 to look similar to other lines, but I have no way of telling whether this
4378 is the right thing to do, as I do not use Windows. No doubt I'll get told
4379 if it's wrong...
4380
4381
4382 Version 4.3 21-May-03
4383 ---------------------
4384
4385 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
4386 Makefile.
4387
4388 2. Some refactoring to improve the quality of the code:
4389
4390 (i) The utf8_table... variables are now declared "const".
4391
4392 (ii) The code for \cx, which used the "case flipping" table to upper case
4393 lower case letters, now just substracts 32. This is ASCII-specific,
4394 but the whole concept of \cx is ASCII-specific, so it seems
4395 reasonable.
4396
4397 (iii) PCRE was using its character types table to recognize decimal and
4398 hexadecimal digits in the pattern. This is silly, because it handles
4399 only 0-9, a-f, and A-F, but the character types table is locale-
4400 specific, which means strange things might happen. A private
4401 table is now used for this - though it costs 256 bytes, a table is
4402 much faster than multiple explicit tests. Of course, the standard
4403 character types table is still used for matching digits in subject
4404 strings against \d.
4405
4406 (iv) Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
4407 ending in _t are). So I've renamed it as ESC_tee.
4408
4409 3. The first argument for regexec() in the POSIX wrapper should have been
4410 defined as "const".
4411
4412 4. Changed pcretest to use malloc() for its buffers so that they can be
4413 Electric Fenced for debugging.
4414
4415 5. There were several places in the code where, in UTF-8 mode, PCRE would try
4416 to read one or more bytes before the start of the subject string. Often this
4417 had no effect on PCRE's behaviour, but in some circumstances it could
4418 provoke a segmentation fault.
4419
4420 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
4421 to try to read one or more bytes before the start of the subject string.
4422
4423 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
4424 UTF-8 support could misbehave in various ways if the subject string
4425 contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
4426 area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
4427 back over UTF-8 characters.)
4428
4429
4430 Version 4.2 14-Apr-03
4431 ---------------------
4432
4433 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
4434
4435 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
4436 [ON_WINDOWS]: new variable, "#" on non-Windows platforms
4437 [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
4438 [WIN_PREFIX]: new variable, "cyg" for Cygwin
4439 * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
4440 and BUILD_EXEEXT
4441 Note: automatic setting of the BUILD variables is not yet working
4442 set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
4443 compile-time but not at link-time
4444 [LINK]: use for linking executables only
4445 make different versions for Windows and non-Windows
4446 [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
4447 libraries
4448 [LINK_FOR_BUILD]: new variable
4449 [OBJEXT]: use throughout
4450 [EXEEXT]: use throughout
4451 <winshared>: new target
4452 <wininstall>: new target
4453 <dftables.o>: use native compiler
4454 <dftables>: use native linker
4455 <install>: handle Windows platform correctly
4456 <clean>: ditto
4457 <check>: ditto
4458 copy DLL to top builddir before testing
4459
4460 As part of these changes, -no-undefined was removed again. This was reported
4461 to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
4462 in any case.
4463
4464 3. Some tidies to get rid of compiler warnings:
4465
4466 . In the match_data structure, match_limit was an unsigned long int, whereas
4467 match_call_count was an int. I've made them both unsigned long ints.
4468
4469 . In pcretest the fact that a const uschar * doesn't automatically cast to
4470 a void * provoked a warning.
4471
4472 . Turning on some more compiler warnings threw up some "shadow" variables
4473 and a few more missing casts.
4474
4475 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
4476 option, a class that contained a single character with a value between 128
4477 and 255 (e.g. /[\xFF]/) caused PCRE to crash.
4478
4479 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
4480 option, a class that contained several characters, but with at least one
4481 whose value was between 128 and 255 caused PCRE to crash.
4482
4483
4484 Version 4.1 12-Mar-03
4485 ---------------------
4486
4487 1. Compiling with gcc -pedantic found a couple of places where casts were
4488 needed, and a string in dftables.c that was longer than standard compilers are
4489 required to support.
4490
4491 2. Compiling with Sun's compiler found a few more places where the code could
4492 be tidied up in order to avoid warnings.
4493
4494 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
4495 first of these names is deprecated in the latest Autoconf in favour of the name
4496 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
4497 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
4498 analogy I have changed it to CFLAGS_FOR_BUILD.
4499
4500 4. Added -no-undefined to the linking command in the Makefile, because this is
4501 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
4502 linking step for the pcreposix library.
4503
4504 5. PCRE was failing to diagnose the case of two named groups with the same
4505 name.
4506
4507 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
4508 literal character that is needed in the subject for a match, and scans along to
4509 ensure that it is present before embarking on the full matching process. This
4510 saves time in cases of nested unlimited repeats that are never going to match.
4511 Problem: the scan can take a lot of time if the subject is very long (e.g.
4512 megabytes), thus penalizing straightforward matches. It is now done only if the
4513 amount of subject to be scanned is less than 1000 bytes.
4514
4515 7. A lesser problem with the same optimization is that it was recording the
4516 first character of an anchored pattern as "needed", thus provoking a search
4517 right along the subject, even when the first match of the pattern was going to
4518 fail. The "needed" character is now not set for anchored patterns, unless it
4519 follows something in the pattern that is of non-fixed length. Thus, it still
4520 fulfils its original purpose of finding quick non-matches in cases of nested
4521 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
4522
4523
4524 Version 4.0 17-Feb-03
4525 ---------------------
4526
4527 1. If a comment in an extended regex that started immediately after a meta-item
4528 extended to the end of string, PCRE compiled incorrect data. This could lead to
4529 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
4530
4531 2. Moved to autoconf 2.53 and libtool 1.4.2.
4532
4533 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
4534 the special perltest8 script is no longer needed - all the tests can be run
4535 from a single perltest script.
4536
4537 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
4538 by \s. It has now been removed in PCRE. This means it isn't recognized as
4539 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
4540 class [:space:] *does* include VT, thereby creating a mess.
4541
4542 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
4543 space and tab.
4544
4545 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
4546 its new features into the main test script, reducing the number of scripts.
4547
4548 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
4549 were backward compatible, and made the (?i) apply to the whole pattern, as if
4550 /i were given. Now it behaves more logically, and applies the option setting
4551 only to what follows. PCRE has been changed to follow suit. However, if it
4552 finds options settings right at the start of the pattern, it extracts them into
4553 the global options, as before. Thus, they show up in the info data.
4554
4555 8. Added support for the \Q...\E escape sequence. Characters in between are
4556 treated as literals. This is slightly different from Perl in that $ and @ are
4557 also handled as literals inside the quotes. In Perl, they will cause variable
4558 interpolation. Note the following examples:
4559
4560 Pattern PCRE matches Perl matches
4561
4562 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
4563 \Qabc\$xyz\E abc\$xyz abc\$xyz
4564 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
4565
4566 For compatibility with Perl, \Q...\E sequences are recognized inside character
4567 classes as well as outside them.
4568
4569 9. Re-organized 3 code statements in pcretest to avoid "overflow in
4570 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
4571 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
4572 signed/unsigned warnings.
4573
4574 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
4575 option for pcretest, so I've replaced it by a simple function that does just
4576 that job.
4577
4578 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
4579 "pcregrep -".
4580
4581 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
4582 Java package. This provides some syntactic sugar for simple cases of what my
4583 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
4584 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
4585 item, you can use this simplified notation. Note that only makes sense with
4586 greedy quantifiers. Consequently, the use of the possessive quantifier forces
4587 greediness, whatever the setting of the PCRE_UNGREEDY option.
4588
4589 13. A change of greediness default within a pattern was not taking effect at
4590 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
4591 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
4592 was abstracted outside.
4593
4594 14. PCRE now supports the \G assertion. It is true when the current matching
4595 position is at the start point of the match. This differs from \A when the
4596 starting offset is non-zero. Used with the /g option of pcretest (or similar
4597 code), it works in the same way as it does for Perl's /g option. If all
4598 alternatives of a regex begin with \G, the expression is anchored to the start
4599 match position, and the "anchored" flag is set in the compiled expression.
4600
4601 15. Some bugs concerning the handling of certain option changes within patterns
4602 have been fixed. These applied to options other than (?ims). For example,
4603 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
4604 been the other way round. Some of this was related to change 7 above.
4605
4606 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
4607 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
4608 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
4609 POSIX classes only within a class (e.g. /[[:alpha:]]/).
4610
4611 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
4612 mode. Unlike ".", it always matches newline, whatever the setting of
4613 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
4614 assertions. Perl allows it, but it doesn't (in general) work because it can't
4615 calculate the length of the lookbehind. At least, that's the case for Perl
4616 5.8.0 - I've been told they are going to document that it doesn't work in
4617 future.
4618
4619 18. Added an error diagnosis for escapes that PCRE does not support: these are
4620 \L, \l, \N, \P, \p, \U, \u, and \X.
4621
4622 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
4623 reading past the end of the pattern in cases such as /[abcd/.
4624
4625 20. PCRE was getting more memory than necessary for patterns with classes that
4626 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
4627
4628 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
4629 compiling PCRE for use with Virtual Pascal.
4630
4631 22. Small fix to the Makefile to make it work properly if the build is done
4632 outside the source tree.
4633
4634 23. Added a new extension: a condition to go with recursion. If a conditional
4635 subpattern starts with (?(R) the "true" branch is used if recursion has
4636 happened, whereas the "false" branch is used only at the top level.
4637
4638 24. When there was a very long string of literal characters (over 255 bytes
4639 without UTF support, over 250 bytes with UTF support), the computation of how
4640 much memory was required could be incorrect, leading to segfaults or other
4641 strange effects.
4642
4643 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
4644 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
4645 there was a subsequent back reference to those brackets. This meant that, for
4646 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
4647 possible to check for precisely this case. All we can do is abandon the
4648 optimization if .* occurs inside capturing brackets when there are any back
4649 references whatsoever. (See below for a better fix that came later.)
4650
4651 26. The handling of the optimization for finding the first character of a
4652 non-anchored pattern, and for finding a character that is required later in the
4653 match were failing in some cases. This didn't break the matching; it just
4654 failed to optimize when it could. The way this is done has been re-implemented.
4655
4656 27. Fixed typo in error message for invalid (?R item (it said "(?p").
4657
4658 28. Added a new feature that provides some of the functionality that Perl
4659 provides with (?{...}). The facility is termed a "callout". The way it is done
4660 in PCRE is for the caller to provide an optional function, by setting
4661 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
4662 global variable. By default it is unset, which disables all calling out. To get
4663 the function called, the regex must include (?C) at appropriate points. This
4664 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
4665 This provides a means of identifying different callout points. When PCRE
4666 reaches such a point in the regex, if pcre_callout has been set, the external
4667 function is called. It is provided with data in a structure called
4668 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
4669 matching continues; if it returns a non-zero value, the match at the current
4670 point fails. However, backtracking will occur if possible. [This was changed
4671 later and other features added - see item 49 below.]
4672
4673 29. pcretest is upgraded to test the callout functionality. It provides a
4674 callout function that displays information. By default, it shows the start of
4675 the match and the current position in the text. There are some new data escapes
4676 to vary what happens:
4677
4678 \C+ in addition, show current contents of captured substrings
4679 \C- do not supply a callout function
4680 \C!n return 1 when callout number n is reached
4681 \C!n!m return 1 when callout number n is reached for the mth time
4682
4683 30. If pcregrep was called with the -l option and just a single file name, it
4684 output "<stdin>" if a match was found, instead of the file name.
4685
4686 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
4687 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
4688 pcre_exec(). This saves a malloc/free per call. The default value of
4689 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
4690 when configuring.
4691
4692 32. The default maximum size of a compiled pattern is 64K. There have been a
4693 few cases of people hitting this limit. The code now uses macros to handle the
4694 storing of links as offsets within the compiled pattern. It defaults to 2-byte
4695 links, but this can be changed to 3 or 4 bytes by --with-link-size when
4696 configuring. Tests 2 and 5 work only with 2-byte links because they output
4697 debugging information about compiled patterns.
4698
4699 33. Internal code re-arrangements:
4700
4701 (a) Moved the debugging function for printing out a compiled regex into
4702 its own source file (printint.c) and used #include to pull it into
4703 pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
4704 separate copies.
4705
4706 (b) Defined the list of op-code names for debugging as a macro in
4707 internal.h so that it is next to the definition of the opcodes.
4708
4709 (c) Defined a table of op-code lengths for simpler skipping along compiled
4710 code. This is again a macro in internal.h so that it is next to the
4711 definition of the opcodes.
4712
4713 34. Added support for recursive calls to individual subpatterns, along the
4714 lines of Robin Houston's patch (but implemented somewhat differently).
4715
4716 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
4717 allow it to read and process whole directories in Win32. This code was
4718 contributed by Lionel Fourquaux; it has not been tested by me.
4719
4720 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
4721 used to name a group. Names consist of alphanumerics and underscores, and must
4722 be unique. Back references use the syntax (?P=name) and recursive calls use
4723 (?P>name) which is a PCRE extension to the Python extension. Groups still have
4724 numbers. The function pcre_fullinfo() can be used after compilation to extract
4725 a name/number map. There are three relevant calls:
4726
4727 PCRE_INFO_NAMEENTRYSIZE yields the size of each entry in the map
4728 PCRE_INFO_NAMECOUNT yields the number of entries
4729 PCRE_INFO_NAMETABLE yields a pointer to the map.
4730
4731 The map is a vector of fixed-size entries. The size of each entry depends on
4732 the length of the longest name used. The first two bytes of each entry are the
4733 group number, most significant byte first. There follows the corresponding
4734 name, zero terminated. The names are in alphabetical order.
4735
4736 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
4737 case instead of 255. Making it the same both with and without UTF-8 support
4738 means that the same test output works with both.
4739
4740 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
4741 calling malloc() with a zero argument.
4742
4743 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
4744 optimization. I've improved things by keeping a bitmap of backreferences with
4745 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
4746 fact referenced, the optimization can be applied. It is unlikely that a
4747 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
4748 the match to follow \n) will appear inside brackets with a number greater than
4749 31, but if it does, any back reference > 31 suppresses the optimization.
4750
4751 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
4752 of disabling numbered capturing parentheses. Any opening parenthesis that is
4753 not followed by ? behaves as if it were followed by ?: but named parentheses
4754 can still be used for capturing (and they will acquire numbers in the usual
4755 way).
4756
4757 41. Redesigned the return codes from the match() function into yes/no/error so
4758 that errors can be passed back from deep inside the nested calls. A malloc
4759 failure while inside a recursive subpattern call now causes the
4760 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
4761
4762 42. It is now possible to set a limit on the number of times the match()
4763 function is called in a call to pcre_exec(). This facility makes it possible to
4764 limit the amount of recursion and backtracking, though not in a directly
4765 obvious way, because the match() function is used in a number of different
4766 circumstances. The count starts from zero for each position in the subject
4767 string (for non-anchored patterns). The default limit is, for compatibility, a
4768 large number, namely 10 000 000. You can change this in two ways:
4769
4770 (a) When configuring PCRE before making, you can use --with-match-limit=n
4771 to set a default value for the compiled library.
4772
4773 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
4774 a different value is set. See 45 below.
4775
4776 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
4777
4778 43. Added a new function pcre_config(int, void *) to enable run-time extraction
4779 of things that can be changed at compile time. The first argument specifies
4780 what is wanted and the second points to where the information is to be placed.
4781 The current list of available information is:
4782
4783 PCRE_CONFIG_UTF8
4784
4785 The output is an integer that is set to one if UTF-8 support is available;
4786 otherwise it is set to zero.
4787
4788 PCRE_CONFIG_NEWLINE
4789
4790 The output is an integer that it set to the value of the code that is used for
4791 newline. It is either LF (10) or CR (13).
4792
4793 PCRE_CONFIG_LINK_SIZE
4794
4795 The output is an integer that contains the number of bytes used for internal
4796 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
4797
4798 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
4799
4800 The output is an integer that contains the threshold above which the POSIX
4801 interface uses malloc() for output vectors. See item 31 above.
4802
4803 PCRE_CONFIG_MATCH_LIMIT
4804
4805 The output is an unsigned integer that contains the default limit of the number
4806 of match() calls in a pcre_exec() execution. See 42 above.
4807
4808 44. pcretest has been upgraded by the addition of the -C option. This causes it
4809 to extract all the available output from the new pcre_config() function, and to
4810 output it. The program then exits immediately.
4811
4812 45. A need has arisen to pass over additional data with calls to pcre_exec() in
4813 order to support additional features. One way would have been to define
4814 pcre_exec2() (for example) with extra arguments, but this would not have been
4815 extensible, and would also have required all calls to the original function to
4816 be mapped to the new one. Instead, I have chosen to extend the mechanism that
4817 is used for passing in "extra" data from pcre_study().
4818
4819 The pcre_extra structure is now exposed and defined in pcre.h. It currently
4820 contains the following fields:
4821
4822 flags a bitmap indicating which of the following fields are set
4823 study_data opaque data from pcre_study()
4824 match_limit a way of specifying a limit on match() calls for a specific
4825 call to pcre_exec()
4826 callout_data data for callouts (see 49 below)
4827
4828 The flag bits are also defined in pcre.h, and are
4829
4830 PCRE_EXTRA_STUDY_DATA
4831 PCRE_EXTRA_MATCH_LIMIT
4832 PCRE_EXTRA_CALLOUT_DATA
4833
4834 The pcre_study() function now returns one of these new pcre_extra blocks, with
4835 the actual study data pointed to by the study_data field, and the
4836 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
4837 before. That is, this change is entirely upwards-compatible and requires no
4838 change to existing code.
4839
4840 If you want to pass in additional data to pcre_exec(), you can either place it
4841 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
4842 block.
4843
4844 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
4845 data string contains the escape sequence \M, pcretest calls pcre_exec() several
4846 times with different match limits, until it finds the minimum value needed for
4847 pcre_exec() to complete. The value is then output. This can be instructive; for
4848 most simple matches the number is quite small, but for pathological cases it
4849 gets very large very quickly.
4850
4851 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
4852 returns the size of the data block pointed to by the study_data field in a
4853 pcre_extra block, that is, the value that was passed as the argument to
4854 pcre_malloc() when PCRE was getting memory in which to place the information
4855 created by pcre_study(). The fourth argument should point to a size_t variable.
4856 pcretest has been extended so that this information is shown after a successful
4857 pcre_study() call when information about the compiled regex is being displayed.
4858
4859 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
4860 because what follows is always an absolute path. (Later: it turns out that this
4861 is more than cosmetic for MinGW, because it doesn't like empty path
4862 components.)
4863
4864 49. Some changes have been made to the callout feature (see 28 above):
4865
4866 (i) A callout function now has three choices for what it returns:
4867
4868 0 => success, carry on matching
4869 > 0 => failure at this point, but backtrack if possible
4870 < 0 => serious error, return this value from pcre_exec()
4871
4872 Negative values should normally be chosen from the set of PCRE_ERROR_xxx
4873 values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
4874 "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
4875 use by callout functions. It will never be used by PCRE itself.
4876
4877 (ii) The pcre_extra structure (see 45 above) has a void * field called
4878 callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
4879 pcre_callout_block structure has a field of the same name. The contents of
4880 the field passed in the pcre_extra structure are passed to the callout
4881 function in the corresponding field in the callout block. This makes it
4882 easier to use the same callout-containing regex from multiple threads. For
4883 testing, the pcretest program has a new data escape
4884
4885 \C*n pass the number n (may be negative) as callout_data
4886
4887 If the callout function in pcretest receives a non-zero value as
4888 callout_data, it returns that value.
4889
4890 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
4891 there were some redundant $(CFLAGS) in commands that are now specified as
4892 $(LINK), which already includes $(CFLAGS).
4893
4894 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
4895 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
4896 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
4897 one-byte characters throughout. Note that case-insensitive matching applies
4898 only to characters whose values are less than 256. PCRE doesn't support the
4899 notion of cases for higher-valued characters.
4900
4901 (i) A character class whose characters are all within 0-255 is handled as
4902 a bit map, and the map is inverted for negative classes. Previously, a
4903 character > 255 always failed to match such a class; however it should
4904 match if the class was a negative one (e.g. [^ab]). This has been fixed.
4905
4906 (ii) A negated character class with a single character < 255 is coded as
4907 "not this character" (OP_NOT). This wasn't working properly when the test
4908 character was multibyte, either singly or repeated.
4909
4910 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
4911 mode, for example: \x{100}{2,3}.
4912
4913 (iv) The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
4914 singly or repeated) now correctly test multibyte characters. However,
4915 PCRE doesn't recognize any characters with values greater than 255 as
4916 digits, spaces, or word characters. Such characters always match \D, \S,
4917 and \W, and never match \d, \s, or \w.
4918
4919 (v) Classes may now contain characters and character ranges with values
4920 greater than 255. For example: [ab\x{100}-\x{400}].
4921
4922 (vi) pcregrep now has a --utf-8 option (synonym -u) which makes it call
4923 PCRE in UTF-8 mode.
4924
4925 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
4926 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
4927 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
4928 value.)
4929
4930 53. The single man page has become too large. I have therefore split it up into
4931 a number of separate man pages. These also give rise to individual HTML pages;
4932 these are now put in a separate directory, and there is an index.html page that
4933 lists them all. Some hyperlinking between the pages has been installed.
4934
4935 54. Added convenience functions for handling named capturing parentheses.
4936
4937 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
4938 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
4939 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
4940 are faulted.
4941
4942 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
4943 calling configure. These values are used when compiling the dftables.c program
4944 which is run to generate the source of the default character tables. They
4945 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
4946 you will need to set these values.
4947
4948 57. Updated the building process for Windows DLL, as provided by Fred Cox.
4949
4950
4951 Version 3.9 02-Jan-02
4952 ---------------------
4953
4954 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
4955
4956 2. If --disable-static was given, the building process failed when trying to
4957 build pcretest and pcregrep. (For some reason it was using libtool to compile
4958 them, which is not right, as they aren't part of the library.)
4959
4960
4961 Version 3.8 18-Dec-01
4962 ---------------------
4963
4964 1. The experimental UTF-8 code was completely screwed up. It was packing the
4965 bytes in the wrong order. How dumb can you get?
4966
4967
4968 Version 3.7 29-Oct-01
4969 ---------------------
4970
4971 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
4972 This caused pcretest, when used on the test data, to segfault. Unfortunately,
4973 this didn't happen under Solaris 8, where I normally test things.
4974
4975 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
4976 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
4977 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
4978 this fix an hour or so after the initial 3.7 release.)
4979
4980
4981 Version 3.6 23-Oct-01
4982 ---------------------
4983
4984 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
4985 offsets passed as NULL with zero offset count.
4986
4987 2. The config.guess and config.sub files had not been updated when I moved to
4988 the latest autoconf.
4989
4990
4991 Version 3.5 15-Aug-01
4992 ---------------------
4993
4994 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
4995 had been forgotten.
4996
4997 2. By using declared but undefined structures, we can avoid using "void"
4998 definitions in pcre.h while keeping the internal definitions of the structures
4999 private.
5000
5001 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
5002 user point of view, this means that both static and shared libraries are built
5003 by default, but this can be individually controlled. More of the work of
5004 handling this static/shared cases is now inside libtool instead of PCRE's make
5005 file.
5006
5007 4. The pcretest utility is now installed along with pcregrep because it is
5008 useful for users (to test regexs) and by doing this, it automatically gets
5009 relinked by libtool. The documentation has been turned into a man page, so
5010 there are now .1, .txt, and .html versions in /doc.
5011
5012 5. Upgrades to pcregrep:
5013 (i) Added long-form option names like gnu grep.
5014 (ii) Added --help to list all options with an explanatory phrase.
5015 (iii) Added -r, --recursive to recurse into sub-directories.
5016 (iv) Added -f, --file to read patterns from a file.
5017
5018 6. pcre_exec() was referring to its "code" argument before testing that
5019 argument for NULL (and giving an error if it was NULL).
5020
5021 7. Upgraded Makefile.in to allow for compiling in a different directory from
5022 the source directory.
5023
5024 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
5025 options bits, the pointer it was passed was to an int instead of to an unsigned
5026 long int. This mattered only on 64-bit systems.
5027
5028 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
5029 generated) instead of pcre.in, which it its source. Also made the same change
5030 in several of the .c files.
5031
5032 10. A new release of gcc defines printf() as a macro, which broke pcretest
5033 because it had an ifdef in the middle of a string argument for printf(). Fixed
5034 by using separate calls to printf().
5035
5036 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
5037 script, to force use of CR or LF instead of \n in the source. On non-Unix
5038 systems, the value can be set in config.h.
5039
5040 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
5041 absolute limit. Changed the text of the error message to make this clear, and
5042 likewise updated the man page.
5043
5044 13. The limit of 99 on the number of capturing subpatterns has been removed.
5045 The new limit is 65535, which I hope will not be a "real" limit.
5046
5047
5048 Version 3.4 22-Aug-00
5049 ---------------------
5050
5051 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
5052
5053 2. Diagnose condition (?(0) as an error instead of crashing on matching.
5054
5055
5056 Version 3.3 01-Aug-00
5057 ---------------------
5058
5059 1. If an octal character was given, but the value was greater than \377, it
5060 was not getting masked to the least significant bits, as documented. This could
5061 lead to crashes in some systems.
5062
5063 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
5064 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
5065
5066 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
5067 These just pass their arguments on to (pcre_free)(), but they are provided
5068 because some uses of PCRE bind it to non-C systems that can call its functions,
5069 but cannot call free() or pcre_free() directly.
5070
5071 4. Add "make test" as a synonym for "make check". Corrected some comments in
5072 the Makefile.
5073
5074 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
5075 Makefile.
5076
5077 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
5078 command called pgrep for grepping around the active processes.
5079
5080 7. Added the beginnings of support for UTF-8 character strings.
5081
5082 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
5083 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
5084 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
5085 out for the ar command.)
5086
5087
5088 Version 3.2 12-May-00
5089 ---------------------
5090
5091 This is purely a bug fixing release.
5092
5093 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
5094 of ZA. This was just one example of several cases that could provoke this bug,
5095 which was introduced by change 9 of version 2.00. The code for breaking
5096 infinite loops after an iteration that matches an empty string was't working
5097 correctly.
5098
5099 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
5100 when matched against abbab (for example). After matching an empty string, it
5101 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
5102 caused it to match further down the string than it should.
5103
5104 3. The code contained an inclusion of sys/types.h. It isn't clear why this
5105 was there because it doesn't seem to be needed, and it causes trouble on some
5106 systems, as it is not a Standard C header. It has been removed.
5107
5108 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
5109 were reported on the Macintosh. The changes were from
5110
5111 while ((c = *(++ptr)) != 0 && c != '\n');
5112 to
5113 while ((c = *(++ptr)) != 0 && c != '\n') ;
5114
5115 Totally extraordinary, but if that's what it takes...
5116
5117 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
5118 available. Added HAVE_BCOPY and an autoconf test for it; if neither
5119 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
5120 assumes the way PCRE uses memmove() (always moving upwards).
5121
5122 6. PCRE is being used in one environment where strchr() is not available. There
5123 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
5124 faster code anyway.
5125
5126
5127 Version 3.1 09-Feb-00
5128 ---------------------
5129
5130 The only change in this release is the fixing of some bugs in Makefile.in for
5131 the "install" target:
5132
5133 (1) It was failing to install pcreposix.h.
5134
5135 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
5136
5137
5138 Version 3.0 01-Feb-00
5139 ---------------------
5140
5141 1. Add support for the /+ modifier to perltest (to output $` like it does in
5142 pcretest).
5143
5144 2. Add support for the /g modifier to perltest.
5145
5146 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
5147 matches null strings.
5148
5149 4. Fix perltest so that it doesn't do unwanted things when fed an empty
5150 pattern. Perl treats empty patterns specially - it reuses the most recent
5151 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
5152 effect.
5153
5154 5. The POSIX interface was broken in that it was just handing over the POSIX
5155 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
5156 required a bigger vector, with some working space on the end. This means that
5157 the POSIX wrapper now has to get and free some memory, and copy the results.
5158
5159 6. Added some simple autoconf support, placing the test data and the
5160 documentation in separate directories, re-organizing some of the
5161 information files, and making it build pcre-config (a GNU standard). Also added
5162 libtool support for building PCRE as a shared library, which is now the
5163 default.
5164
5165 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
5166 09 are not valid octal constants. Single digits will be used for minor values
5167 less than 10.
5168
5169 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
5170 existing programs that set these in the POSIX interface can use PCRE without
5171 modification.
5172
5173 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
5174 return all that pcre_info() returns, plus additional data. The pcre_info()
5175 function is retained for compatibility, but is considered to be obsolete.
5176
5177 10. Added experimental recursion feature (?R) to handle one common case that
5178 Perl 5.6 will be able to do with (?p{...}).
5179
5180 11. Added support for POSIX character classes like [:alpha:], which Perl is
5181 adopting.
5182
5183
5184 Version 2.08 31-Aug-99
5185 ----------------------
5186
5187 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
5188 trying to match at the startoffset position, but instead was moving forward to
5189 the next newline as if a previous match had failed.
5190
5191 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
5192 and could get into a loop if a null string was matched other than at the start
5193 of the subject.
5194
5195 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
5196 be distinguished at compile time, and for completeness also added PCRE_DATE.
5197
5198 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
5199 in GnuWin32 environments.
5200
5201
5202 Version 2.07 29-Jul-99
5203 ----------------------
5204
5205 1. The documentation is now supplied in plain text form and HTML as well as in
5206 the form of man page sources.
5207
5208 2. C++ compilers don't like assigning (void *) values to other pointer types.
5209 In particular this affects malloc(). Although there is no problem in Standard
5210 C, I've put in casts to keep C++ compilers happy.
5211
5212 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
5213 should be (const char *).
5214
5215 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
5216 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
5217 However, I haven't made this a standard facility. The documentation doesn't
5218 mention it, and the Makefile doesn't support it.
5219
5220 5. The Makefile now contains an "install" target, with editable destinations at
5221 the top of the file. The pcretest program is not installed.
5222
5223 6. pgrep -V now gives the PCRE version number and date.
5224
5225 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
5226 causing the entire string to be ignored, instead of just the last character.
5227
5228 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
5229 non-matching string, it can take a very, very long time, even for strings of
5230 quite modest length, because of the nested recursion. PCRE now does better in
5231 some of these cases. It does this by remembering the last required literal
5232 character in the pattern, and pre-searching the subject to ensure it is present
5233 before running the real match. In other words, it applies a heuristic to detect
5234 some types of certain failure quickly, and in the above example, if presented
5235 with a string that has no trailing " it gives "no match" very quickly.
5236
5237 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
5238 other alternatives are tried instead.
5239
5240
5241 Version 2.06 09-Jun-99
5242 ----------------------
5243
5244 1. Change pcretest's output for amount of store used to show just the code
5245 space, because the remainder (the data block) varies in size between 32-bit and
5246 64-bit systems.
5247
5248 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
5249 start matching at. This allows lookbehinds to work when searching for multiple
5250 occurrences in a string.
5251
5252 3. Added additional options to pcretest for testing multiple occurrences:
5253
5254 /+ outputs the rest of the string that follows a match
5255 /g loops for multiple occurrences, using the new startoffset argument
5256 /G loops for multiple occurrences by passing an incremented pointer
5257
5258 4. PCRE wasn't doing the "first character" optimization for patterns starting
5259 with \b or \B, though it was doing it for other lookbehind assertions. That is,
5260 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
5261 the letter 'x'. On long subject strings, this gives a significant speed-up.
5262
5263
5264 Version 2.05 21-Apr-99
5265 ----------------------
5266
5267 1. Changed the type of magic_number from int to long int so that it works
5268 properly on 16-bit systems.
5269
5270 2. Fixed a bug which caused patterns starting with .* not to work correctly
5271 when the subject string contained newline characters. PCRE was assuming
5272 anchoring for such patterns in all cases, which is not correct because .* will
5273 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
5274 DOTALL is set at top level; otherwise it knows that patterns starting with .*
5275 must be retried after every newline in the subject.
5276
5277
5278 Version 2.04 18-Feb-99
5279 ----------------------
5280
5281 1. For parenthesized subpatterns with repeats whose minimum was zero, the
5282 computation of the store needed to hold the pattern was incorrect (too large).
5283 If such patterns were nested a few deep, this could multiply and become a real
5284 problem.
5285
5286 2. Added /M option to pcretest to show the memory requirement of a specific
5287 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
5288
5289 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
5290 compiled in such a way that the backtracking after subsequent failure was
5291 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
5292 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
5293
5294
5295 Version 2.03 02-Feb-99
5296 ----------------------
5297
5298 1. Fixed typo and small mistake in man page.
5299
5300 2. Added 4th condition (GPL supersedes if conflict) and created separate
5301 LICENCE file containing the conditions.
5302
5303 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
5304 Perl, that is the internal \ allows the delimiter to be included in the
5305 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
5306 the final delimiter, add \ to the end of the pattern (to test the error).
5307
5308 4. Added the convenience functions for extracting substrings after a successful
5309 match. Updated pcretest to make it able to test these functions.
5310
5311
5312 Version 2.02 14-Jan-99
5313 ----------------------
5314
5315 1. Initialized the working variables associated with each extraction so that
5316 their saving and restoring doesn't refer to uninitialized store.
5317
5318 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
5319 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
5320 fix the problem.
5321
5322 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
5323 calls, and wasn't printing the correct value for compiling calls. Increased the
5324 default value of LOOPREPEAT, and the number of significant figures in the
5325 times.
5326
5327 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
5328
5329 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
5330 a building problem on Windows NT with a FAT file system.
5331
5332
5333 Version 2.01 21-Oct-98
5334 ----------------------
5335
5336 1. Changed the API for pcre_compile() to allow for the provision of a pointer
5337 to character tables built by pcre_maketables() in the current locale. If NULL
5338 is passed, the default tables are used.
5339
5340
5341 Version 2.00 24-Sep-98
5342 ----------------------
5343
5344 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
5345 it any more.
5346
5347 2. Allow quantification of (?>) groups, and make it work correctly.
5348
5349 3. The first character computation wasn't working for (?>) groups.
5350
5351 4. Correct the implementation of \Z (it is permitted to match on the \n at the
5352 end of the subject) and add 5.005's \z, which really does match only at the
5353 very end of the subject.
5354
5355 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
5356
5357 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
5358 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
5359 localized options. All options to pcre_study() were also removed.
5360
5361 7. Add other new features from 5.005:
5362
5363 $(?<= positive lookbehind
5364 $(?<! negative lookbehind
5365 (?imsx-imsx) added the unsetting capability
5366 such a setting is global if at outer level; local otherwise
5367 (?imsx-imsx:) non-capturing groups with option setting
5368 (?(cond)re|re) conditional pattern matching
5369
5370 A backreference to itself in a repeated group matches the previous
5371 captured string.
5372
5373 8. General tidying up of studying (both automatic and via "study")
5374 consequential on the addition of new assertions.
5375
5376 9. As in 5.005, unlimited repeated groups that could match an empty substring
5377 are no longer faulted at compile time. Instead, the loop is forcibly broken at
5378 runtime if any iteration does actually match an empty substring.
5379
5380 10. Include the RunTest script in the distribution.
5381
5382 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
5383 discrepancies, some of which were old and were also with respect to 5.004. They
5384 have now been fixed.
5385
5386
5387 Version 1.09 28-Apr-98
5388 ----------------------
5389
5390 1. A negated single character class followed by a quantifier with a minimum
5391 value of one (e.g. [^x]{1,6} ) was not compiled correctly. This could lead to
5392 program crashes, or just wrong answers. This did not apply to negated classes
5393 containing more than one character, or to minima other than one.
5394
5395
5396 Version 1.08 27-Mar-98
5397 ----------------------
5398
5399 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
5400
5401 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
5402 latter must appear before anything that relies on it in the pattern.
5403
5404
5405 Version 1.07 16-Feb-98
5406 ----------------------
5407
5408 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
5409 repeat of a potentially empty string).
5410
5411
5412 Version 1.06 23-Jan-98
5413 ----------------------
5414
5415 1. Added Markus Oberhumer's little patches for C++.
5416
5417 2. Literal strings longer than 255 characters were broken.
5418
5419
5420 Version 1.05 23-Dec-97
5421 ----------------------
5422
5423 1. Negated character classes containing more than one character were failing if
5424 PCRE_CASELESS was set at run time.
5425
5426
5427 Version 1.04 19-Dec-97
5428 ----------------------
5429
5430 1. Corrected the man page, where some "const" qualifiers had been omitted.
5431
5432 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
5433 input syntax.
5434
5435 3. Fixed memory leak which occurred when a regex with back references was
5436 matched with an offsets vector that wasn't big enough. The temporary memory
5437 that is used in this case wasn't being freed if the match failed.
5438
5439 4. Tidied pcretest to ensure it frees memory that it gets.
5440
5441 5. Temporary memory was being obtained in the case where the passed offsets
5442 vector was exactly big enough.
5443
5444 6. Corrected definition of offsetof() from change 5 below.
5445
5446 7. I had screwed up change 6 below and broken the rules for the use of
5447 setjmp(). Now fixed.
5448
5449
5450 Version 1.03 18-Dec-97
5451 ----------------------
5452
5453 1. A erroneous regex with a missing opening parenthesis was correctly
5454 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
5455 on some systems.
5456
5457 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
5458 it was reported that one broken compiler failed on the former because "code" is
5459 also an independent variable.
5460
5461 3. The erroneous regex a[]b caused an array overrun reference.
5462
5463 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
5464 fail on data ending with that character. (It was going on too far, and checking
5465 the next character, typically a binary zero.) This was specific to the
5466 optimized code for single-character negative classes.
5467
5468 5. Added a contributed patch from the TIN world which does the following:
5469
5470 + Add an undef for memmove, in case the the system defines a macro for it.
5471
5472 + Add a definition of offsetof(), in case there isn't one. (I don't know
5473 the reason behind this - offsetof() is part of the ANSI standard - but
5474 it does no harm).
5475
5476 + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
5477 most of the places where whitespace preceded '#'. I have given up and
5478 allowed the remaining 2 cases to be at the margin.
5479
5480 + Rename some variables in pcre to eliminate shadowing. This seems very
5481 pedantic, but does no harm, of course.
5482
5483 6. Moved the call to setjmp() into its own function, to get rid of warnings
5484 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
5485
5486 7. Constructs such as \d{8,} were compiling into the equivalent of
5487 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
5488 outcome, but in this particular case used more store than had been allocated,
5489 which caused the bug to be discovered because it threw up an internal error.
5490
5491 8. The debugging code in both pcre and pcretest for outputting the compiled
5492 form of a regex was going wrong in the case of back references followed by
5493 curly-bracketed repeats.
5494
5495
5496 Version 1.02 12-Dec-97
5497 ----------------------
5498
5499 1. Typos in pcre.3 and comments in the source fixed.
5500
5501 2. Applied a contributed patch to get rid of places where it used to remove
5502 'const' from variables, and fixed some signed/unsigned and uninitialized
5503 variable warnings.
5504
5505 3. Added the "runtest" target to Makefile.
5506
5507 4. Set default compiler flag to -O2 rather than just -O.
5508
5509
5510 Version 1.01 19-Nov-97
5511 ----------------------
5512
5513 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
5514 like /([ab]*)*/, that is, for classes with more than one character in them.
5515
5516 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
5517 as /((?>a*))*/ (a PCRE_EXTRA facility).
5518
5519
5520 Version 1.00 18-Nov-97
5521 ----------------------
5522
5523 1. Added compile-time macros to support systems such as SunOS4 which don't have
5524 memmove() or strerror() but have other things that can be used instead.
5525
5526 2. Arranged that "make clean" removes the executables.
5527
5528
5529 Version 0.99 27-Oct-97
5530 ----------------------
5531
5532 1. Fixed bug in code for optimizing classes with only one character. It was
5533 initializing a 32-byte map regardless, which could cause it to run off the end
5534 of the memory it had got.
5535
5536 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
5537
5538
5539 Version 0.98 22-Oct-97
5540 ----------------------
5541
5542 1. Fixed bug in code for handling temporary memory usage when there are more
5543 back references than supplied space in the ovector. This could cause segfaults.
5544
5545
5546 Version 0.97 21-Oct-97
5547 ----------------------
5548
5549 1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
5550
5551 2. Optimized negated single characters not to use a bit map.
5552
5553 3. Brought error texts together as macro definitions; clarified some of them;
5554 fixed one that was wrong - it said "range out of order" when it meant "invalid
5555 escape sequence".
5556
5557 4. Changed some char * arguments to const char *.
5558
5559 5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
5560
5561 6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
5562 pcretest.
5563
5564
5565 Version 0.96 16-Oct-97
5566 ----------------------
5567
5568 1. Added a simple "pgrep" utility to the distribution.
5569
5570 2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
5571 unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
5572 where "ddd" means "one or more decimal digits".
5573
5574 3. Fixed serious bug. If a pattern had a back reference, but the call to
5575 pcre_exec() didn't supply a large enough ovector to record the related
5576 identifying subpattern, the match always failed. PCRE now remembers the number
5577 of the largest back reference, and gets some temporary memory in which to save
5578 the offsets during matching if necessary, in order to ensure that
5579 backreferences always work.
5580
5581 4. Increased the compatibility with Perl in a number of ways:
5582
5583 (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
5584 to request this handling. The option can be set at compile or exec time.
5585
5586 (b) $ matches before a terminating newline by default; an option
5587 PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
5588 mode). The option can be set at compile or exec time.
5589
5590 (c) The handling of \ followed by a digit other than 0 is now supposed to be
5591 the same as Perl's. If the decimal number it represents is less than 10
5592 or there aren't that many previous left capturing parentheses, an octal
5593 escape is read. Inside a character class, it's always an octal escape,
5594 even if it is a single digit.
5595
5596 (d) An escaped but undefined alphabetic character is taken as a literal,
5597 unless PCRE_EXTRA is set. Currently this just reserves the remaining
5598 escapes.
5599
5600 (e) {0} is now permitted. (The previous item is removed from the compiled
5601 pattern).
5602
5603 5. Changed all the names of code files so that the basic parts are no longer
5604 than 10 characters, and abolished the teeny "globals.c" file.
5605
5606 6. Changed the handling of character classes; they are now done with a 32-byte
5607 bit map always.
5608
5609 7. Added the -d and /D options to pcretest to make it possible to look at the
5610 internals of compilation without having to recompile pcre.
5611
5612
5613 Version 0.95 23-Sep-97
5614 ----------------------
5615
5616 1. Fixed bug in pre-pass concerning escaped "normal" characters such as \x5c or
5617 \x20 at the start of a run of normal characters. These were being treated as
5618 real characters, instead of the source characters being re-checked.
5619
5620
5621 Version 0.94 18-Sep-97
5622 ----------------------
5623
5624 1. The functions are now thread-safe, with the caveat that the global variables
5625 containing pointers to malloc() and free() or alternative functions are the
5626 same for all threads.
5627
5628 2. Get pcre_study() to generate a bitmap of initial characters for non-
5629 anchored patterns when this is possible, and use it if passed to pcre_exec().
5630
5631
5632 Version 0.93 15-Sep-97
5633 ----------------------
5634
5635 1. /(b)|(:+)/ was computing an incorrect first character.
5636
5637 2. Add pcre_study() to the API and the passing of pcre_extra to pcre_exec(),
5638 but not actually doing anything yet.
5639
5640 3. Treat "-" characters in classes that cannot be part of ranges as literals,
5641 as Perl does (e.g. [-az] or [az-]).
5642
5643 4. Set the anchored flag if a branch starts with .* or .*? because that tests
5644 all possible positions.
5645
5646 5. Split up into different modules to avoid including unneeded functions in a
5647 compiled binary. However, compile and exec are still in one module. The "study"
5648 function is split off.
5649
5650 6. The character tables are now in a separate module whose source is generated
5651 by an auxiliary program - but can then be edited by hand if required. There are
5652 now no calls to isalnum(), isspace(), isdigit(), isxdigit(), tolower() or
5653 toupper() in the code.
5654
5655 7. Turn the malloc/free funtions variables into pcre_malloc and pcre_free and
5656 make them global. Abolish the function for setting them, as the caller can now
5657 set them directly.
5658
5659
5660 Version 0.92 11-Sep-97
5661 ----------------------
5662
5663 1. A repeat with a fixed maximum and a minimum of 1 for an ordinary character
5664 (e.g. /a{1,3}/) was broken (I mis-optimized it).
5665
5666 2. Caseless matching was not working in character classes if the characters in
5667 the pattern were in upper case.
5668
5669 3. Make ranges like [W-c] work in the same way as Perl for caseless matching.
5670
5671 4. Make PCRE_ANCHORED public and accept as a compile option.
5672
5673 5. Add an options word to pcre_exec() and accept PCRE_ANCHORED and
5674 PCRE_CASELESS at run time. Add escapes \A and \I to pcretest to cause it to
5675 pass them.
5676
5677 6. Give an error if bad option bits passed at compile or run time.
5678
5679 7. Add PCRE_MULTILINE at compile and exec time, and (?m) as well. Add \M to
5680 pcretest to cause it to pass that flag.
5681
5682 8. Add pcre_info(), to get the number of identifying subpatterns, the stored
5683 options, and the first character, if set.
5684
5685 9. Recognize C+ or C{n,m} where n >= 1 as providing a fixed starting character.
5686
5687
5688 Version 0.91 10-Sep-97
5689 ----------------------
5690
5691 1. PCRE was failing to diagnose unlimited repeats of subpatterns that could
5692 match the empty string as in /(a*)*/. It was looping and ultimately crashing.
5693
5694 2. PCRE was looping on encountering an indefinitely repeated back reference to
5695 a subpattern that had matched an empty string, e.g. /(a|)\1*/. It now does what
5696 Perl does - treats the match as successful.
5697
5698 ****

Properties

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

  ViewVC Help
Powered by ViewVC 1.1.5