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

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1542 - (show annotations) (download)
Wed Apr 1 15:43:53 2015 UTC (15 hours, 2 minutes ago) by ph10
File size: 276200 byte(s)
Fix stack overflow instead of diagnostic for mutual recursion inside a 
lookbehind assertion.

1 ChangeLog for PCRE
2 ------------------
3
4 Version 8.37 xx-xxx-2015
5 ------------------------
6
7 1. When an (*ACCEPT) is triggered inside capturing parentheses, it arranges
8 for those parentheses to be closed with whatever has been captured so far.
9 However, it was failing to mark any other groups between the hightest
10 capture so far and the currrent group as "unset". Thus, the ovector for
11 those groups contained whatever was previously there. An example is the
12 pattern /(x)|((*ACCEPT))/ when matched against "abcd".
13
14 2. If an assertion condition was quantified with a minimum of zero (an odd
15 thing to do, but it happened), SIGSEGV or other misbehaviour could occur.
16
17 3. If a pattern in pcretest input had the P (POSIX) modifier followed by an
18 unrecognized modifier, a crash could occur.
19
20 4. An attempt to do global matching in pcretest with a zero-length ovector
21 caused a crash.
22
23 5. Fixed a memory leak during matching that could occur for a subpattern
24 subroutine call (recursive or otherwise) if the number of captured groups
25 that had to be saved was greater than ten.
26
27 6. Catch a bad opcode during auto-possessification after compiling a bad UTF
28 string with NO_UTF_CHECK. This is a tidyup, not a bug fix, as passing bad
29 UTF with NO_UTF_CHECK is documented as having an undefined outcome.
30
31 7. A UTF pattern containing a "not" match of a non-ASCII character and a
32 subroutine reference could loop at compile time. Example: /[^\xff]((?1))/.
33
34 8. When a pattern is compiled, it remembers the highest back reference so that
35 when matching, if the ovector is too small, extra memory can be obtained to
36 use instead. A conditional subpattern whose condition is a check on a
37 capture having happened, such as, for example in the pattern
38 /^(?:(a)|b)(?(1)A|B)/, is another kind of back reference, but it was not
39 setting the highest backreference number. This mattered only if pcre_exec()
40 was called with an ovector that was too small to hold the capture, and there
41 was no other kind of back reference (a situation which is probably quite
42 rare). The effect of the bug was that the condition was always treated as
43 FALSE when the capture could not be consulted, leading to a incorrect
44 behaviour by pcre2_match(). This bug has been fixed.
45
46 9. A reference to a duplicated named group (either a back reference or a test
47 for being set in a conditional) that occurred in a part of the pattern where
48 PCRE_DUPNAMES was not set caused the amount of memory needed for the pattern
49 to be incorrectly calculated, leading to overwriting.
50
51 10. A mutually recursive set of back references such as (\2)(\1) caused a
52 segfault at study time (while trying to find the minimum matching length).
53 The infinite loop is now broken (with the minimum length unset, that is,
54 zero).
55
56 11. If an assertion that was used as a condition was quantified with a minimum
57 of zero, matching went wrong. In particular, if the whole group had
58 unlimited repetition and could match an empty string, a segfault was
59 likely. The pattern (?(?=0)?)+ is an example that caused this. Perl allows
60 assertions to be quantified, but not if they are being used as conditions,
61 so the above pattern is faulted by Perl. PCRE has now been changed so that
62 it also rejects such patterns.
63
64 12. A possessive capturing group such as (a)*+ with a minimum repeat of zero
65 failed to allow the zero-repeat case if pcre2_exec() was called with an
66 ovector too small to capture the group.
67
68 13. Fixed two bugs in pcretest that were discovered by fuzzing and reported by
69 Red Hat Product Security:
70
71 (a) A crash if /K and /F were both set with the option to save the compiled
72 pattern.
73
74 (b) Another crash if the option to print captured substrings in a callout
75 was combined with setting a null ovector, for example \O\C+ as a subject
76 string.
77
78 14. A pattern such as "((?2){0,1999}())?", which has a group containing a
79 forward reference repeated a large (but limited) number of times within a
80 repeated outer group that has a zero minimum quantifier, caused incorrect
81 code to be compiled, leading to the error "internal error:
82 previously-checked referenced subpattern not found" when an incorrect
83 memory address was read. This bug was reported as "heap overflow",
84 discovered by Kai Lu of Fortinet's FortiGuard Labs and given the CVE number
85 CVE-2015-2325.
86
87 23. A pattern such as "((?+1)(\1))/" containing a forward reference subroutine
88 call within a group that also contained a recursive back reference caused
89 incorrect code to be compiled. This bug was reported as "heap overflow",
90 discovered by Kai Lu of Fortinet's FortiGuard Labs, and given the CVE
91 number CVE-2015-2326.
92
93 24. Computing the size of the JIT read-only data in advance has been a source
94 of various issues, and new ones are still appear unfortunately. To fix
95 existing and future issues, size computation is eliminated from the code,
96 and replaced by on-demand memory allocation.
97
98 25. A pattern such as /(?i)[A-`]/, where characters in the other case are
99 adjacent to the end of the range, and the range contained characters with
100 more than one other case, caused incorrect behaviour when compiled in UTF
101 mode. In that example, the range a-j was left out of the class.
102
103 26. Fix JIT compilation of conditional blocks, which assertion
104 is converted to (*FAIL). E.g: /(?(?!))/.
105
106 27. The pattern /(?(?!)^)/ caused references to random memory. This bug was
107 discovered by the LLVM fuzzer.
108
109 28. The assertion (?!) is optimized to (*FAIL). This was not handled correctly
110 when this assertion was used as a condition, for example (?(?!)a|b). In
111 pcre2_match() it worked by luck; in pcre2_dfa_match() it gave an incorrect
112 error about an unsupported item.
113
114 29. For some types of pattern, for example /Z*(|d*){216}/, the auto-
115 possessification code could take exponential time to complete. A recursion
116 depth limit of 1000 has been imposed to limit the resources used by this
117 optimization.
118
119 30. A pattern such as /(*UTF)[\S\V\H]/, which contains a negated special class
120 such as \S in non-UCP mode, explicit wide characters (> 255) can be ignored
121 because \S ensures they are all in the class. The code for doing this was
122 interacting badly with the code for computing the amount of space needed to
123 compile the pattern, leading to a buffer overflow. This bug was discovered
124 by the LLVM fuzzer.
125
126 31. A pattern such as /((?2)+)((?1))/ which has mutual recursion nested inside
127 other kinds of group caused stack overflow at compile time. This bug was
128 discovered by the LLVM fuzzer.
129
130 32. A pattern such as /(?1)(?#?'){8}(a)/ which had a parenthesized comment
131 between a subroutine call and its quantifier was incorrectly compiled,
132 leading to buffer overflow or other errors. This bug was discovered by the
133 LLVM fuzzer.
134
135 33. The illegal pattern /(?(?<E>.*!.*)?)/ was not being diagnosed as missing an
136 assertion after (?(. The code was failing to check the character after
137 (?(?< for the ! or = that would indicate a lookbehind assertion. This bug
138 was discovered by the LLVM fuzzer.
139
140 34. A pattern such as /X((?2)()*+){2}+/ which has a possessive quantifier with
141 a fixed maximum following a group that contains a subroutine reference was
142 incorrectly compiled and could trigger buffer overflow. This bug was
143 discovered by the LLVM fuzzer.
144
145 35. A mutual recursion within a lookbehind assertion such as (?<=((?2))((?1)))
146 caused a stack overflow instead of the diagnosis of a non-fixed length
147 lookbehind assertion. This bug was discovered by the LLVM fuzzer.
148
149
150 Version 8.36 26-September-2014
151 ------------------------------
152
153 1. Got rid of some compiler warnings in the C++ modules that were shown up by
154 -Wmissing-field-initializers and -Wunused-parameter.
155
156 2. The tests for quantifiers being too big (greater than 65535) were being
157 applied after reading the number, and stupidly assuming that integer
158 overflow would give a negative number. The tests are now applied as the
159 numbers are read.
160
161 3. Tidy code in pcre_exec.c where two branches that used to be different are
162 now the same.
163
164 4. The JIT compiler did not generate match limit checks for certain
165 bracketed expressions with quantifiers. This may lead to exponential
166 backtracking, instead of returning with PCRE_ERROR_MATCHLIMIT. This
167 issue should be resolved now.
168
169 5. Fixed an issue, which occures when nested alternatives are optimized
170 with table jumps.
171
172 6. Inserted two casts and changed some ints to size_t in the light of some
173 reported 64-bit compiler warnings (Bugzilla 1477).
174
175 7. Fixed a bug concerned with zero-minimum possessive groups that could match
176 an empty string, which sometimes were behaving incorrectly in the
177 interpreter (though correctly in the JIT matcher). This pcretest input is
178 an example:
179
180 '\A(?:[^"]++|"(?:[^"]*+|"")*+")++'
181 NON QUOTED "QUOT""ED" AFTER "NOT MATCHED
182
183 the interpreter was reporting a match of 'NON QUOTED ' only, whereas the
184 JIT matcher and Perl both matched 'NON QUOTED "QUOT""ED" AFTER '. The test
185 for an empty string was breaking the inner loop and carrying on at a lower
186 level, when possessive repeated groups should always return to a higher
187 level as they have no backtrack points in them. The empty string test now
188 occurs at the outer level.
189
190 8. Fixed a bug that was incorrectly auto-possessifying \w+ in the pattern
191 ^\w+(?>\s*)(?<=\w) which caused it not to match "test test".
192
193 9. Give a compile-time error for \o{} (as Perl does) and for \x{} (which Perl
194 doesn't).
195
196 10. Change 8.34/15 introduced a bug that caused the amount of memory needed
197 to hold a pattern to be incorrectly computed (too small) when there were
198 named back references to duplicated names. This could cause "internal
199 error: code overflow" or "double free or corruption" or other memory
200 handling errors.
201
202 11. When named subpatterns had the same prefixes, back references could be
203 confused. For example, in this pattern:
204
205 /(?P<Name>a)?(?P<Name2>b)?(?(<Name>)c|d)*l/
206
207 the reference to 'Name' was incorrectly treated as a reference to a
208 duplicate name.
209
210 12. A pattern such as /^s?c/mi8 where the optional character has more than
211 one "other case" was incorrectly compiled such that it would only try to
212 match starting at "c".
213
214 13. When a pattern starting with \s was studied, VT was not included in the
215 list of possible starting characters; this should have been part of the
216 8.34/18 patch.
217
218 14. If a character class started [\Qx]... where x is any character, the class
219 was incorrectly terminated at the ].
220
221 15. If a pattern that started with a caseless match for a character with more
222 than one "other case" was studied, PCRE did not set up the starting code
223 unit bit map for the list of possible characters. Now it does. This is an
224 optimization improvement, not a bug fix.
225
226 16. The Unicode data tables have been updated to Unicode 7.0.0.
227
228 17. Fixed a number of memory leaks in pcregrep.
229
230 18. Avoid a compiler warning (from some compilers) for a function call with
231 a cast that removes "const" from an lvalue by using an intermediate
232 variable (to which the compiler does not object).
233
234 19. Incorrect code was compiled if a group that contained an internal recursive
235 back reference was optional (had quantifier with a minimum of zero). This
236 example compiled incorrect code: /(((a\2)|(a*)\g<-1>))*/ and other examples
237 caused segmentation faults because of stack overflows at compile time.
238
239 20. A pattern such as /((?(R)a|(?1)))+/, which contains a recursion within a
240 group that is quantified with an indefinite repeat, caused a compile-time
241 loop which used up all the system stack and provoked a segmentation fault.
242 This was not the same bug as 19 above.
243
244 21. Add PCRECPP_EXP_DECL declaration to operator<< in pcre_stringpiece.h.
245 Patch by Mike Frysinger.
246
247
248 Version 8.35 04-April-2014
249 --------------------------
250
251 1. A new flag is set, when property checks are present in an XCLASS.
252 When this flag is not set, PCRE can perform certain optimizations
253 such as studying these XCLASS-es.
254
255 2. The auto-possessification of character sets were improved: a normal
256 and an extended character set can be compared now. Furthermore
257 the JIT compiler optimizes more character set checks.
258
259 3. Got rid of some compiler warnings for potentially uninitialized variables
260 that show up only when compiled with -O2.
261
262 4. A pattern such as (?=ab\K) that uses \K in an assertion can set the start
263 of a match later then the end of the match. The pcretest program was not
264 handling the case sensibly - it was outputting from the start to the next
265 binary zero. It now reports this situation in a message, and outputs the
266 text from the end to the start.
267
268 5. Fast forward search is improved in JIT. Instead of the first three
269 characters, any three characters with fixed position can be searched.
270 Search order: first, last, middle.
271
272 6. Improve character range checks in JIT. Characters are read by an inprecise
273 function now, which returns with an unknown value if the character code is
274 above a certain threshold (e.g: 256). The only limitation is that the value
275 must be bigger than the threshold as well. This function is useful when
276 the characters above the threshold are handled in the same way.
277
278 7. The macros whose names start with RAWUCHAR are placeholders for a future
279 mode in which only the bottom 21 bits of 32-bit data items are used. To
280 make this more memorable for those maintaining the code, the names have
281 been changed to start with UCHAR21, and an extensive comment has been added
282 to their definition.
283
284 8. Add missing (new) files sljitNativeTILEGX.c and sljitNativeTILEGX-encoder.c
285 to the export list in Makefile.am (they were accidentally omitted from the
286 8.34 tarball).
287
288 9. The informational output from pcretest used the phrase "starting byte set"
289 which is inappropriate for the 16-bit and 32-bit libraries. As the output
290 for "first char" and "need char" really means "non-UTF-char", I've changed
291 "byte" to "char", and slightly reworded the output. The documentation about
292 these values has also been (I hope) clarified.
293
294 10. Another JIT related optimization: use table jumps for selecting the correct
295 backtracking path, when more than four alternatives are present inside a
296 bracket.
297
298 11. Empty match is not possible, when the minimum length is greater than zero,
299 and there is no \K in the pattern. JIT should avoid empty match checks in
300 such cases.
301
302 12. In a caseless character class with UCP support, when a character with more
303 than one alternative case was not the first character of a range, not all
304 the alternative cases were added to the class. For example, s and \x{17f}
305 are both alternative cases for S: the class [RST] was handled correctly,
306 but [R-T] was not.
307
308 13. The configure.ac file always checked for pthread support when JIT was
309 enabled. This is not used in Windows, so I have put this test inside a
310 check for the presence of windows.h (which was already tested for).
311
312 14. Improve pattern prefix search by a simplified Boyer-Moore algorithm in JIT.
313 The algorithm provides a way to skip certain starting offsets, and usually
314 faster than linear prefix searches.
315
316 15. Change 13 for 8.20 updated RunTest to check for the 'fr' locale as well
317 as for 'fr_FR' and 'french'. For some reason, however, it then used the
318 Windows-specific input and output files, which have 'french' screwed in.
319 So this could never have worked. One of the problems with locales is that
320 they aren't always the same. I have now updated RunTest so that it checks
321 the output of the locale test (test 3) against three different output
322 files, and it allows the test to pass if any one of them matches. With luck
323 this should make the test pass on some versions of Solaris where it was
324 failing. Because of the uncertainty, the script did not used to stop if
325 test 3 failed; it now does. If further versions of a French locale ever
326 come to light, they can now easily be added.
327
328 16. If --with-pcregrep-bufsize was given a non-integer value such as "50K",
329 there was a message during ./configure, but it did not stop. This now
330 provokes an error. The invalid example in README has been corrected.
331 If a value less than the minimum is given, the minimum value has always
332 been used, but now a warning is given.
333
334 17. If --enable-bsr-anycrlf was set, the special 16/32-bit test failed. This
335 was a bug in the test system, which is now fixed. Also, the list of various
336 configurations that are tested for each release did not have one with both
337 16/32 bits and --enable-bar-anycrlf. It now does.
338
339 18. pcretest was missing "-C bsr" for displaying the \R default setting.
340
341 19. Little endian PowerPC systems are supported now by the JIT compiler.
342
343 20. The fast forward newline mechanism could enter to an infinite loop on
344 certain invalid UTF-8 input. Although we don't support these cases
345 this issue can be fixed by a performance optimization.
346
347 21. Change 33 of 8.34 is not sufficient to ensure stack safety because it does
348 not take account if existing stack usage. There is now a new global
349 variable called pcre_stack_guard that can be set to point to an external
350 function to check stack availability. It is called at the start of
351 processing every parenthesized group.
352
353 22. A typo in the code meant that in ungreedy mode the max/min qualifier
354 behaved like a min-possessive qualifier, and, for example, /a{1,3}b/U did
355 not match "ab".
356
357 23. When UTF was disabled, the JIT program reported some incorrect compile
358 errors. These messages are silenced now.
359
360 24. Experimental support for ARM-64 and MIPS-64 has been added to the JIT
361 compiler.
362
363 25. Change all the temporary files used in RunGrepTest to be different to those
364 used by RunTest so that the tests can be run simultaneously, for example by
365 "make -j check".
366
367
368 Version 8.34 15-December-2013
369 -----------------------------
370
371 1. Add pcre[16|32]_jit_free_unused_memory to forcibly free unused JIT
372 executable memory. Patch inspired by Carsten Klein.
373
374 2. ./configure --enable-coverage defined SUPPORT_GCOV in config.h, although
375 this macro is never tested and has no effect, because the work to support
376 coverage involves only compiling and linking options and special targets in
377 the Makefile. The comment in config.h implied that defining the macro would
378 enable coverage support, which is totally false. There was also support for
379 setting this macro in the CMake files (my fault, I just copied it from
380 configure). SUPPORT_GCOV has now been removed.
381
382 3. Make a small performance improvement in strlen16() and strlen32() in
383 pcretest.
384
385 4. Change 36 for 8.33 left some unreachable statements in pcre_exec.c,
386 detected by the Solaris compiler (gcc doesn't seem to be able to diagnose
387 these cases). There was also one in pcretest.c.
388
389 5. Cleaned up a "may be uninitialized" compiler warning in pcre_exec.c.
390
391 6. In UTF mode, the code for checking whether a group could match an empty
392 string (which is used for indefinitely repeated groups to allow for
393 breaking an infinite loop) was broken when the group contained a repeated
394 negated single-character class with a character that occupied more than one
395 data item and had a minimum repetition of zero (for example, [^\x{100}]* in
396 UTF-8 mode). The effect was undefined: the group might or might not be
397 deemed as matching an empty string, or the program might have crashed.
398
399 7. The code for checking whether a group could match an empty string was not
400 recognizing that \h, \H, \v, \V, and \R must match a character.
401
402 8. Implemented PCRE_INFO_MATCH_EMPTY, which yields 1 if the pattern can match
403 an empty string. If it can, pcretest shows this in its information output.
404
405 9. Fixed two related bugs that applied to Unicode extended grapheme clusters
406 that were repeated with a maximizing qualifier (e.g. \X* or \X{2,5}) when
407 matched by pcre_exec() without using JIT:
408
409 (a) If the rest of the pattern did not match after a maximal run of
410 grapheme clusters, the code for backing up to try with fewer of them
411 did not always back up over a full grapheme when characters that do not
412 have the modifier quality were involved, e.g. Hangul syllables.
413
414 (b) If the match point in a subject started with modifier character, and
415 there was no match, the code could incorrectly back up beyond the match
416 point, and potentially beyond the first character in the subject,
417 leading to a segfault or an incorrect match result.
418
419 10. A conditional group with an assertion condition could lead to PCRE
420 recording an incorrect first data item for a match if no other first data
421 item was recorded. For example, the pattern (?(?=ab)ab) recorded "a" as a
422 first data item, and therefore matched "ca" after "c" instead of at the
423 start.
424
425 11. Change 40 for 8.33 (allowing pcregrep to find empty strings) showed up a
426 bug that caused the command "echo a | ./pcregrep -M '|a'" to loop.
427
428 12. The source of pcregrep now includes z/OS-specific code so that it can be
429 compiled for z/OS as part of the special z/OS distribution.
430
431 13. Added the -T and -TM options to pcretest.
432
433 14. The code in pcre_compile.c for creating the table of named capturing groups
434 has been refactored. Instead of creating the table dynamically during the
435 actual compiling pass, the information is remembered during the pre-compile
436 pass (on the stack unless there are more than 20 named groups, in which
437 case malloc() is used) and the whole table is created before the actual
438 compile happens. This has simplified the code (it is now nearly 150 lines
439 shorter) and prepared the way for better handling of references to groups
440 with duplicate names.
441
442 15. A back reference to a named subpattern when there is more than one of the
443 same name now checks them in the order in which they appear in the pattern.
444 The first one that is set is used for the reference. Previously only the
445 first one was inspected. This change makes PCRE more compatible with Perl.
446
447 16. Unicode character properties were updated from Unicode 6.3.0.
448
449 17. The compile-time code for auto-possessification has been refactored, based
450 on a patch by Zoltan Herczeg. It now happens after instead of during
451 compilation. The code is cleaner, and more cases are handled. The option
452 PCRE_NO_AUTO_POSSESS is added for testing purposes, and the -O and /O
453 options in pcretest are provided to set it. It can also be set by
454 (*NO_AUTO_POSSESS) at the start of a pattern.
455
456 18. The character VT has been added to the default ("C" locale) set of
457 characters that match \s and are generally treated as white space,
458 following this same change in Perl 5.18. There is now no difference between
459 "Perl space" and "POSIX space". Whether VT is treated as white space in
460 other locales depends on the locale.
461
462 19. The code for checking named groups as conditions, either for being set or
463 for being recursed, has been refactored (this is related to 14 and 15
464 above). Processing unduplicated named groups should now be as fast at
465 numerical groups, and processing duplicated groups should be faster than
466 before.
467
468 20. Two patches to the CMake build system, by Alexander Barkov:
469
470 (1) Replace the "source" command by "." in CMakeLists.txt because
471 "source" is a bash-ism.
472
473 (2) Add missing HAVE_STDINT_H and HAVE_INTTYPES_H to config-cmake.h.in;
474 without these the CMake build does not work on Solaris.
475
476 21. Perl has changed its handling of \8 and \9. If there is no previously
477 encountered capturing group of those numbers, they are treated as the
478 literal characters 8 and 9 instead of a binary zero followed by the
479 literals. PCRE now does the same.
480
481 22. Following Perl, added \o{} to specify codepoints in octal, making it
482 possible to specify values greater than 0777 and also making them
483 unambiguous.
484
485 23. Perl now gives an error for missing closing braces after \x{... instead of
486 treating the string as literal. PCRE now does the same.
487
488 24. RunTest used to grumble if an inappropriate test was selected explicitly,
489 but just skip it when running all tests. This make it awkward to run ranges
490 of tests when one of them was inappropriate. Now it just skips any
491 inappropriate tests, as it always did when running all tests.
492
493 25. If PCRE_AUTO_CALLOUT and PCRE_UCP were set for a pattern that contained
494 character types such as \d or \w, too many callouts were inserted, and the
495 data that they returned was rubbish.
496
497 26. In UCP mode, \s was not matching two of the characters that Perl matches,
498 namely NEL (U+0085) and MONGOLIAN VOWEL SEPARATOR (U+180E), though they
499 were matched by \h. The code has now been refactored so that the lists of
500 the horizontal and vertical whitespace characters used for \h and \v (which
501 are defined only in one place) are now also used for \s.
502
503 27. Add JIT support for the 64 bit TileGX architecture.
504 Patch by Jiong Wang (Tilera Corporation).
505
506 28. Possessive quantifiers for classes (both explicit and automatically
507 generated) now use special opcodes instead of wrapping in ONCE brackets.
508
509 29. Whereas an item such as A{4}+ ignored the possessivenes of the quantifier
510 (because it's meaningless), this was not happening when PCRE_CASELESS was
511 set. Not wrong, but inefficient.
512
513 30. Updated perltest.pl to add /u (force Unicode mode) when /W (use Unicode
514 properties for \w, \d, etc) is present in a test regex. Otherwise if the
515 test contains no characters greater than 255, Perl doesn't realise it
516 should be using Unicode semantics.
517
518 31. Upgraded the handling of the POSIX classes [:graph:], [:print:], and
519 [:punct:] when PCRE_UCP is set so as to include the same characters as Perl
520 does in Unicode mode.
521
522 32. Added the "forbid" facility to pcretest so that putting tests into the
523 wrong test files can sometimes be quickly detected.
524
525 33. There is now a limit (default 250) on the depth of nesting of parentheses.
526 This limit is imposed to control the amount of system stack used at compile
527 time. It can be changed at build time by --with-parens-nest-limit=xxx or
528 the equivalent in CMake.
529
530 34. Character classes such as [A-\d] or [a-[:digit:]] now cause compile-time
531 errors. Perl warns for these when in warning mode, but PCRE has no facility
532 for giving warnings.
533
534 35. Change 34 for 8.13 allowed quantifiers on assertions, because Perl does.
535 However, this was not working for (?!) because it is optimized to (*FAIL),
536 for which PCRE does not allow quantifiers. The optimization is now disabled
537 when a quantifier follows (?!). I can't see any use for this, but it makes
538 things uniform.
539
540 36. Perl no longer allows group names to start with digits, so I have made this
541 change also in PCRE. It simplifies the code a bit.
542
543 37. In extended mode, Perl ignores spaces before a + that indicates a
544 possessive quantifier. PCRE allowed a space before the quantifier, but not
545 before the possessive +. It now does.
546
547 38. The use of \K (reset reported match start) within a repeated possessive
548 group such as (a\Kb)*+ was not working.
549
550 40. Document that the same character tables must be used at compile time and
551 run time, and that the facility to pass tables to pcre_exec() and
552 pcre_dfa_exec() is for use only with saved/restored patterns.
553
554 41. Applied Jeff Trawick's patch CMakeLists.txt, which "provides two new
555 features for Builds with MSVC:
556
557 1. Support pcre.rc and/or pcreposix.rc (as is already done for MinGW
558 builds). The .rc files can be used to set FileDescription and many other
559 attributes.
560
561 2. Add an option (-DINSTALL_MSVC_PDB) to enable installation of .pdb files.
562 This allows higher-level build scripts which want .pdb files to avoid
563 hard-coding the exact files needed."
564
565 42. Added support for [[:<:]] and [[:>:]] as used in the BSD POSIX library to
566 mean "start of word" and "end of word", respectively, as a transition aid.
567
568 43. A minimizing repeat of a class containing codepoints greater than 255 in
569 non-UTF 16-bit or 32-bit modes caused an internal error when PCRE was
570 compiled to use the heap for recursion.
571
572 44. Got rid of some compiler warnings for unused variables when UTF but not UCP
573 is configured.
574
575
576 Version 8.33 28-May-2013
577 ------------------------
578
579 1. Added 'U' to some constants that are compared to unsigned integers, to
580 avoid compiler signed/unsigned warnings. Added (int) casts to unsigned
581 variables that are added to signed variables, to ensure the result is
582 signed and can be negated.
583
584 2. Applied patch by Daniel Richard G for quashing MSVC warnings to the
585 CMake config files.
586
587 3. Revise the creation of config.h.generic so that all boolean macros are
588 #undefined, whereas non-boolean macros are #ifndef/#endif-ed. This makes
589 overriding via -D on the command line possible.
590
591 4. Changing the definition of the variable "op" in pcre_exec.c from pcre_uchar
592 to unsigned int is reported to make a quite noticeable speed difference in
593 a specific Windows environment. Testing on Linux did also appear to show
594 some benefit (and it is clearly not harmful). Also fixed the definition of
595 Xop which should be unsigned.
596
597 5. Related to (4), changing the definition of the intermediate variable cc
598 in repeated character loops from pcre_uchar to pcre_uint32 also gave speed
599 improvements.
600
601 6. Fix forward search in JIT when link size is 3 or greater. Also removed some
602 unnecessary spaces.
603
604 7. Adjust autogen.sh and configure.ac to lose warnings given by automake 1.12
605 and later.
606
607 8. Fix two buffer over read issues in 16 and 32 bit modes. Affects JIT only.
608
609 9. Optimizing fast_forward_start_bits in JIT.
610
611 10. Adding support for callouts in JIT, and fixing some issues revealed
612 during this work. Namely:
613
614 (a) Unoptimized capturing brackets incorrectly reset on backtrack.
615
616 (b) Minimum length was not checked before the matching is started.
617
618 11. The value of capture_last that is passed to callouts was incorrect in some
619 cases when there was a capture on one path that was subsequently abandoned
620 after a backtrack. Also, the capture_last value is now reset after a
621 recursion, since all captures are also reset in this case.
622
623 12. The interpreter no longer returns the "too many substrings" error in the
624 case when an overflowing capture is in a branch that is subsequently
625 abandoned after a backtrack.
626
627 13. In the pathological case when an offset vector of size 2 is used, pcretest
628 now prints out the matched string after a yield of 0 or 1.
629
630 14. Inlining subpatterns in recursions, when certain conditions are fulfilled.
631 Only supported by the JIT compiler at the moment.
632
633 15. JIT compiler now supports 32 bit Macs thanks to Lawrence Velazquez.
634
635 16. Partial matches now set offsets[2] to the "bumpalong" value, that is, the
636 offset of the starting point of the matching process, provided the offsets
637 vector is large enough.
638
639 17. The \A escape now records a lookbehind value of 1, though its execution
640 does not actually inspect the previous character. This is to ensure that,
641 in partial multi-segment matching, at least one character from the old
642 segment is retained when a new segment is processed. Otherwise, if there
643 are no lookbehinds in the pattern, \A might match incorrectly at the start
644 of a new segment.
645
646 18. Added some #ifdef __VMS code into pcretest.c to help VMS implementations.
647
648 19. Redefined some pcre_uchar variables in pcre_exec.c as pcre_uint32; this
649 gives some modest performance improvement in 8-bit mode.
650
651 20. Added the PCRE-specific property \p{Xuc} for matching characters that can
652 be expressed in certain programming languages using Universal Character
653 Names.
654
655 21. Unicode validation has been updated in the light of Unicode Corrigendum #9,
656 which points out that "non characters" are not "characters that may not
657 appear in Unicode strings" but rather "characters that are reserved for
658 internal use and have only local meaning".
659
660 22. When a pattern was compiled with automatic callouts (PCRE_AUTO_CALLOUT) and
661 there was a conditional group that depended on an assertion, if the
662 assertion was false, the callout that immediately followed the alternation
663 in the condition was skipped when pcre_exec() was used for matching.
664
665 23. Allow an explicit callout to be inserted before an assertion that is the
666 condition for a conditional group, for compatibility with automatic
667 callouts, which always insert a callout at this point.
668
669 24. In 8.31, (*COMMIT) was confined to within a recursive subpattern. Perl also
670 confines (*SKIP) and (*PRUNE) in the same way, and this has now been done.
671
672 25. (*PRUNE) is now supported by the JIT compiler.
673
674 26. Fix infinite loop when /(?<=(*SKIP)ac)a/ is matched against aa.
675
676 27. Fix the case where there are two or more SKIPs with arguments that may be
677 ignored.
678
679 28. (*SKIP) is now supported by the JIT compiler.
680
681 29. (*THEN) is now supported by the JIT compiler.
682
683 30. Update RunTest with additional test selector options.
684
685 31. The way PCRE handles backtracking verbs has been changed in two ways.
686
687 (1) Previously, in something like (*COMMIT)(*SKIP), COMMIT would override
688 SKIP. Now, PCRE acts on whichever backtracking verb is reached first by
689 backtracking. In some cases this makes it more Perl-compatible, but Perl's
690 rather obscure rules do not always do the same thing.
691
692 (2) Previously, backtracking verbs were confined within assertions. This is
693 no longer the case for positive assertions, except for (*ACCEPT). Again,
694 this sometimes improves Perl compatibility, and sometimes does not.
695
696 32. A number of tests that were in test 2 because Perl did things differently
697 have been moved to test 1, because either Perl or PCRE has changed, and
698 these tests are now compatible.
699
700 32. Backtracking control verbs are now handled in the same way in JIT and
701 interpreter.
702
703 33. An opening parenthesis in a MARK/PRUNE/SKIP/THEN name in a pattern that
704 contained a forward subroutine reference caused a compile error.
705
706 34. Auto-detect and optimize limited repetitions in JIT.
707
708 35. Implement PCRE_NEVER_UTF to lock out the use of UTF, in particular,
709 blocking (*UTF) etc.
710
711 36. In the interpreter, maximizing pattern repetitions for characters and
712 character types now use tail recursion, which reduces stack usage.
713
714 37. The value of the max lookbehind was not correctly preserved if a compiled
715 and saved regex was reloaded on a host of different endianness.
716
717 38. Implemented (*LIMIT_MATCH) and (*LIMIT_RECURSION). As part of the extension
718 of the compiled pattern block, expand the flags field from 16 to 32 bits
719 because it was almost full.
720
721 39. Try madvise first before posix_madvise.
722
723 40. Change 7 for PCRE 7.9 made it impossible for pcregrep to find empty lines
724 with a pattern such as ^$. It has taken 4 years for anybody to notice! The
725 original change locked out all matches of empty strings. This has been
726 changed so that one match of an empty string per line is recognized.
727 Subsequent searches on the same line (for colouring or for --only-matching,
728 for example) do not recognize empty strings.
729
730 41. Applied a user patch to fix a number of spelling mistakes in comments.
731
732 42. Data lines longer than 65536 caused pcretest to crash.
733
734 43. Clarified the data type for length and startoffset arguments for pcre_exec
735 and pcre_dfa_exec in the function-specific man pages, where they were
736 explicitly stated to be in bytes, never having been updated. I also added
737 some clarification to the pcreapi man page.
738
739 44. A call to pcre_dfa_exec() with an output vector size less than 2 caused
740 a segmentation fault.
741
742
743 Version 8.32 30-November-2012
744 -----------------------------
745
746 1. Improved JIT compiler optimizations for first character search and single
747 character iterators.
748
749 2. Supporting IBM XL C compilers for PPC architectures in the JIT compiler.
750 Patch by Daniel Richard G.
751
752 3. Single character iterator optimizations in the JIT compiler.
753
754 4. Improved JIT compiler optimizations for character ranges.
755
756 5. Rename the "leave" variable names to "quit" to improve WinCE compatibility.
757 Reported by Giuseppe D'Angelo.
758
759 6. The PCRE_STARTLINE bit, indicating that a match can occur only at the start
760 of a line, was being set incorrectly in cases where .* appeared inside
761 atomic brackets at the start of a pattern, or where there was a subsequent
762 *PRUNE or *SKIP.
763
764 7. Improved instruction cache flush for POWER/PowerPC.
765 Patch by Daniel Richard G.
766
767 8. Fixed a number of issues in pcregrep, making it more compatible with GNU
768 grep:
769
770 (a) There is now no limit to the number of patterns to be matched.
771
772 (b) An error is given if a pattern is too long.
773
774 (c) Multiple uses of --exclude, --exclude-dir, --include, and --include-dir
775 are now supported.
776
777 (d) --exclude-from and --include-from (multiple use) have been added.
778
779 (e) Exclusions and inclusions now apply to all files and directories, not
780 just to those obtained from scanning a directory recursively.
781
782 (f) Multiple uses of -f and --file-list are now supported.
783
784 (g) In a Windows environment, the default for -d has been changed from
785 "read" (the GNU grep default) to "skip", because otherwise the presence
786 of a directory in the file list provokes an error.
787
788 (h) The documentation has been revised and clarified in places.
789
790 9. Improve the matching speed of capturing brackets.
791
792 10. Changed the meaning of \X so that it now matches a Unicode extended
793 grapheme cluster.
794
795 11. Patch by Daniel Richard G to the autoconf files to add a macro for sorting
796 out POSIX threads when JIT support is configured.
797
798 12. Added support for PCRE_STUDY_EXTRA_NEEDED.
799
800 13. In the POSIX wrapper regcomp() function, setting re_nsub field in the preg
801 structure could go wrong in environments where size_t is not the same size
802 as int.
803
804 14. Applied user-supplied patch to pcrecpp.cc to allow PCRE_NO_UTF8_CHECK to be
805 set.
806
807 15. The EBCDIC support had decayed; later updates to the code had included
808 explicit references to (e.g.) \x0a instead of CHAR_LF. There has been a
809 general tidy up of EBCDIC-related issues, and the documentation was also
810 not quite right. There is now a test that can be run on ASCII systems to
811 check some of the EBCDIC-related things (but is it not a full test).
812
813 16. The new PCRE_STUDY_EXTRA_NEEDED option is now used by pcregrep, resulting
814 in a small tidy to the code.
815
816 17. Fix JIT tests when UTF is disabled and both 8 and 16 bit mode are enabled.
817
818 18. If the --only-matching (-o) option in pcregrep is specified multiple
819 times, each one causes appropriate output. For example, -o1 -o2 outputs the
820 substrings matched by the 1st and 2nd capturing parentheses. A separating
821 string can be specified by --om-separator (default empty).
822
823 19. Improving the first n character searches.
824
825 20. Turn case lists for horizontal and vertical white space into macros so that
826 they are defined only once.
827
828 21. This set of changes together give more compatible Unicode case-folding
829 behaviour for characters that have more than one other case when UCP
830 support is available.
831
832 (a) The Unicode property table now has offsets into a new table of sets of
833 three or more characters that are case-equivalent. The MultiStage2.py
834 script that generates these tables (the pcre_ucd.c file) now scans
835 CaseFolding.txt instead of UnicodeData.txt for character case
836 information.
837
838 (b) The code for adding characters or ranges of characters to a character
839 class has been abstracted into a generalized function that also handles
840 case-independence. In UTF-mode with UCP support, this uses the new data
841 to handle characters with more than one other case.
842
843 (c) A bug that is fixed as a result of (b) is that codepoints less than 256
844 whose other case is greater than 256 are now correctly matched
845 caselessly. Previously, the high codepoint matched the low one, but not
846 vice versa.
847
848 (d) The processing of \h, \H, \v, and \ in character classes now makes use
849 of the new class addition function, using character lists defined as
850 macros alongside the case definitions of 20 above.
851
852 (e) Caseless back references now work with characters that have more than
853 one other case.
854
855 (f) General caseless matching of characters with more than one other case
856 is supported.
857
858 22. Unicode character properties were updated from Unicode 6.2.0
859
860 23. Improved CMake support under Windows. Patch by Daniel Richard G.
861
862 24. Add support for 32-bit character strings, and UTF-32
863
864 25. Major JIT compiler update (code refactoring and bugfixing).
865 Experimental Sparc 32 support is added.
866
867 26. Applied a modified version of Daniel Richard G's patch to create
868 pcre.h.generic and config.h.generic by "make" instead of in the
869 PrepareRelease script.
870
871 27. Added a definition for CHAR_NULL (helpful for the z/OS port), and use it in
872 pcre_compile.c when checking for a zero character.
873
874 28. Introducing a native interface for JIT. Through this interface, the compiled
875 machine code can be directly executed. The purpose of this interface is to
876 provide fast pattern matching, so several sanity checks are not performed.
877 However, feature tests are still performed. The new interface provides
878 1.4x speedup compared to the old one.
879
880 29. If pcre_exec() or pcre_dfa_exec() was called with a negative value for
881 the subject string length, the error given was PCRE_ERROR_BADOFFSET, which
882 was confusing. There is now a new error PCRE_ERROR_BADLENGTH for this case.
883
884 30. In 8-bit UTF-8 mode, pcretest failed to give an error for data codepoints
885 greater than 0x7fffffff (which cannot be represented in UTF-8, even under
886 the "old" RFC 2279). Instead, it ended up passing a negative length to
887 pcre_exec().
888
889 31. Add support for GCC's visibility feature to hide internal functions.
890
891 32. Running "pcretest -C pcre8" or "pcretest -C pcre16" gave a spurious error
892 "unknown -C option" after outputting 0 or 1.
893
894 33. There is now support for generating a code coverage report for the test
895 suite in environments where gcc is the compiler and lcov is installed. This
896 is mainly for the benefit of the developers.
897
898 34. If PCRE is built with --enable-valgrind, certain memory regions are marked
899 unaddressable using valgrind annotations, allowing valgrind to detect
900 invalid memory accesses. This is mainly for the benefit of the developers.
901
902 25. (*UTF) can now be used to start a pattern in any of the three libraries.
903
904 26. Give configure error if --enable-cpp but no C++ compiler found.
905
906
907 Version 8.31 06-July-2012
908 -------------------------
909
910 1. Fixing a wrong JIT test case and some compiler warnings.
911
912 2. Removed a bashism from the RunTest script.
913
914 3. Add a cast to pcre_exec.c to fix the warning "unary minus operator applied
915 to unsigned type, result still unsigned" that was given by an MS compiler
916 on encountering the code "-sizeof(xxx)".
917
918 4. Partial matching support is added to the JIT compiler.
919
920 5. Fixed several bugs concerned with partial matching of items that consist
921 of more than one character:
922
923 (a) /^(..)\1/ did not partially match "aba" because checking references was
924 done on an "all or nothing" basis. This also applied to repeated
925 references.
926
927 (b) \R did not give a hard partial match if \r was found at the end of the
928 subject.
929
930 (c) \X did not give a hard partial match after matching one or more
931 characters at the end of the subject.
932
933 (d) When newline was set to CRLF, a pattern such as /a$/ did not recognize
934 a partial match for the string "\r".
935
936 (e) When newline was set to CRLF, the metacharacter "." did not recognize
937 a partial match for a CR character at the end of the subject string.
938
939 6. If JIT is requested using /S++ or -s++ (instead of just /S+ or -s+) when
940 running pcretest, the text "(JIT)" added to the output whenever JIT is
941 actually used to run the match.
942
943 7. Individual JIT compile options can be set in pcretest by following -s+[+]
944 or /S+[+] with a digit between 1 and 7.
945
946 8. OP_NOT now supports any UTF character not just single-byte ones.
947
948 9. (*MARK) control verb is now supported by the JIT compiler.
949
950 10. The command "./RunTest list" lists the available tests without actually
951 running any of them. (Because I keep forgetting what they all are.)
952
953 11. Add PCRE_INFO_MAXLOOKBEHIND.
954
955 12. Applied a (slightly modified) user-supplied patch that improves performance
956 when the heap is used for recursion (compiled with --disable-stack-for-
957 recursion). Instead of malloc and free for each heap frame each time a
958 logical recursion happens, frames are retained on a chain and re-used where
959 possible. This sometimes gives as much as 30% improvement.
960
961 13. As documented, (*COMMIT) is now confined to within a recursive subpattern
962 call.
963
964 14. As documented, (*COMMIT) is now confined to within a positive assertion.
965
966 15. It is now possible to link pcretest with libedit as an alternative to
967 libreadline.
968
969 16. (*COMMIT) control verb is now supported by the JIT compiler.
970
971 17. The Unicode data tables have been updated to Unicode 6.1.0.
972
973 18. Added --file-list option to pcregrep.
974
975 19. Added binary file support to pcregrep, including the -a, --binary-files,
976 -I, and --text options.
977
978 20. The madvise function is renamed for posix_madvise for QNX compatibility
979 reasons. Fixed by Giuseppe D'Angelo.
980
981 21. Fixed a bug for backward assertions with REVERSE 0 in the JIT compiler.
982
983 22. Changed the option for creating symbolic links for 16-bit man pages from
984 -s to -sf so that re-installing does not cause issues.
985
986 23. Support PCRE_NO_START_OPTIMIZE in JIT as (*MARK) support requires it.
987
988 24. Fixed a very old bug in pcretest that caused errors with restarted DFA
989 matches in certain environments (the workspace was not being correctly
990 retained). Also added to pcre_dfa_exec() a simple plausibility check on
991 some of the workspace data at the beginning of a restart.
992
993 25. \s*\R was auto-possessifying the \s* when it should not, whereas \S*\R
994 was not doing so when it should - probably a typo introduced by SVN 528
995 (change 8.10/14).
996
997 26. When PCRE_UCP was not set, \w+\x{c4} was incorrectly auto-possessifying the
998 \w+ when the character tables indicated that \x{c4} was a word character.
999 There were several related cases, all because the tests for doing a table
1000 lookup were testing for characters less than 127 instead of 255.
1001
1002 27. If a pattern contains capturing parentheses that are not used in a match,
1003 their slots in the ovector are set to -1. For those that are higher than
1004 any matched groups, this happens at the end of processing. In the case when
1005 there were back references that the ovector was too small to contain
1006 (causing temporary malloc'd memory to be used during matching), and the
1007 highest capturing number was not used, memory off the end of the ovector
1008 was incorrectly being set to -1. (It was using the size of the temporary
1009 memory instead of the true size.)
1010
1011 28. To catch bugs like 27 using valgrind, when pcretest is asked to specify an
1012 ovector size, it uses memory at the end of the block that it has got.
1013
1014 29. Check for an overlong MARK name and give an error at compile time. The
1015 limit is 255 for the 8-bit library and 65535 for the 16-bit library.
1016
1017 30. JIT compiler update.
1018
1019 31. JIT is now supported on jailbroken iOS devices. Thanks for Ruiger
1020 Rill for the patch.
1021
1022 32. Put spaces around SLJIT_PRINT_D in the JIT compiler. Required by CXX11.
1023
1024 33. Variable renamings in the PCRE-JIT compiler. No functionality change.
1025
1026 34. Fixed typos in pcregrep: in two places there was SUPPORT_LIBZ2 instead of
1027 SUPPORT_LIBBZ2. This caused a build problem when bzip2 but not gzip (zlib)
1028 was enabled.
1029
1030 35. Improve JIT code generation for greedy plus quantifier.
1031
1032 36. When /((?:a?)*)*c/ or /((?>a?)*)*c/ was matched against "aac", it set group
1033 1 to "aa" instead of to an empty string. The bug affected repeated groups
1034 that could potentially match an empty string.
1035
1036 37. Optimizing single character iterators in JIT.
1037
1038 38. Wide characters specified with \uxxxx in JavaScript mode are now subject to
1039 the same checks as \x{...} characters in non-JavaScript mode. Specifically,
1040 codepoints that are too big for the mode are faulted, and in a UTF mode,
1041 disallowed codepoints are also faulted.
1042
1043 39. If PCRE was compiled with UTF support, in three places in the DFA
1044 matcher there was code that should only have been obeyed in UTF mode, but
1045 was being obeyed unconditionally. In 8-bit mode this could cause incorrect
1046 processing when bytes with values greater than 127 were present. In 16-bit
1047 mode the bug would be provoked by values in the range 0xfc00 to 0xdc00. In
1048 both cases the values are those that cannot be the first data item in a UTF
1049 character. The three items that might have provoked this were recursions,
1050 possessively repeated groups, and atomic groups.
1051
1052 40. Ensure that libpcre is explicitly listed in the link commands for pcretest
1053 and pcregrep, because some OS require shared objects to be explicitly
1054 passed to ld, causing the link step to fail if they are not.
1055
1056 41. There were two incorrect #ifdefs in pcre_study.c, meaning that, in 16-bit
1057 mode, patterns that started with \h* or \R* might be incorrectly matched.
1058
1059
1060 Version 8.30 04-February-2012
1061 -----------------------------
1062
1063 1. Renamed "isnumber" as "is_a_number" because in some Mac environments this
1064 name is defined in ctype.h.
1065
1066 2. Fixed a bug in fixed-length calculation for lookbehinds that would show up
1067 only in quite long subpatterns.
1068
1069 3. Removed the function pcre_info(), which has been obsolete and deprecated
1070 since it was replaced by pcre_fullinfo() in February 2000.
1071
1072 4. For a non-anchored pattern, if (*SKIP) was given with a name that did not
1073 match a (*MARK), and the match failed at the start of the subject, a
1074 reference to memory before the start of the subject could occur. This bug
1075 was introduced by fix 17 of release 8.21.
1076
1077 5. A reference to an unset group with zero minimum repetition was giving
1078 totally wrong answers (in non-JavaScript-compatibility mode). For example,
1079 /(another)?(\1?)test/ matched against "hello world test". This bug was
1080 introduced in release 8.13.
1081
1082 6. Add support for 16-bit character strings (a large amount of work involving
1083 many changes and refactorings).
1084
1085 7. RunGrepTest failed on msys because \r\n was replaced by whitespace when the
1086 command "pattern=`printf 'xxx\r\njkl'`" was run. The pattern is now taken
1087 from a file.
1088
1089 8. Ovector size of 2 is also supported by JIT based pcre_exec (the ovector size
1090 rounding is not applied in this particular case).
1091
1092 9. The invalid Unicode surrogate codepoints U+D800 to U+DFFF are now rejected
1093 if they appear, or are escaped, in patterns.
1094
1095 10. Get rid of a number of -Wunused-but-set-variable warnings.
1096
1097 11. The pattern /(?=(*:x))(q|)/ matches an empty string, and returns the mark
1098 "x". The similar pattern /(?=(*:x))((*:y)q|)/ did not return a mark at all.
1099 Oddly, Perl behaves the same way. PCRE has been fixed so that this pattern
1100 also returns the mark "x". This bug applied to capturing parentheses,
1101 non-capturing parentheses, and atomic parentheses. It also applied to some
1102 assertions.
1103
1104 12. Stephen Kelly's patch to CMakeLists.txt allows it to parse the version
1105 information out of configure.ac instead of relying on pcre.h.generic, which
1106 is not stored in the repository.
1107
1108 13. Applied Dmitry V. Levin's patch for a more portable method for linking with
1109 -lreadline.
1110
1111 14. ZH added PCRE_CONFIG_JITTARGET; added its output to pcretest -C.
1112
1113 15. Applied Graycode's patch to put the top-level frame on the stack rather
1114 than the heap when not using the stack for recursion. This gives a
1115 performance improvement in many cases when recursion is not deep.
1116
1117 16. Experimental code added to "pcretest -C" to output the stack frame size.
1118
1119
1120 Version 8.21 12-Dec-2011
1121 ------------------------
1122
1123 1. Updating the JIT compiler.
1124
1125 2. JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
1126 are added as well.
1127
1128 3. Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
1129 PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
1130 calling _pcre_jit_exec. Some extra comments are added.
1131
1132 4. (*MARK) settings inside atomic groups that do not contain any capturing
1133 parentheses, for example, (?>a(*:m)), were not being passed out. This bug
1134 was introduced by change 18 for 8.20.
1135
1136 5. Supporting of \x, \U and \u in JavaScript compatibility mode based on the
1137 ECMA-262 standard.
1138
1139 6. Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
1140 erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
1141 This bug was probably introduced by change 9 of 8.13.
1142
1143 7. While fixing 6 above, I noticed that a number of other items were being
1144 incorrectly rejected as "not fixed length". This arose partly because newer
1145 opcodes had not been added to the fixed-length checking code. I have (a)
1146 corrected the bug and added tests for these items, and (b) arranged for an
1147 error to occur if an unknown opcode is encountered while checking for fixed
1148 length instead of just assuming "not fixed length". The items that were
1149 rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
1150 (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
1151 repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
1152
1153 8. A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
1154 being incorrectly compiled and would have given unpredicatble results.
1155
1156 9. A possessively repeated subpattern with minimum repeat count greater than
1157 one behaved incorrectly. For example, (A){2,}+ behaved as if it was
1158 (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
1159 the first (A) could occur when it should not.
1160
1161 10. Add a cast and remove a redundant test from the code.
1162
1163 11. JIT should use pcre_malloc/pcre_free for allocation.
1164
1165 12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
1166 best practice nowadays, and helps with cross-compiling. (If the exec_prefix
1167 is anything other than /usr, -L is still shown).
1168
1169 13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
1170
1171 14. Perl does not support \N without a following name in a [] class; PCRE now
1172 also gives an error.
1173
1174 15. If a forward reference was repeated with an upper limit of around 2000,
1175 it caused the error "internal error: overran compiling workspace". The
1176 maximum number of forward references (including repeats) was limited by the
1177 internal workspace, and dependent on the LINK_SIZE. The code has been
1178 rewritten so that the workspace expands (via pcre_malloc) if necessary, and
1179 the default depends on LINK_SIZE. There is a new upper limit (for safety)
1180 of around 200,000 forward references. While doing this, I also speeded up
1181 the filling in of repeated forward references.
1182
1183 16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
1184 incorrectly expecting the subject to contain another "a" after the start.
1185
1186 17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
1187 in the match, the SKIP should be ignored. This was not happening; instead
1188 the SKIP was being treated as NOMATCH. For patterns such as
1189 /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
1190 tested.
1191
1192 18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
1193 now much more compatible with Perl, in particular in cases where the result
1194 is a non-match for a non-anchored pattern. For example, if
1195 /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
1196 "m", where previously it did not return a name. A side effect of this
1197 change is that for partial matches, the last encountered mark name is
1198 returned, as for non matches. A number of tests that were previously not
1199 Perl-compatible have been moved into the Perl-compatible test files. The
1200 refactoring has had the pleasing side effect of removing one argument from
1201 the match() function, thus reducing its stack requirements.
1202
1203 19. If the /S+ option was used in pcretest to study a pattern using JIT,
1204 subsequent uses of /S (without +) incorrectly behaved like /S+.
1205
1206 21. Retrieve executable code size support for the JIT compiler and fixing
1207 some warnings.
1208
1209 22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
1210 not work when the shorter character appeared right at the end of the
1211 subject string.
1212
1213 23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
1214 systems.
1215
1216 24. Added PCRE_INFO_JITSIZE to pass on the value from (21) above, and also
1217 output it when the /M option is used in pcretest.
1218
1219 25. The CheckMan script was not being included in the distribution. Also, added
1220 an explicit "perl" to run Perl scripts from the PrepareRelease script
1221 because this is reportedly needed in Windows.
1222
1223 26. If study data was being save in a file and studying had not found a set of
1224 "starts with" bytes for the pattern, the data written to the file (though
1225 never used) was taken from uninitialized memory and so caused valgrind to
1226 complain.
1227
1228 27. Updated RunTest.bat as provided by Sheri Pierce.
1229
1230 28. Fixed a possible uninitialized memory bug in pcre_jit_compile.c.
1231
1232 29. Computation of memory usage for the table of capturing group names was
1233 giving an unnecessarily large value.
1234
1235
1236 Version 8.20 21-Oct-2011
1237 ------------------------
1238
1239 1. Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
1240 a POSIX class. After further experiments with Perl, which convinced me that
1241 Perl has bugs and confusions, a closing square bracket is no longer allowed
1242 in a POSIX name. This bug also affected patterns with classes that started
1243 with full stops.
1244
1245 2. If a pattern such as /(a)b|ac/ is matched against "ac", there is no
1246 captured substring, but while checking the failing first alternative,
1247 substring 1 is temporarily captured. If the output vector supplied to
1248 pcre_exec() was not big enough for this capture, the yield of the function
1249 was still zero ("insufficient space for captured substrings"). This cannot
1250 be totally fixed without adding another stack variable, which seems a lot
1251 of expense for a edge case. However, I have improved the situation in cases
1252 such as /(a)(b)x|abc/ matched against "abc", where the return code
1253 indicates that fewer than the maximum number of slots in the ovector have
1254 been set.
1255
1256 3. Related to (2) above: when there are more back references in a pattern than
1257 slots in the output vector, pcre_exec() uses temporary memory during
1258 matching, and copies in the captures as far as possible afterwards. It was
1259 using the entire output vector, but this conflicts with the specification
1260 that only 2/3 is used for passing back captured substrings. Now it uses
1261 only the first 2/3, for compatibility. This is, of course, another edge
1262 case.
1263
1264 4. Zoltan Herczeg's just-in-time compiler support has been integrated into the
1265 main code base, and can be used by building with --enable-jit. When this is
1266 done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
1267 runtime --no-jit option is given.
1268
1269 5. When the number of matches in a pcre_dfa_exec() run exactly filled the
1270 ovector, the return from the function was zero, implying that there were
1271 other matches that did not fit. The correct "exactly full" value is now
1272 returned.
1273
1274 6. If a subpattern that was called recursively or as a subroutine contained
1275 (*PRUNE) or any other control that caused it to give a non-standard return,
1276 invalid errors such as "Error -26 (nested recursion at the same subject
1277 position)" or even infinite loops could occur.
1278
1279 7. If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
1280 computing the minimum length on reaching *ACCEPT, and so ended up with the
1281 wrong value of 1 rather than 0. Further investigation indicates that
1282 computing a minimum subject length in the presence of *ACCEPT is difficult
1283 (think back references, subroutine calls), and so I have changed the code
1284 so that no minimum is registered for a pattern that contains *ACCEPT.
1285
1286 8. If (*THEN) was present in the first (true) branch of a conditional group,
1287 it was not handled as intended. [But see 16 below.]
1288
1289 9. Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
1290 Sheri Pierce.
1291
1292 10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
1293 the first byte in a match must be "a".
1294
1295 11. Change 17 for 8.13 increased the recursion depth for patterns like
1296 /a(?:.)*?a/ drastically. I've improved things by remembering whether a
1297 pattern contains any instances of (*THEN). If it does not, the old
1298 optimizations are restored. It would be nice to do this on a per-group
1299 basis, but at the moment that is not feasible.
1300
1301 12. In some environments, the output of pcretest -C is CRLF terminated. This
1302 broke RunTest's code that checks for the link size. A single white space
1303 character after the value is now allowed for.
1304
1305 13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
1306 For "fr", it uses the Windows-specific input and output files.
1307
1308 14. If (*THEN) appeared in a group that was called recursively or as a
1309 subroutine, it did not work as intended. [But see next item.]
1310
1311 15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
1312 pattern fragments (but not containing any | characters). If A and B are
1313 matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
1314 was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
1315 D. In other words, Perl considers parentheses that do not contain any |
1316 characters to be part of a surrounding alternative, whereas PCRE was
1317 treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
1318 differently. PCRE now behaves in the same way as Perl, except in the case
1319 of subroutine/recursion calls such as (?1) which have in any case always
1320 been different (but PCRE had them first :-).
1321
1322 16. Related to 15 above: Perl does not treat the | in a conditional group as
1323 creating alternatives. Such a group is treated in the same way as an
1324 ordinary group without any | characters when processing (*THEN). PCRE has
1325 been changed to match Perl's behaviour.
1326
1327 17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
1328 RunGrepTest script failed.
1329
1330 18. Change 22 for version 13 caused atomic groups to use more stack. This is
1331 inevitable for groups that contain captures, but it can lead to a lot of
1332 stack use in large patterns. The old behaviour has been restored for atomic
1333 groups that do not contain any capturing parentheses.
1334
1335 19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
1336 suppress the check for a minimum subject length at run time. (If it was
1337 given to pcre_exec() or pcre_dfa_exec() it did work.)
1338
1339 20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
1340 fail to work when decoding hex characters in data strings in EBCDIC
1341 environments.
1342
1343 21. It appears that in at least one Mac OS environment, the isxdigit() function
1344 is implemented as a macro that evaluates to its argument more than once,
1345 contravening the C 90 Standard (I haven't checked a later standard). There
1346 was an instance in pcretest which caused it to go wrong when processing
1347 \x{...} escapes in subject strings. The has been rewritten to avoid using
1348 things like p++ in the argument of isxdigit().
1349
1350
1351 Version 8.13 16-Aug-2011
1352 ------------------------
1353
1354 1. The Unicode data tables have been updated to Unicode 6.0.0.
1355
1356 2. Two minor typos in pcre_internal.h have been fixed.
1357
1358 3. Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
1359 pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
1360 in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
1361
1362 4. There were a number of related bugs in the code for matching backrefences
1363 caselessly in UTF-8 mode when codes for the characters concerned were
1364 different numbers of bytes. For example, U+023A and U+2C65 are an upper
1365 and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
1366 (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
1367 code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
1368 2-byte code at the end of the subject (it thought there wasn't enough data
1369 left).
1370
1371 5. Comprehensive information about what went wrong is now returned by
1372 pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
1373 as the output vector has at least 2 elements. The offset of the start of
1374 the failing character and a reason code are placed in the vector.
1375
1376 6. When the UTF-8 string check fails for pcre_compile(), the offset that is
1377 now returned is for the first byte of the failing character, instead of the
1378 last byte inspected. This is an incompatible change, but I hope it is small
1379 enough not to be a problem. It makes the returned offset consistent with
1380 pcre_exec() and pcre_dfa_exec().
1381
1382 7. pcretest now gives a text phrase as well as the error number when
1383 pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
1384 failure, the offset and reason code are output.
1385
1386 8. When \R was used with a maximizing quantifier it failed to skip backwards
1387 over a \r\n pair if the subsequent match failed. Instead, it just skipped
1388 back over a single character (\n). This seems wrong (because it treated the
1389 two characters as a single entity when going forwards), conflicts with the
1390 documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
1391 behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
1392 has been changed.
1393
1394 9. Some internal refactoring has changed the processing so that the handling
1395 of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
1396 time (the PCRE_DOTALL option was changed this way some time ago: version
1397 7.7 change 16). This has made it possible to abolish the OP_OPT op code,
1398 which was always a bit of a fudge. It also means that there is one less
1399 argument for the match() function, which reduces its stack requirements
1400 slightly. This change also fixes an incompatibility with Perl: the pattern
1401 (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
1402
1403 10. More internal refactoring has drastically reduced the number of recursive
1404 calls to match() for possessively repeated groups such as (abc)++ when
1405 using pcre_exec().
1406
1407 11. While implementing 10, a number of bugs in the handling of groups were
1408 discovered and fixed:
1409
1410 (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
1411 (a|)*(?1) gave a compile-time internal error.
1412 ((a|)+)+ did not notice that the outer group could match an empty string.
1413 (^a|^)+ was not marked as anchored.
1414 (.*a|.*)+ was not marked as matching at start or after a newline.
1415
1416 12. Yet more internal refactoring has removed another argument from the match()
1417 function. Special calls to this function are now indicated by setting a
1418 value in a variable in the "match data" data block.
1419
1420 13. Be more explicit in pcre_study() instead of relying on "default" for
1421 opcodes that mean there is no starting character; this means that when new
1422 ones are added and accidentally left out of pcre_study(), testing should
1423 pick them up.
1424
1425 14. The -s option of pcretest has been documented for ages as being an old
1426 synonym of -m (show memory usage). I have changed it to mean "force study
1427 for every regex", that is, assume /S for every regex. This is similar to -i
1428 and -d etc. It's slightly incompatible, but I'm hoping nobody is still
1429 using it. It makes it easier to run collections of tests with and without
1430 study enabled, and thereby test pcre_study() more easily. All the standard
1431 tests are now run with and without -s (but some patterns can be marked as
1432 "never study" - see 20 below).
1433
1434 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
1435 restoration of the capturing data to the outer values was not happening
1436 correctly.
1437
1438 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
1439 empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
1440 pattern had matched an empty string, and so incorrectly returned a no
1441 match.
1442
1443 17. There was optimizing code for the last branch of non-capturing parentheses,
1444 and also for the obeyed branch of a conditional subexpression, which used
1445 tail recursion to cut down on stack usage. Unfortunately, now that there is
1446 the possibility of (*THEN) occurring in these branches, tail recursion is
1447 no longer possible because the return has to be checked for (*THEN). These
1448 two optimizations have therefore been removed. [But see 8.20/11 above.]
1449
1450 18. If a pattern containing \R was studied, it was assumed that \R always
1451 matched two bytes, thus causing the minimum subject length to be
1452 incorrectly computed because \R can also match just one byte.
1453
1454 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
1455 was incorrectly computed.
1456
1457 20. If /S is present twice on a test pattern in pcretest input, it now
1458 *disables* studying, thereby overriding the use of -s on the command line
1459 (see 14 above). This is necessary for one or two tests to keep the output
1460 identical in both cases.
1461
1462 21. When (*ACCEPT) was used in an assertion that matched an empty string and
1463 PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
1464
1465 22. When an atomic group that contained a capturing parenthesis was
1466 successfully matched, but the branch in which it appeared failed, the
1467 capturing was not being forgotten if a higher numbered group was later
1468 captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
1469 group 1 to "a", when in fact it should be unset. This applied to multi-
1470 branched capturing and non-capturing groups, repeated or not, and also to
1471 positive assertions (capturing in negative assertions does not happen
1472 in PCRE) and also to nested atomic groups.
1473
1474 23. Add the ++ qualifier feature to pcretest, to show the remainder of the
1475 subject after a captured substring, to make it easier to tell which of a
1476 number of identical substrings has been captured.
1477
1478 24. The way atomic groups are processed by pcre_exec() has been changed so that
1479 if they are repeated, backtracking one repetition now resets captured
1480 values correctly. For example, if ((?>(a+)b)+aabab) is matched against
1481 "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
1482 "aaa". Previously, it would have been "a". As part of this code
1483 refactoring, the way recursive calls are handled has also been changed.
1484
1485 25. If an assertion condition captured any substrings, they were not passed
1486 back unless some other capturing happened later. For example, if
1487 (?(?=(a))a) was matched against "a", no capturing was returned.
1488
1489 26. When studying a pattern that contained subroutine calls or assertions,
1490 the code for finding the minimum length of a possible match was handling
1491 direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
1492 group 1 called group 2 while simultaneously a separate group 2 called group
1493 1). A stack overflow occurred in this case. I have fixed this by limiting
1494 the recursion depth to 10.
1495
1496 27. Updated RunTest.bat in the distribution to the version supplied by Tom
1497 Fortmann. This supports explicit test numbers on the command line, and has
1498 argument validation and error reporting.
1499
1500 28. An instance of \X with an unlimited repeat could fail if at any point the
1501 first character it looked at was a mark character.
1502
1503 29. Some minor code refactoring concerning Unicode properties and scripts
1504 should reduce the stack requirement of match() slightly.
1505
1506 30. Added the '=' option to pcretest to check the setting of unused capturing
1507 slots at the end of the pattern, which are documented as being -1, but are
1508 not included in the return count.
1509
1510 31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
1511 compiled something random. Now it gives a compile-time error (as does
1512 Perl).
1513
1514 32. A *MARK encountered during the processing of a positive assertion is now
1515 recorded and passed back (compatible with Perl).
1516
1517 33. If --only-matching or --colour was set on a pcregrep call whose pattern
1518 had alternative anchored branches, the search for a second match in a line
1519 was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
1520 matched the line "0102" twice. The same bug affected patterns that started
1521 with a backwards assertion. For example /\b01|\b02/ also matched "0102"
1522 twice.
1523
1524 34. Previously, PCRE did not allow quantification of assertions. However, Perl
1525 does, and because of capturing effects, quantifying parenthesized
1526 assertions may at times be useful. Quantifiers are now allowed for
1527 parenthesized assertions.
1528
1529 35. A minor code tidy in pcre_compile() when checking options for \R usage.
1530
1531 36. \g was being checked for fancy things in a character class, when it should
1532 just be a literal "g".
1533
1534 37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
1535 appearance of a nested POSIX class supersedes an apparent external class.
1536 For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
1537 unescaped square brackets may also appear as part of class names. For
1538 example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
1539 more like Perl. (But see 8.20/1 above.)
1540
1541 38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
1542 was because it thought it was \N{name}, which is not supported).
1543
1544 39. Add minix to OS list not supporting the -S option in pcretest.
1545
1546 40. PCRE tries to detect cases of infinite recursion at compile time, but it
1547 cannot analyze patterns in sufficient detail to catch mutual recursions
1548 such as ((?1))((?2)). There is now a runtime test that gives an error if a
1549 subgroup is called recursively as a subpattern for a second time at the
1550 same position in the subject string. In previous releases this might have
1551 been caught by the recursion limit, or it might have run out of stack.
1552
1553 41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
1554 happen only once. PCRE was, however incorrectly giving a compile time error
1555 "recursive call could loop indefinitely" because it cannot analyze the
1556 pattern in sufficient detail. The compile time test no longer happens when
1557 PCRE is compiling a conditional subpattern, but actual runaway loops are
1558 now caught at runtime (see 40 above).
1559
1560 42. It seems that Perl allows any characters other than a closing parenthesis
1561 to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
1562 has been changed to be the same.
1563
1564 43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
1565 as not to get warnings when autogen.sh is called. Also changed
1566 AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
1567
1568 44. To help people who use pcregrep to scan files containing exceedingly long
1569 lines, the following changes have been made:
1570
1571 (a) The default value of the buffer size parameter has been increased from
1572 8K to 20K. (The actual buffer used is three times this size.)
1573
1574 (b) The default can be changed by ./configure --with-pcregrep-bufsize when
1575 PCRE is built.
1576
1577 (c) A --buffer-size=n option has been added to pcregrep, to allow the size
1578 to be set at run time.
1579
1580 (d) Numerical values in pcregrep options can be followed by K or M, for
1581 example --buffer-size=50K.
1582
1583 (e) If a line being scanned overflows pcregrep's buffer, an error is now
1584 given and the return code is set to 2.
1585
1586 45. Add a pointer to the latest mark to the callout data block.
1587
1588 46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
1589 partial match of an empty string instead of no match. This was specific to
1590 the use of ".".
1591
1592 47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
1593 complete match instead of a partial match. This bug was dependent on both
1594 the PCRE_UTF8 and PCRE_DOTALL options being set.
1595
1596 48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
1597 starting byte set, because \b was not being ignored.
1598
1599
1600 Version 8.12 15-Jan-2011
1601 ------------------------
1602
1603 1. Fixed some typos in the markup of the man pages, and wrote a script that
1604 checks for such things as part of the documentation building process.
1605
1606 2. On a big-endian 64-bit system, pcregrep did not correctly process the
1607 --match-limit and --recursion-limit options (added for 8.11). In
1608 particular, this made one of the standard tests fail. (The integer value
1609 went into the wrong half of a long int.)
1610
1611 3. If the --colour option was given to pcregrep with -v (invert match), it
1612 did strange things, either producing crazy output, or crashing. It should,
1613 of course, ignore a request for colour when reporting lines that do not
1614 match.
1615
1616 4. Another pcregrep bug caused similar problems if --colour was specified with
1617 -M (multiline) and the pattern match finished with a line ending.
1618
1619 5. In pcregrep, when a pattern that ended with a literal newline sequence was
1620 matched in multiline mode, the following line was shown as part of the
1621 match. This seems wrong, so I have changed it.
1622
1623 6. Another pcregrep bug in multiline mode, when --colour was specified, caused
1624 the check for further matches in the same line (so they could be coloured)
1625 to overrun the end of the current line. If another match was found, it was
1626 incorrectly shown (and then shown again when found in the next line).
1627
1628 7. If pcregrep was compiled under Windows, there was a reference to the
1629 function pcregrep_exit() before it was defined. I am assuming this was
1630 the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
1631 reported by a user. I've moved the definition above the reference.
1632
1633
1634 Version 8.11 10-Dec-2010
1635 ------------------------
1636
1637 1. (*THEN) was not working properly if there were untried alternatives prior
1638 to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
1639 backtracked to try for "b" instead of moving to the next alternative branch
1640 at the same level (in this case, to look for "c"). The Perl documentation
1641 is clear that when (*THEN) is backtracked onto, it goes to the "next
1642 alternative in the innermost enclosing group".
1643
1644 2. (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
1645 such as (A(*COMMIT)B(*THEN)C|D) any failure after matching A should
1646 result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
1647 (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
1648 (*THEN).
1649
1650 3. If \s appeared in a character class, it removed the VT character from
1651 the class, even if it had been included by some previous item, for example
1652 in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
1653 of \s, but is part of the POSIX "space" class.)
1654
1655 4. A partial match never returns an empty string (because you can always
1656 match an empty string at the end of the subject); however the checking for
1657 an empty string was starting at the "start of match" point. This has been
1658 changed to the "earliest inspected character" point, because the returned
1659 data for a partial match starts at this character. This means that, for
1660 example, /(?<=abc)def/ gives a partial match for the subject "abc"
1661 (previously it gave "no match").
1662
1663 5. Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
1664 of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
1665 previously a full match would be given. However, setting PCRE_PARTIAL_HARD
1666 has an implication that the given string is incomplete (because a partial
1667 match is preferred over a full match). For this reason, these items now
1668 give a partial match in this situation. [Aside: previously, the one case
1669 /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
1670 match rather than a full match, which was wrong by the old rules, but is
1671 now correct.]
1672
1673 6. There was a bug in the handling of #-introduced comments, recognized when
1674 PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
1675 If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
1676 UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
1677 scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
1678 but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
1679 places in pcre_compile().
1680
1681 7. Related to (6) above, when pcre_compile() was skipping #-introduced
1682 comments when looking ahead for named forward references to subpatterns,
1683 the only newline sequence it recognized was NL. It now handles newlines
1684 according to the set newline convention.
1685
1686 8. SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
1687 former, but used strtoul(), whereas pcretest avoided strtoul() but did not
1688 cater for a lack of strerror(). These oversights have been fixed.
1689
1690 9. Added --match-limit and --recursion-limit to pcregrep.
1691
1692 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
1693
1694 11. When the -o option was used, pcregrep was setting a return code of 1, even
1695 when matches were found, and --line-buffered was not being honoured.
1696
1697 12. Added an optional parentheses number to the -o and --only-matching options
1698 of pcregrep.
1699
1700 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
1701 can match an empty string. The code to do it in pcretest and pcredemo
1702 needed fixing:
1703
1704 (a) When the newline convention was "crlf", pcretest got it wrong, skipping
1705 only one byte after an empty string match just before CRLF (this case
1706 just got forgotten; "any" and "anycrlf" were OK).
1707
1708 (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
1709 mode when an empty string match preceded an ASCII character followed by
1710 a non-ASCII character. (The code for advancing by one character rather
1711 than one byte was nonsense.)
1712
1713 (c) The pcredemo.c sample program did not have any code at all to handle
1714 the cases when CRLF is a valid newline sequence.
1715
1716 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
1717 as a starting offset was within the subject string. There is now a new
1718 error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
1719 negative or greater than the length of the string. In order to test this,
1720 pcretest is extended to allow the setting of negative starting offsets.
1721
1722 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
1723 starting offset points to the beginning of a UTF-8 character was
1724 unnecessarily clumsy. I tidied it up.
1725
1726 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
1727 bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
1728
1729 17. Nobody had reported that the --include_dir option, which was added in
1730 release 7.7 should have been called --include-dir (hyphen, not underscore)
1731 for compatibility with GNU grep. I have changed it to --include-dir, but
1732 left --include_dir as an undocumented synonym, and the same for
1733 --exclude-dir, though that is not available in GNU grep, at least as of
1734 release 2.5.4.
1735
1736 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
1737 characters from a string of bytes) have been redefined so as not to use
1738 loops, in order to improve performance in some environments. At the same
1739 time, I abstracted some of the common code into auxiliary macros to save
1740 repetition (this should not affect the compiled code).
1741
1742 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
1743 compile-time error is now given if \c is not followed by an ASCII
1744 character, that is, a byte less than 128. (In EBCDIC mode, the code is
1745 different, and any byte value is allowed.)
1746
1747 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
1748 START_OPTIMIZE option, which is now allowed at compile time - but just
1749 passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
1750 to pcregrep and other applications that have no direct access to PCRE
1751 options. The new /Y option in pcretest sets this option when calling
1752 pcre_compile().
1753
1754 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
1755 back references. Groups containing recursive back references were forced to
1756 be atomic by that change, but in the case of named groups, the amount of
1757 memory required was incorrectly computed, leading to "Failed: internal
1758 error: code overflow". This has been fixed.
1759
1760 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
1761 pcretest.c, to avoid build problems in some Borland environments.
1762
1763
1764 Version 8.10 25-Jun-2010
1765 ------------------------
1766
1767 1. Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
1768 THEN.
1769
1770 2. (*ACCEPT) was not working when inside an atomic group.
1771
1772 3. Inside a character class, \B is treated as a literal by default, but
1773 faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
1774 causes the error). The code is unchanged, but I tidied the documentation.
1775
1776 4. Inside a character class, PCRE always treated \R and \X as literals,
1777 whereas Perl faults them if its -w option is set. I have changed PCRE so
1778 that it faults them when PCRE_EXTRA is set.
1779
1780 5. Added support for \N, which always matches any character other than
1781 newline. (It is the same as "." when PCRE_DOTALL is not set.)
1782
1783 6. When compiling pcregrep with newer versions of gcc which may have
1784 FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
1785 declared with attribute warn_unused_result" were given. Just casting the
1786 result to (void) does not stop the warnings; a more elaborate fudge is
1787 needed. I've used a macro to implement this.
1788
1789 7. Minor change to pcretest.c to avoid a compiler warning.
1790
1791 8. Added four artifical Unicode properties to help with an option to make
1792 \s etc use properties (see next item). The new properties are: Xan
1793 (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
1794
1795 9. Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
1796 use Unicode properties. (*UCP) at the start of a pattern can be used to set
1797 this option. Modified pcretest to add /W to test this facility. Added
1798 REG_UCP to make it available via the POSIX interface.
1799
1800 10. Added --line-buffered to pcregrep.
1801
1802 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
1803 studied, and the match started with a letter with a code point greater than
1804 127 whose first byte was different to the first byte of the other case of
1805 the letter, the other case of this starting letter was not recognized
1806 (#976).
1807
1808 12. If a pattern that was studied started with a repeated Unicode property
1809 test, for example, \p{Nd}+, there was the theoretical possibility of
1810 setting up an incorrect bitmap of starting bytes, but fortunately it could
1811 not have actually happened in practice until change 8 above was made (it
1812 added property types that matched character-matching opcodes).
1813
1814 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
1815 possible starting bytes for non-anchored patterns.
1816
1817 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
1818 \R, and also a number of cases that involve Unicode properties, both
1819 explicit and implicit when PCRE_UCP is set.
1820
1821 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
1822 input, it could crash or give wrong results if characters with values
1823 greater than 0xc0 were present in the subject string. (Detail: it assumed
1824 UTF-8 input when processing these items.)
1825
1826 16. Added a lot of (int) casts to avoid compiler warnings in systems where
1827 size_t is 64-bit (#991).
1828
1829 17. Added a check for running out of memory when PCRE is compiled with
1830 --disable-stack-for-recursion (#990).
1831
1832 18. If the last data line in a file for pcretest does not have a newline on
1833 the end, a newline was missing in the output.
1834
1835 19. The default pcre_chartables.c file recognizes only ASCII characters (values
1836 less than 128) in its various bitmaps. However, there is a facility for
1837 generating tables according to the current locale when PCRE is compiled. It
1838 turns out that in some environments, 0x85 and 0xa0, which are Unicode space
1839 characters, are recognized by isspace() and therefore were getting set in
1840 these tables, and indeed these tables seem to approximate to ISO 8859. This
1841 caused a problem in UTF-8 mode when pcre_study() was used to create a list
1842 of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
1843 which of course cannot start UTF-8 characters. I have changed the code so
1844 that only real ASCII characters (less than 128) and the correct starting
1845 bytes for UTF-8 encodings are set for characters greater than 127 when in
1846 UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
1847 altogether.)
1848
1849 20. Added the /T option to pcretest so as to be able to run tests with non-
1850 standard character tables, thus making it possible to include the tests
1851 used for 19 above in the standard set of tests.
1852
1853 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
1854 reference to a subpattern the other side of a comment that contains an
1855 opening parenthesis caused either an internal compiling error, or a
1856 reference to the wrong subpattern.
1857
1858
1859 Version 8.02 19-Mar-2010
1860 ------------------------
1861
1862 1. The Unicode data tables have been updated to Unicode 5.2.0.
1863
1864 2. Added the option --libs-cpp to pcre-config, but only when C++ support is
1865 configured.
1866
1867 3. Updated the licensing terms in the pcregexp.pas file, as agreed with the
1868 original author of that file, following a query about its status.
1869
1870 4. On systems that do not have stdint.h (e.g. Solaris), check for and include
1871 inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
1872
1873 5. A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
1874 quantifier applied to a forward-referencing subroutine call, could compile
1875 incorrect code or give the error "internal error: previously-checked
1876 referenced subpattern not found".
1877
1878 6. Both MS Visual Studio and Symbian OS have problems with initializing
1879 variables to point to external functions. For these systems, therefore,
1880 pcre_malloc etc. are now initialized to local functions that call the
1881 relevant global functions.
1882
1883 7. There were two entries missing in the vectors called coptable and poptable
1884 in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
1885 I've fixed the data, and added a kludgy way of testing at compile time that
1886 the lengths are correct (equal to the number of opcodes).
1887
1888 8. Following on from 7, I added a similar kludge to check the length of the
1889 eint vector in pcreposix.c.
1890
1891 9. Error texts for pcre_compile() are held as one long string to avoid too
1892 much relocation at load time. To find a text, the string is searched,
1893 counting zeros. There was no check for running off the end of the string,
1894 which could happen if a new error number was added without updating the
1895 string.
1896
1897 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
1898
1899 11. \K was not working if it appeared in an atomic group or in a group that
1900 was called as a "subroutine", or in an assertion. Perl 5.11 documents that
1901 \K is "not well defined" if used in an assertion. PCRE now accepts it if
1902 the assertion is positive, but not if it is negative.
1903
1904 12. Change 11 fortuitously reduced the size of the stack frame used in the
1905 "match()" function of pcre_exec.c by one pointer. Forthcoming
1906 implementation of support for (*MARK) will need an extra pointer on the
1907 stack; I have reserved it now, so that the stack frame size does not
1908 decrease.
1909
1910 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
1911 item in branch that calls a recursion is a subroutine call - as in the
1912 second branch in the above example - was incorrectly given the compile-
1913 time error "recursive call could loop indefinitely" because pcre_compile()
1914 was not correctly checking the subroutine for matching a non-empty string.
1915
1916 14. The checks for overrunning compiling workspace could trigger after an
1917 overrun had occurred. This is a "should never occur" error, but it can be
1918 triggered by pathological patterns such as hundreds of nested parentheses.
1919 The checks now trigger 100 bytes before the end of the workspace.
1920
1921 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
1922
1923
1924 Version 8.01 19-Jan-2010
1925 ------------------------
1926
1927 1. If a pattern contained a conditional subpattern with only one branch (in
1928 particular, this includes all (*DEFINE) patterns), a call to pcre_study()
1929 computed the wrong minimum data length (which is of course zero for such
1930 subpatterns). This could cause incorrect "no match" results.
1931
1932 2. For patterns such as (?i)a(?-i)b|c where an option setting at the start of
1933 the pattern is reset in the first branch, pcre_compile() failed with
1934 "internal error: code overflow at offset...". This happened only when
1935 the reset was to the original external option setting. (An optimization
1936 abstracts leading options settings into an external setting, which was the
1937 cause of this.)
1938
1939 3. A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
1940 of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
1941 assertion pattern did not match (meaning that the assertion was true), it
1942 was incorrectly treated as false if the SKIP had been reached during the
1943 matching. This also applied to assertions used as conditions.
1944
1945 4. If an item that is not supported by pcre_dfa_exec() was encountered in an
1946 assertion subpattern, including such a pattern used as a condition,
1947 unpredictable results occurred, instead of the error return
1948 PCRE_ERROR_DFA_UITEM.
1949
1950 5. The C++ GlobalReplace function was not working like Perl for the special
1951 situation when an empty string is matched. It now does the fancy magic
1952 stuff that is necessary.
1953
1954 6. In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
1955 removed. (These were left over from very, very early versions of PCRE.)
1956
1957 7. Some cosmetic changes to the code to make life easier when compiling it
1958 as part of something else:
1959
1960 (a) Change DEBUG to PCRE_DEBUG.
1961
1962 (b) In pcre_compile(), rename the member of the "branch_chain" structure
1963 called "current" as "current_branch", to prevent a collision with the
1964 Linux macro when compiled as a kernel module.
1965
1966 (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
1967 prevent a collision with the Linux macro when compiled as a kernel
1968 module.
1969
1970 8. In pcre_compile() there are some checks for integer overflows that used to
1971 cast potentially large values to (double). This has been changed to that
1972 when building, a check for int64_t is made, and if it is found, it is used
1973 instead, thus avoiding the use of floating point arithmetic. (There is no
1974 other use of FP in PCRE.) If int64_t is not found, the fallback is to
1975 double.
1976
1977 9. Added two casts to avoid signed/unsigned warnings from VS Studio Express
1978 2005 (difference between two addresses compared to an unsigned value).
1979
1980 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
1981 custom one, because of the following reported problem in Windows:
1982
1983 - libbz2 uses the Pascal calling convention (WINAPI) for the functions
1984 under Win32.
1985 - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
1986 therefore missing the function definition.
1987 - The compiler thus generates a "C" signature for the test function.
1988 - The linker fails to find the "C" function.
1989 - PCRE fails to configure if asked to do so against libbz2.
1990
1991 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
1992 messages were output:
1993
1994 Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
1995 rerunning libtoolize, to keep the correct libtool macros in-tree.
1996 Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
1997
1998 I have done both of these things.
1999
2000 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
2001 most of the time, it *can* run out if it is given a pattern that contains a
2002 runaway infinite recursion. I updated the discussion in the pcrestack man
2003 page.
2004
2005 13. Now that we have gone to the x.xx style of version numbers, the minor
2006 version may start with zero. Using 08 or 09 is a bad idea because users
2007 might check the value of PCRE_MINOR in their code, and 08 or 09 may be
2008 interpreted as invalid octal numbers. I've updated the previous comment in
2009 configure.ac, and also added a check that gives an error if 08 or 09 are
2010 used.
2011
2012 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
2013 causing partial matching to fail when the end of the subject matched \W
2014 in a UTF-8 pattern where \W was quantified with a minimum of 3.
2015
2016 15. There were some discrepancies between the declarations in pcre_internal.h
2017 of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
2018 their definitions. The declarations used "const uschar *" and the
2019 definitions used USPTR. Even though USPTR is normally defined as "const
2020 unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
2021 reported that: "This difference in casting confuses some C++ compilers, for
2022 example, SunCC recognizes above declarations as different functions and
2023 generates broken code for hbpcre." I have changed the declarations to use
2024 USPTR.
2025
2026 16. GNU libtool is named differently on some systems. The autogen.sh script now
2027 tries several variants such as glibtoolize (MacOSX) and libtoolize1x
2028 (FreeBSD).
2029
2030 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
2031 (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
2032 comment: "Figure out how to create a longlong from a string: strtoll and
2033 equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
2034 instance, but it only takes 2 args instead of 3!"
2035
2036 18. A subtle bug concerned with back references has been fixed by a change of
2037 specification, with a corresponding code fix. A pattern such as
2038 ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
2039 refers, was giving matches when it shouldn't. For example, xa=xaaa would
2040 match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
2041 same bug. Such groups have to be quantified to be useful, or contained
2042 inside another quantified group. (If there's no repetition, the reference
2043 can never match.) The problem arises because, having left the group and
2044 moved on to the rest of the pattern, a later failure that backtracks into
2045 the group uses the captured value from the final iteration of the group
2046 rather than the correct earlier one. I have fixed this in PCRE by forcing
2047 any group that contains a reference to itself to be an atomic group; that
2048 is, there cannot be any backtracking into it once it has completed. This is
2049 similar to recursive and subroutine calls.
2050
2051
2052 Version 8.00 19-Oct-09
2053 ----------------------
2054
2055 1. The table for translating pcre_compile() error codes into POSIX error codes
2056 was out-of-date, and there was no check on the pcre_compile() error code
2057 being within the table. This could lead to an OK return being given in
2058 error.
2059
2060 2. Changed the call to open a subject file in pcregrep from fopen(pathname,
2061 "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
2062 in a Windows environment.
2063
2064 3. The pcregrep --count option prints the count for each file even when it is
2065 zero, as does GNU grep. However, pcregrep was also printing all files when
2066 --files-with-matches was added. Now, when both options are given, it prints
2067 counts only for those files that have at least one match. (GNU grep just
2068 prints the file name in this circumstance, but including the count seems
2069 more useful - otherwise, why use --count?) Also ensured that the
2070 combination -clh just lists non-zero counts, with no names.
2071
2072 4. The long form of the pcregrep -F option was incorrectly implemented as
2073 --fixed_strings instead of --fixed-strings. This is an incompatible change,
2074 but it seems right to fix it, and I didn't think it was worth preserving
2075 the old behaviour.
2076
2077 5. The command line items --regex=pattern and --regexp=pattern were not
2078 recognized by pcregrep, which required --regex pattern or --regexp pattern
2079 (with a space rather than an '='). The man page documented the '=' forms,
2080 which are compatible with GNU grep; these now work.
2081
2082 6. No libpcreposix.pc file was created for pkg-config; there was just
2083 libpcre.pc and libpcrecpp.pc. The omission has been rectified.
2084
2085 7. Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
2086 when UCP support is not needed, by modifying the Python script that
2087 generates it from Unicode data files. This should not matter if the module
2088 is correctly used as a library, but I received one complaint about 50K of
2089 unwanted data. My guess is that the person linked everything into his
2090 program rather than using a library. Anyway, it does no harm.
2091
2092 8. A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
2093 was a minimum greater than 1 for a wide character in a possessive
2094 repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
2095 which had an unlimited repeat of a nested, fixed maximum repeat of a wide
2096 character. Chaos in the form of incorrect output or a compiling loop could
2097 result.
2098
2099 9. The restrictions on what a pattern can contain when partial matching is
2100 requested for pcre_exec() have been removed. All patterns can now be
2101 partially matched by this function. In addition, if there are at least two
2102 slots in the offset vector, the offset of the earliest inspected character
2103 for the match and the offset of the end of the subject are set in them when
2104 PCRE_ERROR_PARTIAL is returned.
2105
2106 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
2107 synonymous with PCRE_PARTIAL, for backwards compatibility, and
2108 PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
2109 and may be more useful for multi-segment matching.
2110
2111 11. Partial matching with pcre_exec() is now more intuitive. A partial match
2112 used to be given if ever the end of the subject was reached; now it is
2113 given only if matching could not proceed because another character was
2114 needed. This makes a difference in some odd cases such as Z(*FAIL) with the
2115 string "Z", which now yields "no match" instead of "partial match". In the
2116 case of pcre_dfa_exec(), "no match" is given if every matching path for the
2117 final character ended with (*FAIL).
2118
2119 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
2120 if the pattern had a "must contain" character that was already found in the
2121 earlier partial match, unless partial matching was again requested. For
2122 example, with the pattern /dog.(body)?/, the "must contain" character is
2123 "g". If the first part-match was for the string "dog", restarting with
2124 "sbody" failed. This bug has been fixed.
2125
2126 13. The string returned by pcre_dfa_exec() after a partial match has been
2127 changed so that it starts at the first inspected character rather than the
2128 first character of the match. This makes a difference only if the pattern
2129 starts with a lookbehind assertion or \b or \B (\K is not supported by
2130 pcre_dfa_exec()). It's an incompatible change, but it makes the two
2131 matching functions compatible, and I think it's the right thing to do.
2132
2133 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
2134 so that the demonstration program is easily available in environments where
2135 PCRE has not been installed from source.
2136
2137 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
2138 libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
2139 library.
2140
2141 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
2142 It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
2143 is not the first non-POSIX option to be added. Clearly some people find
2144 these options useful.
2145
2146 17. If a caller to the POSIX matching function regexec() passes a non-zero
2147 value for nmatch with a NULL value for pmatch, the value of
2148 nmatch is forced to zero.
2149
2150 18. RunGrepTest did not have a test for the availability of the -u option of
2151 the diff command, as RunTest does. It now checks in the same way as
2152 RunTest, and also checks for the -b option.
2153
2154 19. If an odd number of negated classes containing just a single character
2155 interposed, within parentheses, between a forward reference to a named
2156 subpattern and the definition of the subpattern, compilation crashed with
2157 an internal error, complaining that it could not find the referenced
2158 subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
2159 [The bug was that it was starting one character too far in when skipping
2160 over the character class, thus treating the ] as data rather than
2161 terminating the class. This meant it could skip too much.]
2162
2163 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
2164 /g option in pcretest when the pattern contains \K, which makes it possible
2165 to have an empty string match not at the start, even when the pattern is
2166 anchored. Updated pcretest and pcredemo to use this option.
2167
2168 21. If the maximum number of capturing subpatterns in a recursion was greater
2169 than the maximum at the outer level, the higher number was returned, but
2170 with unset values at the outer level. The correct (outer level) value is
2171 now given.
2172
2173 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
2174 PCRE did not set those parentheses (unlike Perl). I have now found a way to
2175 make it do so. The string so far is captured, making this feature
2176 compatible with Perl.
2177
2178 23. The tests have been re-organized, adding tests 11 and 12, to make it
2179 possible to check the Perl 5.10 features against Perl 5.10.
2180
2181 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
2182 pattern matches a fixed length string. PCRE did not allow this; now it
2183 does. Neither allows recursion.
2184
2185 25. I finally figured out how to implement a request to provide the minimum
2186 length of subject string that was needed in order to match a given pattern.
2187 (It was back references and recursion that I had previously got hung up
2188 on.) This code has now been added to pcre_study(); it finds a lower bound
2189 to the length of subject needed. It is not necessarily the greatest lower
2190 bound, but using it to avoid searching strings that are too short does give
2191 some useful speed-ups. The value is available to calling programs via
2192 pcre_fullinfo().
2193
2194 26. While implementing 25, I discovered to my embarrassment that pcretest had
2195 not been passing the result of pcre_study() to pcre_dfa_exec(), so the
2196 study optimizations had never been tested with that matching function.
2197 Oops. What is worse, even when it was passed study data, there was a bug in
2198 pcre_dfa_exec() that meant it never actually used it. Double oops. There
2199 were also very few tests of studied patterns with pcre_dfa_exec().
2200
2201 27. If (?| is used to create subpatterns with duplicate numbers, they are now
2202 allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
2203 on the other side of the coin, they are no longer allowed to have different
2204 names, because these cannot be distinguished in PCRE, and this has caused
2205 confusion. (This is a difference from Perl.)
2206
2207 28. When duplicate subpattern names are present (necessarily with different
2208 numbers, as required by 27 above), and a test is made by name in a
2209 conditional pattern, either for a subpattern having been matched, or for
2210 recursion in such a pattern, all the associated numbered subpatterns are
2211 tested, and the overall condition is true if the condition is true for any
2212 one of them. This is the way Perl works, and is also more like the way
2213 testing by number works.
2214
2215
2216 Version 7.9 11-Apr-09
2217 ---------------------
2218
2219 1. When building with support for bzlib/zlib (pcregrep) and/or readline
2220 (pcretest), all targets were linked against these libraries. This included
2221 libpcre, libpcreposix, and libpcrecpp, even though they do not use these
2222 libraries. This caused unwanted dependencies to be created. This problem
2223 has been fixed, and now only pcregrep is linked with bzlib/zlib and only
2224 pcretest is linked with readline.
2225
2226 2. The "typedef int BOOL" in pcre_internal.h that was included inside the
2227 "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
2228 moved outside it again, because FALSE and TRUE are already defined in AIX,
2229 but BOOL is not.
2230
2231 3. The pcre_config() function was treating the PCRE_MATCH_LIMIT and
2232 PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
2233
2234 4. The pcregrep documentation said spaces were inserted as well as colons (or
2235 hyphens) following file names and line numbers when outputting matching
2236 lines. This is not true; no spaces are inserted. I have also clarified the
2237 wording for the --colour (or --color) option.
2238
2239 5. In pcregrep, when --colour was used with -o, the list of matching strings
2240 was not coloured; this is different to GNU grep, so I have changed it to be
2241 the same.
2242
2243 6. When --colo(u)r was used in pcregrep, only the first matching substring in
2244 each matching line was coloured. Now it goes on to look for further matches
2245 of any of the test patterns, which is the same behaviour as GNU grep.
2246
2247 7. A pattern that could match an empty string could cause pcregrep to loop; it
2248 doesn't make sense to accept an empty string match in pcregrep, so I have
2249 locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
2250 seems to be how GNU grep behaves. [But see later change 40 for release
2251 8.33.]
2252
2253 8. The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
2254 start or after a newline", because the conditional assertion was not being
2255 correctly handled. The rule now is that both the assertion and what follows
2256 in the first alternative must satisfy the test.
2257
2258 9. If auto-callout was enabled in a pattern with a conditional group whose
2259 condition was an assertion, PCRE could crash during matching, both with
2260 pcre_exec() and pcre_dfa_exec().
2261
2262 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
2263 used for matching.
2264
2265 11. Unicode property support in character classes was not working for
2266 characters (bytes) greater than 127 when not in UTF-8 mode.
2267
2268 12. Added the -M command line option to pcretest.
2269
2270 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
2271
2272 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
2273
2274 16. Added comments and documentation about mis-use of no_arg in the C++
2275 wrapper.
2276
2277 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
2278 from Martin Jerabek that uses macro names for all relevant character and
2279 string constants.
2280
2281 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
2282 SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
2283 SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
2284 these, but not everybody uses configure.
2285
2286 19. A conditional group that had only one branch was not being correctly
2287 recognized as an item that could match an empty string. This meant that an
2288 enclosing group might also not be so recognized, causing infinite looping
2289 (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
2290 with the subject "ab", where knowledge that the repeated group can match
2291 nothing is needed in order to break the loop.
2292
2293 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
2294 exec(), but without supplying a callout function, matching went wrong.
2295
2296 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
2297 leak if the size of the offset vector was greater than 30. When the vector
2298 is smaller, the saved offsets during recursion go onto a local stack
2299 vector, but for larger vectors malloc() is used. It was failing to free
2300 when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
2301 error, in fact).
2302
2303 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
2304 heapframe that is used only when UTF-8 support is enabled. This caused no
2305 problem, but was untidy.
2306
2307 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
2308 CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
2309 included within another project.
2310
2311 24. Steven Van Ingelgem's patches to add more options to the CMake support,
2312 slightly modified by me:
2313
2314 (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
2315 not building pcregrep.
2316
2317 (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
2318 if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
2319
2320 25. Forward references, both numeric and by name, in patterns that made use of
2321 duplicate group numbers, could behave incorrectly or give incorrect errors,
2322 because when scanning forward to find the reference group, PCRE was not
2323 taking into account the duplicate group numbers. A pattern such as
2324 ^X(?3)(a)(?|(b)|(q))(Y) is an example.
2325
2326 26. Changed a few more instances of "const unsigned char *" to USPTR, making
2327 the feature of a custom pointer more persuasive (as requested by a user).
2328
2329 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
2330 pcretest.c, inside #ifndefs, because it seems they are sometimes already
2331 pre-defined.
2332
2333 28. Added support for (*UTF8) at the start of a pattern.
2334
2335 29. Arrange for flags added by the "release type" setting in CMake to be shown
2336 in the configuration summary.
2337
2338
2339 Version 7.8 05-Sep-08
2340 ---------------------
2341
2342 1. Replaced UCP searching code with optimized version as implemented for Ad
2343 Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
2344 stage table and inline lookup instead of a function, giving speed ups of 2
2345 to 5 times on some simple patterns that I tested. Permission was given to
2346 distribute the MultiStage2.py script that generates the tables (it's not in
2347 the tarball, but is in the Subversion repository).
2348
2349 2. Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
2350 scripts.
2351
2352 3. Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
2353 a group with a zero qualifier. The result of the study could be incorrect,
2354 or the function might crash, depending on the pattern.
2355
2356 4. Caseless matching was not working for non-ASCII characters in back
2357 references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
2358 It now works when Unicode Property Support is available.
2359
2360 5. In pcretest, an escape such as \x{de} in the data was always generating
2361 a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
2362 non-UTF-8 mode. If the value is greater than 255, it gives a warning about
2363 truncation.
2364
2365 6. Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
2366
2367 7. Added two (int) casts to pcregrep when printing the difference of two
2368 pointers, in case they are 64-bit values.
2369
2370 8. Added comments about Mac OS X stack usage to the pcrestack man page and to
2371 test 2 if it fails.
2372
2373 9. Added PCRE_CALL_CONVENTION just before the names of all exported functions,
2374 and a #define of that name to empty if it is not externally set. This is to
2375 allow users of MSVC to set it if necessary.
2376
2377 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
2378 the convenience functions in the pcre_get.c source file.
2379
2380 11. An option change at the start of a pattern that had top-level alternatives
2381 could cause overwriting and/or a crash. This command provoked a crash in
2382 some environments:
2383
2384 printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
2385
2386 This potential security problem was recorded as CVE-2008-2371.
2387
2388 12. For a pattern where the match had to start at the beginning or immediately
2389 after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
2390 pcre_dfa_exec() could read past the end of the passed subject if there was
2391 no match. To help with detecting such bugs (e.g. with valgrind), I modified
2392 pcretest so that it places the subject at the end of its malloc-ed buffer.
2393
2394 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
2395 exec() might read past the end of the data buffer in UTF-8 mode.
2396
2397 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
2398 the data contained the byte 0x85 as part of a UTF-8 character within its
2399 first line. This applied both to normal and DFA matching.
2400
2401 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
2402 /^[^d]*?$/8 failed to match "abc".
2403
2404 16. Added a missing copyright notice to pcrecpp_internal.h.
2405
2406 17. Make it more clear in the documentation that values returned from
2407 pcre_exec() in ovector are byte offsets, not character counts.
2408
2409 18. Tidied a few places to stop certain compilers from issuing warnings.
2410
2411 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
2412 supplied by Stefan Weber. I made a further small update for 7.8 because
2413 there is a change of source arrangements: the pcre_searchfuncs.c module is
2414 replaced by pcre_ucd.c.
2415
2416
2417 Version 7.7 07-May-08
2418 ---------------------
2419
2420 1. Applied Craig's patch to sort out a long long problem: "If we can't convert
2421 a string to a long long, pretend we don't even have a long long." This is
2422 done by checking for the strtoq, strtoll, and _strtoi64 functions.
2423
2424 2. Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
2425 pre-7.6 versions, which defined a global no_arg variable instead of putting
2426 it in the RE class. (See also #8 below.)
2427
2428 3. Remove a line of dead code, identified by coverity and reported by Nuno
2429 Lopes.
2430
2431 4. Fixed two related pcregrep bugs involving -r with --include or --exclude:
2432
2433 (1) The include/exclude patterns were being applied to the whole pathnames
2434 of files, instead of just to the final components.
2435
2436 (2) If there was more than one level of directory, the subdirectories were
2437 skipped unless they satisfied the include/exclude conditions. This is
2438 inconsistent with GNU grep (and could even be seen as contrary to the
2439 pcregrep specification - which I improved to make it absolutely clear).
2440 The action now is always to scan all levels of directory, and just
2441 apply the include/exclude patterns to regular files.
2442
2443 5. Added the --include_dir and --exclude_dir patterns to pcregrep, and used
2444 --exclude_dir in the tests to avoid scanning .svn directories.
2445
2446 6. Applied Craig's patch to the QuoteMeta function so that it escapes the
2447 NUL character as backslash + 0 rather than backslash + NUL, because PCRE
2448 doesn't support NULs in patterns.
2449
2450 7. Added some missing "const"s to declarations of static tables in
2451 pcre_compile.c and pcre_dfa_exec.c.
2452
2453 8. Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
2454 caused by fix #2 above. (Subsequently also a second patch to fix the
2455 first patch. And a third patch - this was a messy problem.)
2456
2457 9. Applied Craig's patch to remove the use of push_back().
2458
2459 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
2460 matching function regexec().
2461
2462 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
2463 which, however, unlike Perl's \g{...}, are subroutine calls, not back
2464 references. PCRE supports relative numbers with this syntax (I don't think
2465 Oniguruma does).
2466
2467 12. Previously, a group with a zero repeat such as (...){0} was completely
2468 omitted from the compiled regex. However, this means that if the group
2469 was called as a subroutine from elsewhere in the pattern, things went wrong
2470 (an internal error was given). Such groups are now left in the compiled
2471 pattern, with a new opcode that causes them to be skipped at execution
2472 time.
2473
2474 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
2475 to the way PCRE behaves:
2476
2477 (a) A lone ] character is dis-allowed (Perl treats it as data).
2478
2479 (b) A back reference to an unmatched subpattern matches an empty string
2480 (Perl fails the current match path).
2481
2482 (c) A data ] in a character class must be notated as \] because if the
2483 first data character in a class is ], it defines an empty class. (In
2484 Perl it is not possible to have an empty class.) The empty class []
2485 never matches; it forces failure and is equivalent to (*FAIL) or (?!).
2486 The negative empty class [^] matches any one character, independently
2487 of the DOTALL setting.
2488
2489 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
2490 non-existent subpattern following a character class starting with ']' and
2491 containing () gave an internal compiling error instead of "reference to
2492 non-existent subpattern". Fortunately, when the pattern did exist, the
2493 compiled code was correct. (When scanning forwards to check for the
2494 existence of the subpattern, it was treating the data ']' as terminating
2495 the class, so got the count wrong. When actually compiling, the reference
2496 was subsequently set up correctly.)
2497
2498 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
2499 it was being rejected as not supported by pcre_dfa_exec(), even though
2500 other assertions are supported. I have made pcre_dfa_exec() support
2501 (*FAIL).
2502
2503 16. The implementation of 13c above involved the invention of a new opcode,
2504 OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
2505 cannot be changed at match time, I realized I could make a small
2506 improvement to matching performance by compiling OP_ALLANY instead of
2507 OP_ANY for "." when DOTALL was set, and then removing the runtime tests
2508 on the OP_ANY path.
2509
2510 17. Compiling pcretest on Windows with readline support failed without the
2511 following two fixes: (1) Make the unistd.h include conditional on
2512 HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
2513
2514 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
2515 ncurses library to be included for pcretest when ReadLine support is
2516 requested, but also to allow for it to be overridden. This patch came from
2517 Daniel Bergström.
2518
2519 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
2520 as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
2521 any errors with the current Unicode tables. Thanks to Peter Kankowski for
2522 spotting this.
2523
2524
2525 Version 7.6 28-Jan-08
2526 ---------------------
2527
2528 1. A character class containing a very large number of characters with
2529 codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
2530 overflow.
2531
2532 2. Patch to cut out the "long long" test in pcrecpp_unittest when
2533 HAVE_LONG_LONG is not defined.
2534
2535 3. Applied Christian Ehrlicher's patch to update the CMake build files to
2536 bring them up to date and include new features. This patch includes:
2537
2538 - Fixed PH's badly added libz and libbz2 support.
2539 - Fixed a problem with static linking.
2540 - Added pcredemo. [But later removed - see 7 below.]
2541 - Fixed dftables problem and added an option.
2542 - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
2543 HAVE_LONG_LONG.
2544 - Added readline support for pcretest.
2545 - Added an listing of the option settings after cmake has run.
2546
2547 4. A user submitted a patch to Makefile that makes it easy to create
2548 "pcre.dll" under mingw when using Configure/Make. I added stuff to
2549 Makefile.am that cause it to include this special target, without
2550 affecting anything else. Note that the same mingw target plus all
2551 the other distribution libraries and programs are now supported
2552 when configuring with CMake (see 6 below) instead of with
2553 Configure/Make.
2554
2555 5. Applied Craig's patch that moves no_arg into the RE class in the C++ code.
2556 This is an attempt to solve the reported problem "pcrecpp::no_arg is not
2557 exported in the Windows port". It has not yet been confirmed that the patch
2558 solves the problem, but it does no harm.
2559
2560 6. Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
2561 NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
2562 with CMake, and also correct the comment about stack recursion.
2563
2564 7. Remove the automatic building of pcredemo from the ./configure system and
2565 from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
2566 of a program that users should build themselves after PCRE is installed, so
2567 building it automatically is not really right. What is more, it gave
2568 trouble in some build environments.
2569
2570 8. Further tidies to CMakeLists.txt from Sheri and Christian.
2571
2572
2573 Version 7.5 10-Jan-08
2574 ---------------------
2575
2576 1. Applied a patch from Craig: "This patch makes it possible to 'ignore'
2577 values in parens when parsing an RE using the C++ wrapper."
2578
2579 2. Negative specials like \S did not work in character classes in UTF-8 mode.
2580 Characters greater than 255 were excluded from the class instead of being
2581 included.
2582
2583 3. The same bug as (2) above applied to negated POSIX classes such as
2584 [:^space:].
2585
2586 4. PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
2587 defined or documented. It seems to have been a typo for PCRE_STATIC, so
2588 I have changed it.
2589
2590 5. The construct (?&) was not diagnosed as a syntax error (it referenced the
2591 first named subpattern) and a construct such as (?&a) would reference the
2592 first named subpattern whose name started with "a" (in other words, the
2593 length check was missing). Both these problems are fixed. "Subpattern name
2594 expected" is now given for (?&) (a zero-length name), and this patch also
2595 makes it give the same error for \k'' (previously it complained that that
2596 was a reference to a non-existent subpattern).
2597
2598 6. The erroneous patterns (?+-a) and (?-+a) give different error messages;
2599 this is right because (?- can be followed by option settings as well as by
2600 digits. I have, however, made the messages clearer.
2601
2602 7. Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
2603 than the number used in the conditional) now cause a compile-time error.
2604 This is actually not compatible with Perl, which accepts such patterns, but
2605 treats the conditional as always being FALSE (as PCRE used to), but it
2606 seems to me that giving a diagnostic is better.
2607
2608 8. Change "alphameric" to the more common word "alphanumeric" in comments
2609 and messages.
2610
2611 9. Fix two occurrences of "backslash" in comments that should have been
2612 "backspace".
2613
2614 10. Remove two redundant lines of code that can never be obeyed (their function
2615 was moved elsewhere).
2616
2617 11. The program that makes PCRE's Unicode character property table had a bug
2618 which caused it to generate incorrect table entries for sequences of
2619 characters that have the same character type, but are in different scripts.
2620 It amalgamated them into a single range, with the script of the first of
2621 them. In other words, some characters were in the wrong script. There were
2622 thirteen such cases, affecting characters in the following ranges:
2623
2624 U+002b0 - U+002c1
2625 U+0060c - U+0060d
2626 U+0061e - U+00612
2627 U+0064b - U+0065e
2628 U+0074d - U+0076d
2629 U+01800 - U+01805
2630 U+01d00 - U+01d77
2631 U+01d9b - U+01dbf
2632 U+0200b - U+0200f
2633 U+030fc - U+030fe
2634 U+03260 - U+0327f
2635 U+0fb46 - U+0fbb1
2636 U+10450 - U+1049d
2637
2638 12. The -o option (show only the matching part of a line) for pcregrep was not
2639 compatible with GNU grep in that, if there was more than one match in a
2640 line, it showed only the first of them. It now behaves in the same way as
2641 GNU grep.
2642
2643 13. If the -o and -v options were combined for pcregrep, it printed a blank
2644 line for every non-matching line. GNU grep prints nothing, and pcregrep now
2645 does the same. The return code can be used to tell if there were any
2646 non-matching lines.
2647
2648 14. Added --file-offsets and --line-offsets to pcregrep.
2649
2650 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
2651 infinitely looping recursion. The bug was that positive lookaheads were not
2652 being skipped when checking for a possible empty match (negative lookaheads
2653 and both kinds of lookbehind were skipped).
2654
2655 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
2656 inclusion of <windows.h> to before rather than after the definition of
2657 INVALID_FILE_ATTRIBUTES (patch from David Byron).
2658
2659 17. Specifying a possessive quantifier with a specific limit for a Unicode
2660 character property caused pcre_compile() to compile bad code, which led at
2661 runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
2662 are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
2663 caused the error; without that there was no problem.
2664
2665 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
2666
2667 19. Added --enable-pcretest-libreadline.
2668
2669 20. In pcrecpp.cc, the variable 'count' was incremented twice in
2670 RE::GlobalReplace(). As a result, the number of replacements returned was
2671 double what it should be. I removed one of the increments, but Craig sent a
2672 later patch that removed the other one (the right fix) and added unit tests
2673 that check the return values (which was not done before).
2674
2675 21. Several CMake things:
2676
2677 (1) Arranged that, when cmake is used on Unix, the libraries end up with
2678 the names libpcre and libpcreposix, not just pcre and pcreposix.
2679
2680 (2) The above change means that pcretest and pcregrep are now correctly
2681 linked with the newly-built libraries, not previously installed ones.
2682
2683 (3) Added PCRE_SUPPORT_LIBREADLINE, PCRE_SUPPORT_LIBZ, PCRE_SUPPORT_LIBBZ2.
2684
2685 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
2686 crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
2687 UTF-8 newline character). The key issue is that the pattern starts .*;
2688 this means that the match must be either at the beginning, or after a
2689 newline. The bug was in the code for advancing after a failed match and
2690 checking that the new position followed a newline. It was not taking
2691 account of UTF-8 characters correctly.
2692
2693 23. PCRE was behaving differently from Perl in the way it recognized POSIX
2694 character classes. PCRE was not treating the sequence [:...:] as a
2695 character class unless the ... were all letters. Perl, however, seems to
2696 allow any characters between [: and :], though of course it rejects as
2697 unknown any "names" that contain non-letters, because all the known class
2698 names consist only of letters. Thus, Perl gives an error for [[:1234:]],
2699 for example, whereas PCRE did not - it did not recognize a POSIX character
2700 class. This seemed a bit dangerous, so the code has been changed to be
2701 closer to Perl. The behaviour is not identical to Perl, because PCRE will
2702 diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
2703 treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
2704 Perl does, and where it didn't before.
2705
2706 24. Rewrite so as to remove the single use of %n from pcregrep because in some
2707 Windows environments %n is disabled by default.
2708
2709
2710 Version 7.4 21-Sep-07
2711 ---------------------
2712
2713 1. Change 7.3/28 was implemented for classes by looking at the bitmap. This
2714 means that a class such as [\s] counted as "explicit reference to CR or
2715 LF". That isn't really right - the whole point of the change was to try to
2716 help when there was an actual mention of one of the two characters. So now
2717 the change happens only if \r or \n (or a literal CR or LF) character is
2718 encountered.
2719
2720 2. The 32-bit options word was also used for 6 internal flags, but the numbers
2721 of both had grown to the point where there were only 3 bits left.
2722 Fortunately, there was spare space in the data structure, and so I have
2723 moved the internal flags into a new 16-bit field to free up more option
2724 bits.
2725
2726 3. The appearance of (?J) at the start of a pattern set the DUPNAMES option,
2727 but did not set the internal JCHANGED flag - either of these is enough to
2728 control the way the "get" function works - but the PCRE_INFO_JCHANGED
2729 facility is supposed to tell if (?J) was ever used, so now (?J) at the
2730 start sets both bits.
2731
2732 4. Added options (at build time, compile time, exec time) to change \R from
2733 matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
2734
2735 5. doc/pcresyntax.html was missing from the distribution.
2736
2737 6. Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
2738 compatibility, even though it is no longer used.
2739
2740 7. Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
2741 strtoull to pcrecpp.cc to select the available functions in WIN32 when the
2742 windows.h file is present (where different names are used). [This was
2743 reversed later after testing - see 16 below.]
2744
2745 8. Changed all #include <config.h> to #include "config.h". There were also
2746 some further <pcre.h> cases that I changed to "pcre.h".
2747
2748 9. When pcregrep was used with the --colour option, it missed the line ending
2749 sequence off the lines that it output.
2750
2751 10. It was pointed out to me that arrays of string pointers cause lots of
2752 relocations when a shared library is dynamically loaded. A technique of
2753 using a single long string with a table of offsets can drastically reduce
2754 these. I have refactored PCRE in four places to do this. The result is
2755 dramatic:
2756
2757 Originally: 290
2758 After changing UCP table: 187
2759 After changing error message table: 43
2760 After changing table of "verbs" 36
2761 After changing table of Posix names 22
2762
2763 Thanks to the folks working on Gregex for glib for this insight.
2764
2765 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
2766 unicode-properties was also set.
2767
2768 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
2769
2770 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
2771 checked only for CRLF.
2772
2773 14. Added casts to pcretest.c to avoid compiler warnings.
2774
2775 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
2776
2777 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
2778 and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
2779 entirely. This removes changes made in 7 above.
2780
2781 17. The CMake files have been updated, and there is now more information about
2782 building with CMake in the NON-UNIX-USE document.
2783
2784
2785 Version 7.3 28-Aug-07
2786 ---------------------
2787
2788 1. In the rejigging of the build system that eventually resulted in 7.1, the
2789 line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
2790 brackets there is not right, since it causes compilers to look for an
2791 installed pcre.h, not the version that is in the source that is being
2792 compiled (which of course may be different). I have changed it back to:
2793
2794 #include "pcre.h"
2795
2796 I have a vague recollection that the change was concerned with compiling in
2797 different directories, but in the new build system, that is taken care of
2798 by the VPATH setting the Makefile.
2799
2800 2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
2801 when the subject happened to end in the byte 0x85 (e.g. if the last
2802 character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
2803 characters but of course it shouldn't be taken as a newline when it is part
2804 of another character. The bug was that, for an unlimited repeat of . in
2805 not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
2806 characters when looking for a newline.
2807
2808 3. A small performance improvement in the DOTALL UTF-8 mode .* case.
2809
2810 4. Debugging: adjusted the names of opcodes for different kinds of parentheses
2811 in debug output.
2812
2813 5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
2814 long printing in the pcrecpp unittest when running under MinGW.
2815
2816 6. ESC_K was left out of the EBCDIC table.
2817
2818 7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
2819 parentheses; I made it 1000, which seemed large enough. Unfortunately, the
2820 limit also applies to "virtual nesting" when a pattern is recursive, and in
2821 this case 1000 isn't so big. I have been able to remove this limit at the
2822 expense of backing off one optimization in certain circumstances. Normally,
2823 when pcre_exec() would call its internal match() function recursively and
2824 immediately return the result unconditionally, it uses a "tail recursion"
2825 feature to save stack. However, when a subpattern that can match an empty
2826 string has an unlimited repetition quantifier, it no longer makes this
2827 optimization. That gives it a stack frame in which to save the data for
2828 checking that an empty string has been matched. Previously this was taken
2829 from the 1000-entry workspace that had been reserved. So now there is no
2830 explicit limit, but more stack is used.
2831
2832 8. Applied Daniel's patches to solve problems with the import/export magic
2833 syntax that is required for Windows, and which was going wrong for the
2834 pcreposix and pcrecpp parts of the library. These were overlooked when this
2835 problem was solved for the main library.
2836
2837 9. There were some crude static tests to avoid integer overflow when computing
2838 the size of patterns that contain repeated groups with explicit upper
2839 limits. As the maximum quantifier is 65535, the maximum group length was
2840 set at 30,000 so that the product of these two numbers did not overflow a
2841 32-bit integer. However, it turns out that people want to use groups that
2842 are longer than 30,000 bytes (though not repeat them that many times).
2843 Change 7.0/17 (the refactoring of the way the pattern size is computed) has
2844 made it possible to implement the integer overflow checks in a much more
2845 dynamic way, which I have now done. The artificial limitation on group
2846 length has been removed - we now have only the limit on the total length of
2847 the compiled pattern, which depends on the LINK_SIZE setting.
2848
2849 10. Fixed a bug in the documentation for get/copy named substring when
2850 duplicate names are permitted. If none of the named substrings are set, the
2851 functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
2852 empty string.
2853
2854 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
2855 instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
2856 because the ] is interpreted as the first data character and the
2857 terminating ] is not found. PCRE has been made compatible with Perl in this
2858 regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
2859 cause memory overwriting.
2860
2861 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
2862 string has been matched (to stop an infinite loop). It was not recognizing
2863 a conditional subpattern that could match an empty string if that
2864 subpattern was within another subpattern. For example, it looped when
2865 trying to match (((?(1)X|))*) but it was OK with ((?(1)X|)*) where the
2866 condition was not nested. This bug has been fixed.
2867
2868 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
2869 past the start of the subject in the presence of bytes with the top bit
2870 set, for example "\x8aBCD".
2871
2872 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
2873 (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
2874
2875 14. Optimized (?!) to (*FAIL).
2876
2877 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
2878 This restricts code points to be within the range 0 to 0x10FFFF, excluding
2879 the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
2880 full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
2881 does: it's just the validity check that is more restrictive.
2882
2883 16. Inserted checks for integer overflows during escape sequence (backslash)
2884 processing, and also fixed erroneous offset values for syntax errors during
2885 backslash processing.
2886
2887 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
2888 for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
2889
2890 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
2891 caused an overrun.
2892
2893 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
2894 something other than just ASCII characters) inside a group that had an
2895 unlimited repeat caused a loop at compile time (while checking to see
2896 whether the group could match an empty string).
2897
2898 20. Debugging a pattern containing \p or \P could cause a crash. For example,
2899 [\P{Any}] did so. (Error in the code for printing property names.)
2900
2901 21. An orphan \E inside a character class could cause a crash.
2902
2903 22. A repeated capturing bracket such as (A)? could cause a wild memory
2904 reference during compilation.
2905
2906 23. There are several functions in pcre_compile() that scan along a compiled
2907 expression for various reasons (e.g. to see if it's fixed length for look
2908 behind). There were bugs in these functions when a repeated \p or \P was
2909 present in the pattern. These operators have additional parameters compared
2910 with \d, etc, and these were not being taken into account when moving along
2911 the compiled data. Specifically:
2912
2913 (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
2914 length.
2915
2916 (b) An item such as \pL+ within a repeated group could cause crashes or
2917 loops.
2918
2919 (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
2920 "reference to non-existent subpattern" error.
2921
2922 (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
2923
2924 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
2925 characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
2926
2927 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
2928
2929 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
2930 character were causing crashes (broken optimization).
2931
2932 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
2933 \p or \P) caused a compile-time loop.
2934
2935 28. More problems have arisen in unanchored patterns when CRLF is a valid line
2936 break. For example, the unstudied pattern [\r\n]A does not match the string
2937 "\r\nA" because change 7.0/46 below moves the current point on by two
2938 characters after failing to match at the start. However, the pattern \nA
2939 *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
2940 the same is true. There doesn't seem any very clean way out of this, but
2941 what I have chosen to do makes the common cases work: PCRE now takes note
2942 of whether there can be an explicit match for \r or \n anywhere in the
2943 pattern, and if so, 7.0/46 no longer applies. As part of this change,
2944 there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
2945 pattern has explicit CR or LF references.
2946
2947 29. Added (*CR) etc for changing newline setting at start of pattern.
2948
2949
2950 Version 7.2 19-Jun-07
2951 ---------------------
2952
2953 1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
2954 which is apparently normally available under Windows.
2955
2956 2. Re-jig the pcregrep tests with different newline settings in an attempt
2957 to make them independent of the local environment's newline setting.
2958
2959 3. Add code to configure.ac to remove -g from the CFLAGS default settings.
2960
2961 4. Some of the "internals" tests were previously cut out when the link size
2962 was not 2, because the output contained actual offsets. The recent new
2963 "Z" feature of pcretest means that these can be cut out, making the tests
2964 usable with all link sizes.
2965
2966 5. Implemented Stan Switzer's goto replacement for longjmp() when not using
2967 stack recursion. This gives a massive performance boost under BSD, but just
2968 a small improvement under Linux. However, it saves one field in the frame
2969 in all cases.
2970
2971 6. Added more features from the forthcoming Perl 5.10:
2972
2973 (a) (?-n) (where n is a string of digits) is a relative subroutine or
2974 recursion call. It refers to the nth most recently opened parentheses.
2975
2976 (b) (?+n) is also a relative subroutine call; it refers to the nth next
2977 to be opened parentheses.
2978
2979 (c) Conditions that refer to capturing parentheses can be specified
2980 relatively, for example, (?(-2)... or (?(+3)...
2981
2982 (d) \K resets the start of the current match so that everything before
2983 is not part of it.
2984
2985 (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
2986
2987 (f) \g{name} is another synonym - part of Perl 5.10's unification of
2988 reference syntax.
2989
2990 (g) (?| introduces a group in which the numbering of parentheses in each
2991 alternative starts with the same number.
2992
2993 (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
2994
2995 7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
2996 PCRE_INFO_JCHANGED.
2997
2998 8. A pattern such as (.*(.)?)* caused pcre_exec() to fail by either not
2999 terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
3000 for detecting groups that can match an empty string.
3001
3002 9. A pattern with a very large number of alternatives (more than several
3003 hundred) was running out of internal workspace during the pre-compile
3004 phase, where pcre_compile() figures out how much memory will be needed. A
3005 bit of new cunning has reduced the workspace needed for groups with
3006 alternatives. The 1000-alternative test pattern now uses 12 bytes of
3007 workspace instead of running out of the 4096 that are available.
3008
3009 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
3010
3011 11. Applied patch from Google to remove an optimization that didn't quite work.
3012 The report of the bug said:
3013
3014 pcrecpp::RE("a*").FullMatch("aaa") matches, while
3015 pcrecpp::RE("a*?").FullMatch("aaa") does not, and
3016 pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
3017
3018 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
3019 it matched the wrong number of bytes.
3020
3021
3022 Version 7.1 24-Apr-07
3023 ---------------------
3024
3025 1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
3026 that is more "standard", making use of automake and other Autotools. There
3027 is some re-arrangement of the files and adjustment of comments consequent
3028 on this.
3029
3030 2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
3031 for recursive directory scanning broke on some systems because the files
3032 are not scanned in any specific order and on different systems the order
3033 was different. A call to "sort" has been inserted into RunGrepTest for the
3034 approprate test as a short-term fix. In the longer term there may be an
3035 alternative.
3036
3037 3. I had an email from Eric Raymond about problems translating some of PCRE's
3038 man pages to HTML (despite the fact that I distribute HTML pages, some
3039 people do their own conversions for various reasons). The problems
3040 concerned the use of low-level troff macros .br and .in. I have therefore
3041 removed all such uses from the man pages (some were redundant, some could
3042 be replaced by .nf/.fi pairs). The 132html script that I use to generate
3043 HTML has been updated to handle .nf/.fi and to complain if it encounters
3044 .br or .in.
3045
3046 4. Updated comments in configure.ac that get placed in config.h.in and also
3047 arranged for config.h to be included in the distribution, with the name
3048 config.h.generic, for the benefit of those who have to compile without
3049 Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
3050
3051 5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
3052 Weber: (1) pcre_internal.h was missing some function renames; (2) updated
3053 makevp.bat for the current PCRE, using the additional files
3054 makevp_c.txt, makevp_l.txt, and pcregexp.pas.
3055
3056 6. A Windows user reported a minor discrepancy with test 2, which turned out
3057 to be caused by a trailing space on an input line that had got lost in his
3058 copy. The trailing space was an accident, so I've just removed it.
3059
3060 7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
3061 that is needed.
3062
3063 8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
3064 as "const" (a) because they are and (b) because it helps the PHP
3065 maintainers who have recently made a script to detect big data structures
3066 in the php code that should be moved to the .rodata section. I remembered
3067 to update Builducptable as well, so it won't revert if ucptable.h is ever
3068 re-created.
3069
3070 9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
3071 pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
3072 order to be able to cut out the UTF-8 tables in the latter when UTF-8
3073 support is not required. This saves 1.5-2K of code, which is important in
3074 some applications.
3075
3076 Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
3077 so as not to refer to the tables, even though these functions will never be
3078 called when UTF-8 support is disabled. Otherwise there are problems with a
3079 shared library.
3080
3081 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
3082
3083 (a) It was defining its arguments as char * instead of void *.
3084
3085 (b) It was assuming that all moves were upwards in memory; this was true
3086 a long time ago when I wrote it, but is no longer the case.
3087
3088 The emulated memove() is provided for those environments that have neither
3089 memmove() nor bcopy(). I didn't think anyone used it these days, but that
3090 is clearly not the case, as these two bugs were recently reported.
3091
3092 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
3093 and Detrail to create the HTML documentation, the .txt form of the man
3094 pages, and it removes trailing spaces from listed files. It also creates
3095 pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
3096 case, it wraps all the #defines with #ifndefs. This script should be run
3097 before "make dist".
3098
3099 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
3100 with Unicode property support.
3101
3102 (a) For a maximizing quantifier, if the two different cases of the
3103 character were of different lengths in their UTF-8 codings (there are
3104 some cases like this - I found 11), and the matching function had to
3105 back up over a mixture of the two cases, it incorrectly assumed they
3106 were both the same length.
3107
3108 (b) When PCRE was configured to use the heap rather than the stack for
3109 recursion during matching, it was not correctly preserving the data for
3110 the other case of a UTF-8 character when checking ahead for a match
3111 while processing a minimizing repeat. If the check also involved
3112 matching a wide character, but failed, corruption could cause an
3113 erroneous result when trying to check for a repeat of the original
3114 character.
3115
3116 13. Some tidying changes to the testing mechanism:
3117
3118 (a) The RunTest script now detects the internal link size and whether there
3119 is UTF-8 and UCP support by running ./pcretest -C instead of relying on
3120 values substituted by "configure". (The RunGrepTest script already did
3121 this for UTF-8.) The configure.ac script no longer substitutes the
3122 relevant variables.
3123
3124 (b) The debugging options /B and /D in pcretest show the compiled bytecode
3125 with length and offset values. This means that the output is different
3126 for different internal link sizes. Test 2 is skipped for link sizes
3127 other than 2 because of this, bypassing the problem. Unfortunately,
3128 there was also a test in test 3 (the locale tests) that used /B and
3129 failed for link sizes other than 2. Rather than cut the whole test out,
3130 I have added a new /Z option to pcretest that replaces the length and
3131 offset values with spaces. This is now used to make test 3 independent
3132 of link size. (Test 2 will be tidied up later.)
3133
3134 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
3135 segmentation fault instead of returning the appropriate error message.
3136
3137 15. In multiline mode when the newline sequence was set to "any", the pattern
3138 ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
3139 This doesn't seem right; it now treats the CRLF combination as the line
3140 ending, and so does not match in that case. It's only a pattern such as ^$
3141 that would hit this one: something like ^ABC$ would have failed after \r
3142 and then tried again after \r\n.
3143
3144 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
3145 in an attempt to make files that differ only in their line terminators
3146 compare equal. This works on Linux.
3147
3148 17. Under certain error circumstances pcregrep might try to free random memory
3149 as it exited. This is now fixed, thanks to valgrind.
3150
3151 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
3152 "abc\r\n\r\n", it found an unwanted second match after the second \r. This
3153 was because its rules for how to advance for /g after matching an empty
3154 string at the end of a line did not allow for this case. They now check for
3155 it specially.
3156
3157 20. pcretest is supposed to handle patterns and data of any length, by
3158 extending its buffers when necessary. It was getting this wrong when the
3159 buffer for a data line had to be extended.
3160
3161 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
3162 CRLF as a newline sequence.
3163
3164 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
3165 out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
3166 I have nevertheless tidied it up.
3167
3168 23. Added some casts to kill warnings from HP-UX ia64 compiler.
3169
3170 24. Added a man page for pcre-config.
3171
3172
3173 Version 7.0 19-Dec-06
3174 ---------------------
3175
3176 1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
3177 moving to gcc 4.1.1.
3178
3179 2. The -S option for pcretest uses setrlimit(); I had omitted to #include
3180 sys/time.h, which is documented as needed for this function. It doesn't
3181 seem to matter on Linux, but it showed up on some releases of OS X.
3182
3183 3. It seems that there are systems where bytes whose values are greater than
3184 127 match isprint() in the "C" locale. The "C" locale should be the
3185 default when a C program starts up. In most systems, only ASCII printing
3186 characters match isprint(). This difference caused the output from pcretest
3187 to vary, making some of the tests fail. I have changed pcretest so that:
3188
3189 (a) When it is outputting text in the compiled version of a pattern, bytes
3190 other than 32-126 are always shown as hex escapes.
3191
3192 (b) When it is outputting text that is a matched part of a subject string,
3193 it does the same, unless a different locale has been set for the match
3194 (using the /L modifier). In this case, it uses isprint() to decide.
3195
3196 4. Fixed a major bug that caused incorrect computation of the amount of memory
3197 required for a compiled pattern when options that changed within the
3198 pattern affected the logic of the preliminary scan that determines the
3199 length. The relevant options are -x, and -i in UTF-8 mode. The result was
3200 that the computed length was too small. The symptoms of this bug were
3201 either the PCRE error "internal error: code overflow" from pcre_compile(),
3202 or a glibc crash with a message such as "pcretest: free(): invalid next
3203 size (fast)". Examples of patterns that provoked this bug (shown in
3204 pcretest format) are:
3205
3206 /(?-x: )/x
3207 /(?x)(?-x: \s*#\s*)/
3208 /((?i)[\x{c0}])/8
3209 /(?i:[\x{c0}])/8
3210
3211 HOWEVER: Change 17 below makes this fix obsolete as the memory computation
3212 is now done differently.
3213
3214 5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
3215 wrapper classes; (b) implement a new function in the C++ scanner that is
3216 more efficient than the old way of doing things because it avoids levels of
3217 recursion in the regex matching; (c) add a paragraph to the documentation
3218 for the FullMatch() function.
3219
3220 6. The escape sequence \n was being treated as whatever was defined as
3221 "newline". Not only was this contrary to the documentation, which states
3222 that \n is character 10 (hex 0A), but it also went horribly wrong when
3223 "newline" was defined as CRLF. This has been fixed.
3224
3225 7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
3226 was being set to -1 for the "end of line" case (supposedly a value that no
3227 character can have). Though this value is never used (the check for end of
3228 line is "zero bytes in current character"), it caused compiler complaints.
3229 I've changed it to 0xffffffff.
3230
3231 8. In pcre_version.c, the version string was being built by a sequence of
3232 C macros that, in the event of PCRE_PRERELEASE being defined as an empty
3233 string (as it is for production releases) called a macro with an empty
3234 argument. The C standard says the result of this is undefined. The gcc
3235 compiler treats it as an empty string (which was what was wanted) but it is
3236 reported that Visual C gives an error. The source has been hacked around to
3237 avoid this problem.
3238
3239 9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
3240 builds of pcretest, and changed the call to _setmode() to use _O_BINARY
3241 instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
3242 of them did).
3243
3244 10. Originally, pcretest opened its input and output without "b"; then I was
3245 told that "b" was needed in some environments, so it was added for release
3246 5.0 to both the input and output. (It makes no difference on Unix-like
3247 systems.) Later I was told that it is wrong for the input on Windows. I've
3248 now abstracted the modes into two macros, to make it easier to fiddle with
3249 them, and removed "b" from the input mode under Windows.
3250
3251 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
3252
3253 12. Added -help and --help to pcretest as an official way of being reminded
3254 of the options.
3255
3256 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
3257 and pcrecpp.cc because they annoy compilers at high warning levels.
3258
3259 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
3260
3261 15. Fixed an occurrence of == in configure.ac that should have been = (shell
3262 scripts are not C programs :-) and which was not noticed because it works
3263 on Linux.
3264
3265 16. pcretest is supposed to handle any length of pattern and data line (as one
3266 line or as a continued sequence of lines) by extending its input buffer if
3267 necessary. This feature was broken for very long pattern lines, leading to
3268 a string of junk being passed to pcre_compile() if the pattern was longer
3269 than about 50K.
3270
3271 17. I have done a major re-factoring of the way pcre_compile() computes the
3272 amount of memory needed for a compiled pattern. Previously, there was code
3273 that made a preliminary scan of the pattern in order to do this. That was
3274 OK when PCRE was new, but as the facilities have expanded, it has become
3275 harder and harder to keep it in step with the real compile phase, and there
3276 have been a number of bugs (see for example, 4 above). I have now found a
3277 cunning way of running the real compile function in a "fake" mode that
3278 enables it to compute how much memory it would need, while actually only
3279 ever using a few hundred bytes of working memory and without too many
3280 tests of the mode. This should make future maintenance and development
3281 easier. A side effect of this work is that the limit of 200 on the nesting
3282 depth of parentheses has been removed (though this was never a serious
3283 limitation, I suspect). However, there is a downside: pcre_compile() now
3284 runs more slowly than before (30% or more, depending on the pattern). I
3285 hope this isn't a big issue. There is no effect on runtime performance.
3286
3287 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
3288 newline (only possible for the last line of a file) and it was a
3289 pattern that set a locale (followed by /Lsomething), pcretest crashed.
3290
3291 19. Added additional timing features to pcretest. (1) The -tm option now times
3292 matching only, not compiling. (2) Both -t and -tm can be followed, as a
3293 separate command line item, by a number that specifies the number of
3294 repeats to use when timing. The default is 50000; this gives better
3295 precision, but takes uncomfortably long for very large patterns.
3296
3297 20. Extended pcre_study() to be more clever in cases where a branch of a
3298 subpattern has no definite first character. For example, (a*|b*)[cd] would
3299 previously give no result from pcre_study(). Now it recognizes that the
3300 first character must be a, b, c, or d.
3301
3302 21. There was an incorrect error "recursive call could loop indefinitely" if
3303 a subpattern (or the entire pattern) that was being tested for matching an
3304 empty string contained only one non-empty item after a nested subpattern.
3305 For example, the pattern (?>\x{100}*)\d(?R) provoked this error
3306 incorrectly, because the \d was being skipped in the check.
3307
3308 22. The pcretest program now has a new pattern option /B and a command line
3309 option -b, which is equivalent to adding /B to every pattern. This causes
3310 it to show the compiled bytecode, without the additional information that
3311 -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
3312 is the same as /B/I).
3313
3314 23. A new optimization is now able automatically to treat some sequences such
3315 as a*b as a*+b. More specifically, if something simple (such as a character
3316 or a simple class like \d) has an unlimited quantifier, and is followed by
3317 something that cannot possibly match the quantified thing, the quantifier
3318 is automatically "possessified".
3319
3320 24. A recursive reference to a subpattern whose number was greater than 39
3321 went wrong under certain circumstances in UTF-8 mode. This bug could also
3322 have affected the operation of pcre_study().
3323
3324 25. Realized that a little bit of performance could be had by replacing
3325 (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
3326
3327 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
3328
3329 27. Possessive quantifiers such as a++ were previously implemented by turning
3330 them into atomic groups such as ($>a+). Now they have their own opcodes,
3331 which improves performance. This includes the automatically created ones
3332 from 23 above.
3333
3334 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
3335 lookahead was broken if it was not anchored. PCRE was mistakenly expecting
3336 the first matched character to be a colon. This applied both to named and
3337 numbered groups.
3338
3339 29. The ucpinternal.h header file was missing its idempotency #ifdef.
3340
3341 30. I was sent a "project" file called libpcre.a.dev which I understand makes
3342 building PCRE on Windows easier, so I have included it in the distribution.
3343
3344 31. There is now a check in pcretest against a ridiculously large number being
3345 returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
3346 loop, the loop is abandoned.
3347
3348 32. Forward references to subpatterns in conditions such as (?(2)...) where
3349 subpattern 2 is defined later cause pcre_compile() to search forwards in
3350 the pattern for the relevant set of parentheses. This search went wrong
3351 when there were unescaped parentheses in a character class, parentheses
3352 escaped with \Q...\E, or parentheses in a #-comment in /x mode.
3353
3354 33. "Subroutine" calls and backreferences were previously restricted to
3355 referencing subpatterns earlier in the regex. This restriction has now
3356 been removed.
3357
3358 34. Added a number of extra features that are going to be in Perl 5.10. On the
3359 whole, these are just syntactic alternatives for features that PCRE had
3360 previously implemented using the Python syntax or my own invention. The
3361 other formats are all retained for compatibility.
3362
3363 (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
3364 as (?P<name>...). The new forms, as well as being in Perl 5.10, are
3365 also .NET compatible.
3366
3367 (b) A recursion or subroutine call to a named group can now be defined as
3368 (?&name) as well as (?P>name).
3369
3370 (c) A backreference to a named group can now be defined as \k<name> or
3371 \k'name' as well as (?P=name). The new forms, as well as being in Perl
3372 5.10, are also .NET compatible.
3373
3374 (d) A conditional reference to a named group can now use the syntax
3375 (?(<name>) or (?('name') as well as (?(name).
3376
3377 (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
3378 groups (named and numbered) that are never evaluated inline, but can be
3379 called as "subroutines" from elsewhere. In effect, the DEFINE condition
3380 is always false. There may be only one alternative in such a group.
3381
3382 (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
3383 as the simple (?(R). The condition is true only if the most recent
3384 recursion is that of the given number or name. It does not search out
3385 through the entire recursion stack.
3386
3387 (g) The escape \gN or \g{N} has been added, where N is a positive or
3388 negative number, specifying an absolute or relative reference.
3389
3390 35. Tidied to get rid of some further signed/unsigned compiler warnings and
3391 some "unreachable code" warnings.
3392
3393 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
3394 things, this adds five new scripts.
3395
3396 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
3397 There were also incompatibilities regarding the handling of \Q..\E inside
3398 character classes, for example with patterns like [\Qa\E-\Qz\E] where the
3399 hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
3400
3401 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
3402 matches an empty string, and forcibly breaks the loop. There were bugs in
3403 this code in non-simple cases. For a pattern such as ^(a()*)* matched
3404 against aaaa the result was just "a" rather than "aaaa", for example. Two
3405 separate and independent bugs (that affected different cases) have been
3406 fixed.
3407
3408 39. Refactored the code to abolish the use of different opcodes for small
3409 capturing bracket numbers. This is a tidy that I avoided doing when I
3410 removed the limit on the number of capturing brackets for 3.5 back in 2001.
3411 The new approach is not only tidier, it makes it possible to reduce the
3412 memory needed to fix the previous bug (38).
3413
3414 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
3415 sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
3416 processing dot, circumflex, or dollar metacharacters, or #-comments in /x
3417 mode.
3418
3419 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
3420 report.
3421
3422 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
3423 copy construction and assignment in the C++ wrapper.
3424
3425 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
3426 couple of bugs that could have given wrong results in the "--newline=crlf"
3427 case.
3428
3429 44. Added a number of casts and did some reorganization of signed/unsigned int
3430 variables following suggestions from Dair Grant. Also renamed the variable
3431 "this" as "item" because it is a C++ keyword.
3432
3433 45. Arranged for dftables to add
3434
3435 #include "pcre_internal.h"
3436
3437 to pcre_chartables.c because without it, gcc 4.x may remove the array
3438 definition from the final binary if PCRE is built into a static library and
3439 dead code stripping is activated.
3440
3441 46. For an unanchored pattern, if a match attempt fails at the start of a
3442 newline sequence, and the newline setting is CRLF or ANY, and the next two
3443 characters are CRLF, advance by two characters instead of one.
3444
3445
3446 Version 6.7 04-Jul-06
3447 ---------------------
3448
3449 1. In order to handle tests when input lines are enormously long, pcretest has
3450 been re-factored so that it automatically extends its buffers when
3451 necessary. The code is crude, but this _is_ just a test program. The
3452 default size has been increased from 32K to 50K.
3453
3454 2. The code in pcre_study() was using the value of the re argument before
3455 testing it for NULL. (Of course, in any sensible call of the function, it
3456 won't be NULL.)
3457
3458 3. The memmove() emulation function in pcre_internal.h, which is used on
3459 systems that lack both memmove() and bcopy() - that is, hardly ever -
3460 was missing a "static" storage class specifier.
3461
3462 4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
3463 containing an extended class (one that cannot be represented by a bitmap
3464 because it contains high-valued characters or Unicode property items, e.g.
3465 [\pZ]). Almost always one would set UTF-8 mode when processing such a
3466 pattern, but PCRE should not loop if you do not (it no longer does).
3467 [Detail: two cases were found: (a) a repeated subpattern containing an
3468 extended class; (b) a recursive reference to a subpattern that followed a
3469 previous extended class. It wasn't skipping over the extended class
3470 correctly when UTF-8 mode was not set.]
3471
3472 5. A negated single-character class was not being recognized as fixed-length
3473 in lookbehind assertions such as (?<=[^f]), leading to an incorrect
3474 compile error "lookbehind assertion is not fixed length".
3475
3476 6. The RunPerlTest auxiliary script was showing an unexpected difference
3477 between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
3478 write a Perl script that can interpret lines of an input file either as
3479 byte characters or as UTF-8, which is what "perltest" was being required to
3480 do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
3481 can't do is switch easily at run time between having the "use utf8;" pragma
3482 or not. In the end, I fudged it by using the RunPerlTest script to insert
3483 "use utf8;" explicitly for the UTF-8 tests.
3484
3485 7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
3486 the end of the subject string, contrary to the documentation and to what
3487 Perl does. This was true of both matching functions. Now it matches only at
3488 the start of the subject and immediately after *internal* newlines.
3489
3490 8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
3491 a pointer to an int instead of a pointer to an unsigned long int. This
3492 caused problems on 64-bit systems.
3493
3494 9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
3495 instance of the 'standard' template library not being so standard".
3496
3497 10. There was no check on the number of named subpatterns nor the maximum
3498 length of a subpattern name. The product of these values is used to compute
3499 the size of the memory block for a compiled pattern. By supplying a very
3500 long subpattern name and a large number of named subpatterns, the size
3501 computation could be caused to overflow. This is now prevented by limiting
3502 the length of names to 32 characters, and the number of named subpatterns
3503 to 10,000.
3504
3505 11. Subpatterns that are repeated with specific counts have to be replicated in
3506 the compiled pattern. The size of memory for this was computed from the
3507 length of the subpattern and the repeat count. The latter is limited to
3508 65535, but there was no limit on the former, meaning that integer overflow
3509 could in principle occur. The compiled length of a repeated subpattern is
3510 now limited to 30,000 bytes in order to prevent this.
3511
3512 12. Added the optional facility to have named substrings with the same name.
3513
3514 13. Added the ability to use a named substring as a condition, using the
3515 Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
3516 are numbers (not recommended). Forward references are permitted.
3517
3518 14. Added forward references in named backreferences (if you see what I mean).
3519
3520 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
3521 pattern could run off the end of the subject. For example, the pattern
3522 "(?s)(.{1,5})"8 did this with the subject "ab".
3523
3524 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
3525 PCRE_CASELESS was set when matching characters that were quantified with ?
3526 or *.
3527
3528 17. A character class other than a single negated character that had a minimum
3529 but no maximum quantifier - for example [ab]{6,} - was not handled
3530 correctly by pce_dfa_exec(). It would match only one character.
3531
3532 18. A valid (though odd) pattern that looked like a POSIX character
3533 class but used an invalid character after [ (for example [[,abc,]]) caused
3534 pcre_compile() to give the error "Failed: internal error: code overflow" or
3535 in some cases to crash with a glibc free() error. This could even happen if
3536 the pattern terminated after [[ but there just happened to be a sequence of
3537 letters, a binary zero, and a closing ] in the memory that followed.
3538
3539 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
3540 over the years. Originally (before any Unicode support), just the bottom 8
3541 bits were taken. Thus, for example, \500 really meant \100. Nowadays the
3542 output from "man perlunicode" includes this:
3543
3544 The regular expression compiler produces polymorphic opcodes. That
3545 is, the pattern adapts to the data and automatically switches to
3546 the Unicode character scheme when presented with Unicode data--or
3547 instead uses a traditional byte scheme when presented with byte
3548 data.
3549
3550 Sadly, a wide octal escape does not cause a switch, and in a string with
3551 no other multibyte characters, these octal escapes are treated as before.
3552 Thus, in Perl, the pattern /\500/ actually matches \100 but the pattern
3553 /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
3554 Unicode string.
3555
3556 I have not perpetrated such confusion in PCRE. Up till now, it took just
3557 the bottom 8 bits, as in old Perl. I have now made octal escapes with
3558 values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
3559 translate to the appropriate multibyte character.
3560
3561 29. Applied some refactoring to reduce the number of warnings from Microsoft
3562 and Borland compilers. This has included removing the fudge introduced
3563 seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
3564 a warning about an unused variable.
3565
3566 21. PCRE has not included VT (character 0x0b) in the set of whitespace
3567 characters since release 4.0, because Perl (from release 5.004) does not.
3568 [Or at least, is documented not to: some releases seem to be in conflict
3569 with the documentation.] However, when a pattern was studied with
3570 pcre_study() and all its branches started with \s, PCRE still included VT
3571 as a possible starting character. Of course, this did no harm; it just
3572 caused an unnecessary match attempt.
3573
3574 22. Removed a now-redundant internal flag bit that recorded the fact that case
3575 dependency changed within the pattern. This was once needed for "required
3576 byte" processing, but is no longer used. This recovers a now-scarce options
3577 bit. Also moved the least significant internal flag bit to the most-
3578 significant bit of the word, which was not previously used (hangover from
3579 the days when it was an int rather than a uint) to free up another bit for
3580 the future.
3581
3582 23. Added support for CRLF line endings as well as CR and LF. As well as the
3583 default being selectable at build time, it can now be changed at runtime
3584 via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
3585 specify that it is scanning data with non-default line endings.
3586
3587 24. Changed the definition of CXXLINK to make it agree with the definition of
3588 LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
3589
3590 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
3591 recursions. This makes a big different to stack usage for some patterns.
3592
3593 26. If a subpattern containing a named recursion or subroutine reference such
3594 as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
3595 the space required for the compiled pattern went wrong and gave too small a
3596 value. Depending on the environment, this could lead to "Failed: internal
3597 error: code overflow at offset 49" or "glibc detected double free or
3598 corruption" errors.
3599
3600 27. Applied patches from Google (a) to support the new newline modes and (b) to
3601 advance over multibyte UTF-8 characters in GlobalReplace.
3602
3603 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
3604 difference for some implementation of PCRE in some Windows version.
3605
3606 29. Added some extra testing facilities to pcretest:
3607
3608 \q<number> in a data line sets the "match limit" value
3609 \Q<number> in a data line sets the "match recursion limt" value
3610 -S <number> sets the stack size, where <number> is in megabytes
3611
3612 The -S option isn't available for Windows.
3613
3614
3615 Version 6.6 06-Feb-06
3616 ---------------------
3617
3618 1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
3619 in pcreposix.h. I have copied the definition from pcre.h.
3620
3621 2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
3622 because pcre.h is no longer a built file.
3623
3624 3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
3625 not normally included in the compiled code.
3626
3627
3628 Version 6.5 01-Feb-06
3629 ---------------------
3630
3631 1. When using the partial match feature with pcre_dfa_exec(), it was not
3632 anchoring the second and subsequent partial matches at the new starting
3633 point. This could lead to incorrect results. For example, with the pattern
3634 /1234/, partially matching against "123" and then "a4" gave a match.
3635
3636 2. Changes to pcregrep:
3637
3638 (a) All non-match returns from pcre_exec() were being treated as failures
3639 to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
3640 error message is output. Some extra information is given for the
3641 PCRE_ERROR_MATCHLIMIT and PCRE_ERROR_RECURSIONLIMIT errors, which are
3642 probably the only errors that are likely to be caused by users (by
3643 specifying a regex that has nested indefinite repeats, for instance).
3644 If there are more than 20 of these errors, pcregrep is abandoned.
3645
3646 (b) A binary zero was treated as data while matching, but terminated the
3647 output line if it was written out. This has been fixed: binary zeroes
3648 are now no different to any other data bytes.
3649
3650 (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
3651 used to set a locale for matching. The --locale=xxxx long option has
3652 been added (no short equivalent) to specify a locale explicitly on the
3653 pcregrep command, overriding the environment variables.
3654
3655 (d) When -B was used with -n, some line numbers in the output were one less
3656 than they should have been.
3657
3658 (e) Added the -o (--only-matching) option.
3659
3660 (f) If -A or -C was used with -c (count only), some lines of context were
3661 accidentally printed for the final match.
3662
3663 (g) Added the -H (--with-filename) option.
3664
3665 (h) The combination of options -rh failed to suppress file names for files
3666 that were found from directory arguments.
3667
3668 (i) Added the -D (--devices) and -d (--directories) options.
3669
3670 (j) Added the -F (--fixed-strings) option.
3671
3672 (k) Allow "-" to be used as a file name for -f as well as for a data file.
3673
3674 (l) Added the --colo(u)r option.
3675
3676 (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
3677 is not present by default.
3678
3679 3. A nasty bug was discovered in the handling of recursive patterns, that is,
3680 items such as (?R) or (?1), when the recursion could match a number of
3681 alternatives. If it matched one of the alternatives, but subsequently,
3682 outside the recursion, there was a failure, the code tried to back up into
3683 the recursion. However, because of the way PCRE is implemented, this is not
3684 possible, and the result was an incorrect result from the match.
3685
3686 In order to prevent this happening, the specification of recursion has
3687 been changed so that all such subpatterns are automatically treated as
3688 atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
3689
3690 4. I had overlooked the fact that, in some locales, there are characters for
3691 which isalpha() is true but neither isupper() nor islower() are true. In
3692 the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
3693 and ordfeminine) are like this. This affected the treatment of \w and \W
3694 when they appeared in character classes, but not when they appeared outside
3695 a character class. The bit map for "word" characters is now created
3696 separately from the results of isalnum() instead of just taking it from the
3697 upper, lower, and digit maps. (Plus the underscore character, of course.)
3698
3699 5. The above bug also affected the handling of POSIX character classes such as
3700 [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
3701 permanent tables. Instead, the bit maps for such a class were previously
3702 created as the appropriate unions of the upper, lower, and digit bitmaps.
3703 Now they are created by subtraction from the [[:word:]] class, which has
3704 its own bitmap.
3705
3706 6. The [[:blank:]] character class matches horizontal, but not vertical space.
3707 It is created by subtracting the vertical space characters (\x09, \x0a,
3708 \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
3709 subtraction was done in the overall bitmap for a character class, meaning
3710 that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
3711 be recognized. This bug has been fixed.
3712
3713 7. Patches from the folks at Google:
3714
3715 (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
3716 real life, but is still worth protecting against".
3717
3718 (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
3719 regular expressions".
3720
3721 (c) pcre_scanner.cc: avoid use of std::count() because not all systems
3722 have it.
3723
3724 (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
3725 "configure" and the latter not, in order to fix a problem somebody had
3726 with compiling the Arg class on HP-UX.
3727
3728 (e) Improve the error-handling of the C++ wrapper a little bit.
3729
3730 (f) New tests for checking recursion limiting.
3731
3732 8. The pcre_memmove() function, which is used only if the environment does not
3733 have a standard memmove() function (and is therefore rarely compiled),
3734 contained two bugs: (a) use of int instead of size_t, and (b) it was not
3735 returning a result (though PCRE never actually uses the result).
3736
3737 9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
3738 large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
3739 returned instead of calling malloc() with an overflowing number that would
3740 most likely cause subsequent chaos.
3741
3742 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
3743
3744 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
3745 with this option is matched, the nmatch and pmatch options of regexec() are
3746 ignored.
3747
3748 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
3749 provided in case anyone wants to the the POSIX interface with UTF-8
3750 strings.
3751
3752 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
3753 C++ linking (needed for some HP-UX environments).
3754
3755 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
3756 (unused parameter) and in the pcre_printint() function (omitted "default"
3757 switch label when the default is to do nothing).
3758
3759 15. Added some code to make it possible, when PCRE is compiled as a C++
3760 library, to replace subject pointers for pcre_exec() with a smart pointer
3761 class, thus making it possible to process discontinuous strings.
3762
3763 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
3764 much the same function. They were added by different people who were trying
3765 to make PCRE easy to compile on non-Unix systems. It has been suggested
3766 that PCRE_EXPORT be abolished now that there is more automatic apparatus
3767 for compiling on Windows systems. I have therefore replaced it with
3768 PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
3769 defaults to "extern" for C or "extern C" for C++, which works fine on
3770 Unix-like systems. It is now possible to override the value of PCRE_DATA_
3771 SCOPE with something explicit in config.h. In addition:
3772
3773 (a) pcreposix.h still had just "extern" instead of either of these macros;
3774 I have replaced it with PCRE_DATA_SCOPE.
3775
3776 (b) Functions such as _pcre_xclass(), which are internal to the library,
3777 but external in the C sense, all had PCRE_EXPORT in their definitions.
3778 This is apparently wrong for the Windows case, so I have removed it.
3779 (It makes no difference on Unix-like systems.)
3780
3781 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
3782 of recursive calls to match(). This is different to MATCH_LIMIT because
3783 that limits the total number of calls to match(), not all of which increase
3784 the depth of recursion. Limiting the recursion depth limits the amount of
3785 stack (or heap if NO_RECURSE is set) that is used. The default can be set
3786 when PCRE is compiled, and changed at run time. A patch from Google adds
3787 this functionality to the C++ interface.
3788
3789 18. Changes to the handling of Unicode character properties:
3790
3791 (a) Updated the table to Unicode 4.1.0.
3792
3793 (b) Recognize characters that are not in the table as "Cn" (undefined).
3794
3795 (c) I revised the way the table is implemented to a much improved format
3796 which includes recognition of ranges. It now supports the ranges that
3797 are defined in UnicodeData.txt, and it also amalgamates other
3798 characters into ranges. This has reduced the number of entries in the
3799 table from around 16,000 to around 3,000, thus reducing its size
3800 considerably. I realized I did not need to use a tree structure after
3801 all - a binary chop search is just as efficient. Having reduced the
3802 number of entries, I extended their size from 6 bytes to 8 bytes to
3803 allow for more data.
3804
3805 (d) Added support for Unicode script names via properties such as \p{Han}.
3806
3807 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
3808 matching that character.
3809
3810 20. When matching a repeated Unicode property with a minimum greater than zero,
3811 (for example \pL{2,}), PCRE could look past the end of the subject if it
3812 reached it while seeking the minimum number of characters. This could
3813 happen only if some of the characters were more than one byte long, because
3814 there is a check for at least the minimum number of bytes.
3815
3816 21. Refactored the implementation of \p and \P so as to be more general, to
3817 allow for more different types of property in future. This has changed the
3818 compiled form incompatibly. Anybody with saved compiled patterns that use
3819 \p or \P will have to recompile them.
3820
3821 22. Added "Any" and "L&" to the supported property types.
3822
3823 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
3824 but give a compile time error if the value is greater than 0xff.
3825
3826 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
3827 accidentally not being installed or uninstalled.
3828
3829 25. The pcre.h file was built from pcre.h.in, but the only changes that were
3830 made were to insert the current release number. This seemed silly, because
3831 it made things harder for people building PCRE on systems that don't run
3832 "configure". I have turned pcre.h into a distributed file, no longer built
3833 by "configure", with the version identification directly included. There is
3834 no longer a pcre.h.in file.
3835
3836 However, this change necessitated a change to the pcre-config script as
3837 well. It is built from pcre-config.in, and one of the substitutions was the
3838 release number. I have updated configure.ac so that ./configure now finds
3839 the release number by grepping pcre.h.
3840
3841 26. Added the ability to run the tests under valgrind.
3842
3843
3844 Version 6.4 05-Sep-05
3845 ---------------------
3846
3847 1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
3848 "--" to be printed when multiple files were scanned, even when none of the
3849 -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
3850 consider it to be a bug, and have restored the previous behaviour.
3851
3852 2. A couple of code tidies to get rid of compiler warnings.
3853
3854 3. The pcretest program used to cheat by referring to symbols in the library
3855 whose names begin with _pcre_. These are internal symbols that are not
3856 really supposed to be visible externally, and in some environments it is
3857 possible to suppress them. The cheating is now confined to including
3858 certain files from the library's source, which is a bit cleaner.
3859
3860 4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
3861 file's purpose clearer.
3862
3863 5. Reorganized pcre_ucp_findchar().
3864
3865
3866 Version 6.3 15-Aug-05
3867 ---------------------
3868
3869 1. The file libpcre.pc.in did not have general read permission in the tarball.
3870
3871 2. There were some problems when building without C++ support:
3872
3873 (a) If C++ support was not built, "make install" and "make test" still
3874 tried to test it.
3875
3876 (b) There were problems when the value of CXX was explicitly set. Some
3877 changes have been made to try to fix these, and ...
3878
3879 (c) --disable-cpp can now be used to explicitly disable C++ support.
3880
3881 (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
3882 backslash in a target when C++ was disabled. This confuses some
3883 versions of "make", apparently. Using an intermediate variable solves
3884 this. (Same for CPP_LOBJ.)
3885
3886 3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
3887 (non-Windows) now includes $(CFLAGS) because these flags are sometimes
3888 necessary on certain architectures.
3889
3890 4. Added a setting of -export-symbols-regex to the link command to remove
3891 those symbols that are exported in the C sense, but actually are local
3892 within the library, and not documented. Their names all begin with
3893 "_pcre_". This is not a perfect job, because (a) we have to except some
3894 symbols that pcretest ("illegally") uses, and (b) the facility isn't always
3895 available (and never for static libraries). I have made a note to try to
3896 find a way round (a) in the future.
3897
3898
3899 Version 6.2 01-Aug-05
3900 ---------------------
3901
3902 1. There was no test for integer overflow of quantifier values. A construction
3903 such as {1111111111111111} would give undefined results. What is worse, if
3904 a minimum quantifier for a parenthesized subpattern overflowed and became
3905 negative, the calculation of the memory size went wrong. This could have
3906 led to memory overwriting.
3907
3908 2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
3909
3910 3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
3911 operating environments where this matters.
3912
3913 4. Applied Giuseppe Maxia's patch to add additional features for controlling
3914 PCRE options from within the C++ wrapper.
3915
3916 5. Named capturing subpatterns were not being correctly counted when a pattern
3917 was compiled. This caused two problems: (a) If there were more than 100
3918 such subpatterns, the calculation of the memory needed for the whole
3919 compiled pattern went wrong, leading to an overflow error. (b) Numerical
3920 back references of the form \12, where the number was greater than 9, were
3921 not recognized as back references, even though there were sufficient
3922 previous subpatterns.
3923
3924 6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
3925 versions of gcc, e.g. 2.95.4.
3926
3927
3928 Version 6.1 21-Jun-05
3929 ---------------------
3930
3931 1. There was one reference to the variable "posix" in pcretest.c that was not
3932 surrounded by "#if !defined NOPOSIX".
3933
3934 2. Make it possible to compile pcretest without DFA support, UTF8 support, or
3935 the cross-check on the old pcre_info() function, for the benefit of the
3936 cut-down version of PCRE that is currently imported into Exim.
3937
3938 3. A (silly) pattern starting with (?i)(?-i) caused an internal space
3939 allocation error. I've done the easy fix, which wastes 2 bytes for sensible
3940 patterns that start (?i) but I don't think that matters. The use of (?i) is
3941 just an example; this all applies to the other options as well.
3942
3943 4. Since libtool seems to echo the compile commands it is issuing, the output
3944 from "make" can be reduced a bit by putting "@" in front of each libtool
3945 compile command.
3946
3947 5. Patch from the folks at Google for configure.in to be a bit more thorough
3948 in checking for a suitable C++ installation before trying to compile the
3949 C++ stuff. This should fix a reported problem when a compiler was present,
3950 but no suitable headers.
3951
3952 6. The man pages all had just "PCRE" as their title. I have changed them to
3953 be the relevant file name. I have also arranged that these names are
3954 retained in the file doc/pcre.txt, which is a concatenation in text format
3955 of all the man pages except the little individual ones for each function.
3956
3957 7. The NON-UNIX-USE file had not been updated for the different set of source
3958 files that come with release 6. I also added a few comments about the C++
3959 wrapper.
3960
3961
3962 Version 6.0 07-Jun-05
3963 ---------------------
3964
3965 1. Some minor internal re-organization to help with my DFA experiments.
3966
3967 2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
3968 didn't matter for the library itself when fully configured, but did matter
3969 when compiling without UCP support, or within Exim, where the ucp files are
3970 not imported.
3971
3972 3. Refactoring of the library code to split up the various functions into
3973 different source modules. The addition of the new DFA matching code (see
3974 below) to a single monolithic source would have made it really too
3975 unwieldy, quite apart from causing all the code to be include in a
3976 statically linked application, when only some functions are used. This is
3977 relevant even without the DFA addition now that patterns can be compiled in
3978 one application and matched in another.
3979
3980 The downside of splitting up is that there have to be some external
3981 functions and data tables that are used internally in different modules of
3982 the library but which are not part of the API. These have all had their
3983 names changed to start with "_pcre_" so that they are unlikely to clash
3984 with other external names.
3985
3986 4. Added an alternate matching function, pcre_dfa_exec(), which matches using
3987 a different (DFA) algorithm. Although it is slower than the original
3988 function, it does have some advantages for certain types of matching
3989 problem.
3990
3991 5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
3992 including restarting after a partial match.
3993
3994 6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
3995 defined when compiling for Windows was sent to me. I have put it into the
3996 code, though I have no means of testing or verifying it.
3997
3998 7. Added the pcre_refcount() auxiliary function.
3999
4000 8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
4001 match before or at the first newline in the subject string. In pcretest,
4002 the /f option on a pattern can be used to set this.
4003
4004 9. A repeated \w when used in UTF-8 mode with characters greater than 256
4005 would behave wrongly. This has been present in PCRE since release 4.0.
4006
4007 10. A number of changes to the pcregrep command:
4008
4009 (a) Refactored how -x works; insert ^(...)$ instead of setting
4010 PCRE_ANCHORED and checking the length, in preparation for adding
4011 something similar for -w.
4012
4013 (b) Added the -w (match as a word) option.
4014
4015 (c) Refactored the way lines are read and buffered so as to have more
4016 than one at a time available.
4017
4018 (d) Implemented a pcregrep test script.
4019
4020 (e) Added the -M (multiline match) option. This allows patterns to match
4021 over several lines of the subject. The buffering ensures that at least
4022 8K, or the rest of the document (whichever is the shorter) is available
4023 for matching (and similarly the previous 8K for lookbehind assertions).
4024
4025 (f) Changed the --help output so that it now says
4026
4027 -w, --word-regex(p)
4028
4029 instead of two lines, one with "regex" and the other with "regexp"
4030 because that confused at least one person since the short forms are the
4031 same. (This required a bit of code, as the output is generated
4032 automatically from a table. It wasn't just a text change.)
4033
4034 (g) -- can be used to terminate pcregrep options if the next thing isn't an
4035 option but starts with a hyphen. Could be a pattern or a path name
4036 starting with a hyphen, for instance.
4037
4038 (h) "-" can be given as a file name to represent stdin.
4039
4040 (i) When file names are being printed, "(standard input)" is used for
4041 the standard input, for compatibility with GNU grep. Previously
4042 "<stdin>" was used.
4043
4044 (j) The option --label=xxx can be used to supply a name to be used for
4045 stdin when file names are being printed. There is no short form.
4046
4047 (k) Re-factored the options decoding logic because we are going to add
4048 two more options that take data. Such options can now be given in four
4049 different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
4050
4051 (l) Added the -A, -B, and -C options for requesting that lines of context
4052 around matches be printed.
4053
4054 (m) Added the -L option to print the names of files that do not contain
4055 any matching lines, that is, the complement of -l.
4056
4057 (n) The return code is 2 if any file cannot be opened, but pcregrep does
4058 continue to scan other files.
4059
4060 (o) The -s option was incorrectly implemented. For compatibility with other
4061 greps, it now suppresses the error message for a non-existent or non-
4062 accessible file (but not the return code). There is a new option called
4063 -q that suppresses the output of matching lines, which was what -s was
4064 previously doing.
4065
4066 (p) Added --include and --exclude options to specify files for inclusion
4067 and exclusion when recursing.
4068
4069 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
4070 Hopefully, it now does.
4071
4072 12. Missing cast in pcre_study().
4073
4074 13. Added an "uninstall" target to the makefile.
4075
4076 14. Replaced "extern" in the function prototypes in Makefile.in with
4077 "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
4078 world, but is set differently for Windows.
4079
4080 15. Added a second compiling function called pcre_compile2(). The only
4081 difference is that it has an extra argument, which is a pointer to an
4082 integer error code. When there is a compile-time failure, this is set
4083 non-zero, in addition to the error test pointer being set to point to an
4084 error message. The new argument may be NULL if no error number is required
4085 (but then you may as well call pcre_compile(), which is now just a
4086 wrapper). This facility is provided because some applications need a
4087 numeric error indication, but it has also enabled me to tidy up the way
4088 compile-time errors are handled in the POSIX wrapper.
4089
4090 16. Added VPATH=.libs to the makefile; this should help when building with one
4091 prefix path and installing with another. (Or so I'm told by someone who
4092 knows more about this stuff than I do.)
4093
4094 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
4095 passes PCRE_DOTALL to the pcre_compile() function, making the "." character
4096 match everything, including newlines. This is not POSIX-compatible, but
4097 somebody wanted the feature. From pcretest it can be activated by using
4098 both the P and the s flags.
4099
4100 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
4101
4102 19. libpcre.pc was being incorrectly installed as executable.
4103
4104 20. A couple of places in pcretest check for end-of-line by looking for '\n';
4105 it now also looks for '\r' so that it will work unmodified on Windows.
4106
4107 21. Added Google's contributed C++ wrapper to the distribution.
4108
4109 22. Added some untidy missing memory free() calls in pcretest, to keep
4110 Electric Fence happy when testing.
4111
4112
4113
4114 Version 5.0 13-Sep-04
4115 ---------------------
4116
4117 1. Internal change: literal characters are no longer packed up into items
4118 containing multiple characters in a single byte-string. Each character
4119 is now matched using a separate opcode. However, there may be more than one
4120 byte in the character in UTF-8 mode.
4121
4122 2. The pcre_callout_block structure has two new fields: pattern_position and
4123 next_item_length. These contain the offset in the pattern to the next match
4124 item, and its length, respectively.
4125
4126 3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
4127 insertion of callouts before each pattern item. Added the /C option to
4128 pcretest to make use of this.
4129
4130 4. On the advice of a Windows user, the lines
4131
4132 #if defined(_WIN32) || defined(WIN32)
4133 _setmode( _fileno( stdout ), 0x8000 );
4134 #endif /* defined(_WIN32) || defined(WIN32) */
4135
4136 have been added to the source of pcretest. This apparently does useful
4137 magic in relation to line terminators.
4138
4139 5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
4140 for the benefit of those environments where the "b" makes a difference.
4141
4142 6. The icc compiler has the same options as gcc, but "configure" doesn't seem
4143 to know about it. I have put a hack into configure.in that adds in code
4144 to set GCC=yes if CC=icc. This seems to end up at a point in the
4145 generated configure script that is early enough to affect the setting of
4146 compiler options, which is what is needed, but I have no means of testing
4147 whether it really works. (The user who reported this had patched the
4148 generated configure script, which of course I cannot do.)
4149
4150 LATER: After change 22 below (new libtool files), the configure script
4151 seems to know about icc (and also ecc). Therefore, I have commented out
4152 this hack in configure.in.
4153
4154 7. Added support for pkg-config (2 patches were sent in).
4155
4156 8. Negated POSIX character classes that used a combination of internal tables
4157 were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
4158 [[:^ascii]]. Typically, they would match almost any characters. The other
4159 POSIX classes were not broken in this way.
4160
4161 9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
4162 to find the match, as PCRE was deluded into thinking that the match had to
4163 start at the start point or following a newline. The same bug applied to
4164 patterns with negative forward assertions or any backward assertions
4165 preceding ".*" at the start, unless the pattern required a fixed first
4166 character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
4167
4168 10. In UTF-8 mode, when moving forwards in the subject after a failed match
4169 starting at the last subject character, bytes beyond the end of the subject
4170 string were read.
4171
4172 11. Renamed the variable "class" as "classbits" to make life easier for C++
4173 users. (Previously there was a macro definition, but it apparently wasn't
4174 enough.)
4175
4176 12. Added the new field "tables" to the extra data so that tables can be passed
4177 in at exec time, or the internal tables can be re-selected. This allows
4178 a compiled regex to be saved and re-used at a later time by a different
4179 program that might have everything at different addresses.
4180
4181 13. Modified the pcre-config script so that, when run on Solaris, it shows a
4182 -R library as well as a -L library.
4183
4184 14. The debugging options of pcretest (-d on the command line or D on a
4185 pattern) showed incorrect output for anything following an extended class
4186 that contained multibyte characters and which was followed by a quantifier.
4187
4188 15. Added optional support for general category Unicode character properties
4189 via the \p, \P, and \X escapes. Unicode property support implies UTF-8
4190 support. It adds about 90K to the size of the library. The meanings of the
4191 inbuilt class escapes such as \d and \s have NOT been changed.
4192
4193 16. Updated pcredemo.c to include calls to free() to release the memory for the
4194 compiled pattern.
4195
4196 17. The generated file chartables.c was being created in the source directory
4197 instead of in the building directory. This caused the build to fail if the
4198 source directory was different from the building directory, and was
4199 read-only.
4200
4201 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
4202 file. No doubt somebody will tell me if they don't make sense... Also added
4203 Dan Mooney's comments about building on OpenVMS.
4204
4205 19. Added support for partial matching via the PCRE_PARTIAL option for
4206 pcre_exec() and the \P data escape in pcretest.
4207
4208 20. Extended pcretest with 3 new pattern features:
4209
4210 (i) A pattern option of the form ">rest-of-line" causes pcretest to
4211 write the compiled pattern to the file whose name is "rest-of-line".
4212 This is a straight binary dump of the data, with the saved pointer to
4213 the character tables forced to be NULL. The study data, if any, is
4214 written too. After writing, pcretest reads a new pattern.
4215
4216 (ii) If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
4217 compiled pattern from the given file. There must not be any
4218 occurrences of "<" in the file name (pretty unlikely); if there are,
4219 pcretest will instead treat the initial "<" as a pattern delimiter.
4220 After reading in the pattern, pcretest goes on to read data lines as
4221 usual.
4222
4223 (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
4224 and 16-bit fields in a compiled pattern, to simulate a pattern that
4225 was compiled on a host of opposite endianness.
4226
4227 21. The pcre-exec() function can now cope with patterns that were compiled on
4228 hosts of opposite endianness, with this restriction:
4229
4230 As for any compiled expression that is saved and used later, the tables
4231 pointer field cannot be preserved; the extra_data field in the arguments
4232 to pcre_exec() should be used to pass in a tables address if a value
4233 other than the default internal tables were used at compile time.
4234
4235 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
4236 now diagnosed as an error. Previously, most of the time, a negative number
4237 would have been treated as zero, but if in addition "ovector" was passed as
4238 NULL, a crash could occur.
4239
4240 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
4241 new versions from the libtool 1.5 distribution (the last one is a copy of
4242 a file called libtool.m4). This seems to have fixed the need to patch
4243 "configure" to support Darwin 1.3 (which I used to do). However, I still
4244 had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
4245 workstation).
4246
4247 24. Changed the PCRE licence to be the more standard "BSD" licence.
4248
4249
4250 Version 4.5 01-Dec-03
4251 ---------------------
4252
4253 1. There has been some re-arrangement of the code for the match() function so
4254 that it can be compiled in a version that does not call itself recursively.
4255 Instead, it keeps those local variables that need separate instances for
4256 each "recursion" in a frame on the heap, and gets/frees frames whenever it
4257 needs to "recurse". Keeping track of where control must go is done by means
4258 of setjmp/longjmp. The whole thing is implemented by a set of macros that
4259 hide most of the details from the main code, and operates only if
4260 NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
4261 "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
4262 operating.
4263
4264 To make it easier for callers to provide specially tailored get/free
4265 functions for this usage, two new functions, pcre_stack_malloc, and
4266 pcre_stack_free, are used. They are always called in strict stacking order,
4267 and the size of block requested is always the same.
4268
4269 The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
4270 PCRE has been compiled to use the stack or the heap for recursion. The
4271 -C option of pcretest uses this to show which version is compiled.
4272
4273 A new data escape \S, is added to pcretest; it causes the amounts of store
4274 obtained and freed by both kinds of malloc/free at match time to be added
4275 to the output.
4276
4277 2. Changed the locale test to use "fr_FR" instead of "fr" because that's
4278 what's available on my current Linux desktop machine.
4279
4280 3. When matching a UTF-8 string, the test for a valid string at the start has
4281 been extended. If start_offset is not zero, PCRE now checks that it points
4282 to a byte that is the start of a UTF-8 character. If not, it returns
4283 PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
4284 this is necessary because there may be backward assertions in the pattern.
4285 When matching the same subject several times, it may save resources to use
4286 PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
4287
4288 4. The code for checking the validity of UTF-8 strings has been tightened so
4289 that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
4290 containing "overlong sequences".
4291
4292 5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
4293 I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
4294 should have been "&", but it just so happened that all the cases this let
4295 through by mistake were picked up later in the function.
4296
4297 6. I had used a variable called "isblank" - this is a C99 function, causing
4298 some compilers to warn. To avoid this, I renamed it (as "blankclass").
4299
4300 7. Cosmetic: (a) only output another newline at the end of pcretest if it is
4301 prompting; (b) run "./pcretest /dev/null" at the start of the test script
4302 so the version is shown; (c) stop "make test" echoing "./RunTest".
4303
4304 8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
4305
4306 9. The prototype for memmove() for systems that don't have it was using
4307 size_t, but the inclusion of the header that defines size_t was later. I've
4308 moved the #includes for the C headers earlier to avoid this.
4309
4310 10. Added some adjustments to the code to make it easier to compiler on certain
4311 special systems:
4312
4313 (a) Some "const" qualifiers were missing.
4314 (b) Added the macro EXPORT before all exported functions; by default this
4315 is defined to be empty.
4316 (c) Changed the dftables auxiliary program (that builds chartables.c) so
4317 that it reads its output file name as an argument instead of writing
4318 to the standard output and assuming this can be redirected.
4319
4320 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
4321 class containing characters with values greater than 255, PCRE compilation
4322 went into a loop.
4323
4324 12. A recursive reference to a subpattern that was within another subpattern
4325 that had a minimum quantifier of zero caused PCRE to crash. For example,
4326 (x(y(?2))z)? provoked this bug with a subject that got as far as the
4327 recursion. If the recursively-called subpattern itself had a zero repeat,
4328 that was OK.
4329
4330 13. In pcretest, the buffer for reading a data line was set at 30K, but the
4331 buffer into which it was copied (for escape processing) was still set at
4332 1024, so long lines caused crashes.
4333
4334 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
4335 "internal error: code overflow...". This applied to any character class
4336 that was followed by a possessive quantifier.
4337
4338 15. Modified the Makefile to add libpcre.la as a prerequisite for
4339 libpcreposix.la because I was told this is needed for a parallel build to
4340 work.
4341
4342 16. If a pattern that contained .* following optional items at the start was
4343 studied, the wrong optimizing data was generated, leading to matching
4344 errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
4345 matching string must start with a or b or c. The correct conclusion for
4346 this pattern is that a match can start with any character.
4347
4348
4349 Version 4.4 13-Aug-03
4350 ---------------------
4351
4352 1. In UTF-8 mode, a character class containing characters with values between
4353 127 and 255 was not handled correctly if the compiled pattern was studied.
4354 In fixing this, I have also improved the studying algorithm for such
4355 classes (slightly).
4356
4357 2. Three internal functions had redundant arguments passed to them. Removal
4358 might give a very teeny performance improvement.
4359
4360 3. Documentation bug: the value of the capture_top field in a callout is *one
4361 more than* the number of the hightest numbered captured substring.
4362
4363 4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
4364 in incorrectly linking with a previously installed version. They now link
4365 explicitly with libpcre.la.
4366
4367 5. configure.in no longer needs to recognize Cygwin specially.
4368
4369 6. A problem in pcre.in for Windows platforms is fixed.
4370
4371 7. If a pattern was successfully studied, and the -d (or /D) flag was given to
4372 pcretest, it used to include the size of the study block as part of its
4373 output. Unfortunately, the structure contains a field that has a different
4374 size on different hardware architectures. This meant that the tests that
4375 showed this size failed. As the block is currently always of a fixed size,
4376 this information isn't actually particularly useful in pcretest output, so
4377 I have just removed it.
4378
4379 8. Three pre-processor statements accidentally did not start in column 1.
4380 Sadly, there are *still* compilers around that complain, even though
4381 standard C has not required this for well over a decade. Sigh.
4382
4383 9. In pcretest, the code for checking callouts passed small integers in the
4384 callout_data field, which is a void * field. However, some picky compilers
4385 complained about the casts involved for this on 64-bit systems. Now
4386 pcretest passes the address of the small integer instead, which should get
4387 rid of the warnings.
4388
4389 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
4390 both compile and run time, and gives an error if an invalid UTF-8 sequence
4391 is found. There is a option for disabling this check in cases where the
4392 string is known to be correct and/or the maximum performance is wanted.
4393
4394 11. In response to a bug report, I changed one line in Makefile.in from
4395
4396 -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
4397 to
4398 -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
4399
4400 to look similar to other lines, but I have no way of telling whether this
4401 is the right thing to do, as I do not use Windows. No doubt I'll get told
4402 if it's wrong...
4403
4404
4405 Version 4.3 21-May-03
4406 ---------------------
4407
4408 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
4409 Makefile.
4410
4411 2. Some refactoring to improve the quality of the code:
4412
4413 (i) The utf8_table... variables are now declared "const".
4414
4415 (ii) The code for \cx, which used the "case flipping" table to upper case
4416 lower case letters, now just substracts 32. This is ASCII-specific,
4417 but the whole concept of \cx is ASCII-specific, so it seems
4418 reasonable.
4419
4420 (iii) PCRE was using its character types table to recognize decimal and
4421 hexadecimal digits in the pattern. This is silly, because it handles
4422 only 0-9, a-f, and A-F, but the character types table is locale-
4423 specific, which means strange things might happen. A private
4424 table is now used for this - though it costs 256 bytes, a table is
4425 much faster than multiple explicit tests. Of course, the standard
4426 character types table is still used for matching digits in subject
4427 strings against \d.
4428
4429 (iv) Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
4430 ending in _t are). So I've renamed it as ESC_tee.
4431
4432 3. The first argument for regexec() in the POSIX wrapper should have been
4433 defined as "const".
4434
4435 4. Changed pcretest to use malloc() for its buffers so that they can be
4436 Electric Fenced for debugging.
4437
4438 5. There were several places in the code where, in UTF-8 mode, PCRE would try
4439 to read one or more bytes before the start of the subject string. Often this
4440 had no effect on PCRE's behaviour, but in some circumstances it could
4441 provoke a segmentation fault.
4442
4443 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
4444 to try to read one or more bytes before the start of the subject string.
4445
4446 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
4447 UTF-8 support could misbehave in various ways if the subject string
4448 contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
4449 area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
4450 back over UTF-8 characters.)
4451
4452
4453 Version 4.2 14-Apr-03
4454 ---------------------
4455
4456 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
4457
4458 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
4459 [ON_WINDOWS]: new variable, "#" on non-Windows platforms
4460 [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
4461 [WIN_PREFIX]: new variable, "cyg" for Cygwin
4462 * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
4463 and BUILD_EXEEXT
4464 Note: automatic setting of the BUILD variables is not yet working
4465 set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
4466 compile-time but not at link-time
4467 [LINK]: use for linking executables only
4468 make different versions for Windows and non-Windows
4469 [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
4470 libraries
4471 [LINK_FOR_BUILD]: new variable
4472 [OBJEXT]: use throughout
4473 [EXEEXT]: use throughout
4474 <winshared>: new target
4475 <wininstall>: new target
4476 <dftables.o>: use native compiler
4477 <dftables>: use native linker
4478 <install>: handle Windows platform correctly
4479 <clean>: ditto
4480 <check>: ditto
4481 copy DLL to top builddir before testing
4482
4483 As part of these changes, -no-undefined was removed again. This was reported
4484 to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
4485 in any case.
4486
4487 3. Some tidies to get rid of compiler warnings:
4488
4489 . In the match_data structure, match_limit was an unsigned long int, whereas
4490 match_call_count was an int. I've made them both unsigned long ints.
4491
4492 . In pcretest the fact that a const uschar * doesn't automatically cast to
4493 a void * provoked a warning.
4494
4495 . Turning on some more compiler warnings threw up some "shadow" variables
4496 and a few more missing casts.
4497
4498 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
4499 option, a class that contained a single character with a value between 128
4500 and 255 (e.g. /[\xFF]/) caused PCRE to crash.
4501
4502 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
4503 option, a class that contained several characters, but with at least one
4504 whose value was between 128 and 255 caused PCRE to crash.
4505
4506
4507 Version 4.1 12-Mar-03
4508 ---------------------
4509
4510 1. Compiling with gcc -pedantic found a couple of places where casts were
4511 needed, and a string in dftables.c that was longer than standard compilers are
4512 required to support.
4513
4514 2. Compiling with Sun's compiler found a few more places where the code could
4515 be tidied up in order to avoid warnings.
4516
4517 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
4518 first of these names is deprecated in the latest Autoconf in favour of the name
4519 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
4520 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
4521 analogy I have changed it to CFLAGS_FOR_BUILD.
4522
4523 4. Added -no-undefined to the linking command in the Makefile, because this is
4524 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
4525 linking step for the pcreposix library.
4526
4527 5. PCRE was failing to diagnose the case of two named groups with the same
4528 name.
4529
4530 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
4531 literal character that is needed in the subject for a match, and scans along to
4532 ensure that it is present before embarking on the full matching process. This
4533 saves time in cases of nested unlimited repeats that are never going to match.
4534 Problem: the scan can take a lot of time if the subject is very long (e.g.
4535 megabytes), thus penalizing straightforward matches. It is now done only if the
4536 amount of subject to be scanned is less than 1000 bytes.
4537
4538 7. A lesser problem with the same optimization is that it was recording the
4539 first character of an anchored pattern as "needed", thus provoking a search
4540 right along the subject, even when the first match of the pattern was going to
4541 fail. The "needed" character is now not set for anchored patterns, unless it
4542 follows something in the pattern that is of non-fixed length. Thus, it still
4543 fulfils its original purpose of finding quick non-matches in cases of nested
4544 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
4545
4546
4547 Version 4.0 17-Feb-03
4548 ---------------------
4549
4550 1. If a comment in an extended regex that started immediately after a meta-item
4551 extended to the end of string, PCRE compiled incorrect data. This could lead to
4552 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
4553
4554 2. Moved to autoconf 2.53 and libtool 1.4.2.
4555
4556 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
4557 the special perltest8 script is no longer needed - all the tests can be run
4558 from a single perltest script.
4559
4560 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
4561 by \s. It has now been removed in PCRE. This means it isn't recognized as
4562 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
4563 class [:space:] *does* include VT, thereby creating a mess.
4564
4565 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
4566 space and tab.
4567
4568 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
4569 its new features into the main test script, reducing the number of scripts.
4570
4571 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
4572 were backward compatible, and made the (?i) apply to the whole pattern, as if
4573 /i were given. Now it behaves more logically, and applies the option setting
4574 only to what follows. PCRE has been changed to follow suit. However, if it
4575 finds options settings right at the start of the pattern, it extracts them into
4576 the global options, as before. Thus, they show up in the info data.
4577
4578 8. Added support for the \Q...\E escape sequence. Characters in between are
4579 treated as literals. This is slightly different from Perl in that $ and @ are
4580 also handled as literals inside the quotes. In Perl, they will cause variable
4581 interpolation. Note the following examples:
4582
4583 Pattern PCRE matches Perl matches
4584
4585 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
4586 \Qabc\$xyz\E abc\$xyz abc\$xyz
4587 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
4588
4589 For compatibility with Perl, \Q...\E sequences are recognized inside character
4590 classes as well as outside them.
4591
4592 9. Re-organized 3 code statements in pcretest to avoid "overflow in
4593 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
4594 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
4595 signed/unsigned warnings.
4596
4597 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
4598 option for pcretest, so I've replaced it by a simple function that does just
4599 that job.
4600
4601 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
4602 "pcregrep -".
4603
4604 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
4605 Java package. This provides some syntactic sugar for simple cases of what my
4606 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
4607 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
4608 item, you can use this simplified notation. Note that only makes sense with
4609 greedy quantifiers. Consequently, the use of the possessive quantifier forces
4610 greediness, whatever the setting of the PCRE_UNGREEDY option.
4611
4612 13. A change of greediness default within a pattern was not taking effect at
4613 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
4614 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
4615 was abstracted outside.
4616
4617 14. PCRE now supports the \G assertion. It is true when the current matching
4618 position is at the start point of the match. This differs from \A when the
4619 starting offset is non-zero. Used with the /g option of pcretest (or similar
4620 code), it works in the same way as it does for Perl's /g option. If all
4621 alternatives of a regex begin with \G, the expression is anchored to the start
4622 match position, and the "anchored" flag is set in the compiled expression.
4623
4624 15. Some bugs concerning the handling of certain option changes within patterns
4625 have been fixed. These applied to options other than (?ims). For example,
4626 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
4627 been the other way round. Some of this was related to change 7 above.
4628
4629 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
4630 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
4631 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
4632 POSIX classes only within a class (e.g. /[[:alpha:]]/).
4633
4634 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
4635 mode. Unlike ".", it always matches newline, whatever the setting of
4636 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
4637 assertions. Perl allows it, but it doesn't (in general) work because it can't
4638 calculate the length of the lookbehind. At least, that's the case for Perl
4639 5.8.0 - I've been told they are going to document that it doesn't work in
4640 future.
4641
4642 18. Added an error diagnosis for escapes that PCRE does not support: these are
4643 \L, \l, \N, \P, \p, \U, \u, and \X.
4644
4645 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
4646 reading past the end of the pattern in cases such as /[abcd/.
4647
4648 20. PCRE was getting more memory than necessary for patterns with classes that
4649 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
4650
4651 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
4652 compiling PCRE for use with Virtual Pascal.
4653
4654 22. Small fix to the Makefile to make it work properly if the build is done
4655 outside the source tree.
4656
4657 23. Added a new extension: a condition to go with recursion. If a conditional
4658 subpattern starts with (?(R) the "true" branch is used if recursion has
4659 happened, whereas the "false" branch is used only at the top level.
4660
4661 24. When there was a very long string of literal characters (over 255 bytes
4662 without UTF support, over 250 bytes with UTF support), the computation of how
4663 much memory was required could be incorrect, leading to segfaults or other
4664 strange effects.
4665
4666 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
4667 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
4668 there was a subsequent back reference to those brackets. This meant that, for
4669 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
4670 possible to check for precisely this case. All we can do is abandon the
4671 optimization if .* occurs inside capturing brackets when there are any back
4672 references whatsoever. (See below for a better fix that came later.)
4673
4674 26. The handling of the optimization for finding the first character of a
4675 non-anchored pattern, and for finding a character that is required later in the
4676 match were failing in some cases. This didn't break the matching; it just
4677 failed to optimize when it could. The way this is done has been re-implemented.
4678
4679 27. Fixed typo in error message for invalid (?R item (it said "(?p").
4680
4681 28. Added a new feature that provides some of the functionality that Perl
4682 provides with (?{...}). The facility is termed a "callout". The way it is done
4683 in PCRE is for the caller to provide an optional function, by setting
4684 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
4685 global variable. By default it is unset, which disables all calling out. To get
4686 the function called, the regex must include (?C) at appropriate points. This
4687 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
4688 This provides a means of identifying different callout points. When PCRE
4689 reaches such a point in the regex, if pcre_callout has been set, the external
4690 function is called. It is provided with data in a structure called
4691 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
4692 matching continues; if it returns a non-zero value, the match at the current
4693 point fails. However, backtracking will occur if possible. [This was changed
4694 later and other features added - see item 49 below.]
4695
4696 29. pcretest is upgraded to test the callout functionality. It provides a
4697 callout function that displays information. By default, it shows the start of
4698 the match and the current position in the text. There are some new data escapes
4699 to vary what happens:
4700
4701 \C+ in addition, show current contents of captured substrings
4702 \C- do not supply a callout function
4703 \C!n return 1 when callout number n is reached
4704 \C!n!m return 1 when callout number n is reached for the mth time
4705
4706 30. If pcregrep was called with the -l option and just a single file name, it
4707 output "<stdin>" if a match was found, instead of the file name.
4708
4709 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
4710 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
4711 pcre_exec(). This saves a malloc/free per call. The default value of
4712 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
4713 when configuring.
4714
4715 32. The default maximum size of a compiled pattern is 64K. There have been a
4716 few cases of people hitting this limit. The code now uses macros to handle the
4717 storing of links as offsets within the compiled pattern. It defaults to 2-byte
4718 links, but this can be changed to 3 or 4 bytes by --with-link-size when
4719 configuring. Tests 2 and 5 work only with 2-byte links because they output
4720 debugging information about compiled patterns.
4721
4722 33. Internal code re-arrangements:
4723
4724 (a) Moved the debugging function for printing out a compiled regex into
4725 its own source file (printint.c) and used #include to pull it into
4726 pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
4727 separate copies.
4728
4729 (b) Defined the list of op-code names for debugging as a macro in
4730 internal.h so that it is next to the definition of the opcodes.
4731
4732 (c) Defined a table of op-code lengths for simpler skipping along compiled
4733 code. This is again a macro in internal.h so that it is next to the
4734 definition of the opcodes.
4735
4736 34. Added support for recursive calls to individual subpatterns, along the
4737 lines of Robin Houston's patch (but implemented somewhat differently).
4738
4739 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
4740 allow it to read and process whole directories in Win32. This code was
4741 contributed by Lionel Fourquaux; it has not been tested by me.
4742
4743 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
4744 used to name a group. Names consist of alphanumerics and underscores, and must
4745 be unique. Back references use the syntax (?P=name) and recursive calls use
4746 (?P>name) which is a PCRE extension to the Python extension. Groups still have
4747 numbers. The function pcre_fullinfo() can be used after compilation to extract
4748 a name/number map. There are three relevant calls:
4749
4750 PCRE_INFO_NAMEENTRYSIZE yields the size of each entry in the map
4751 PCRE_INFO_NAMECOUNT yields the number of entries
4752 PCRE_INFO_NAMETABLE yields a pointer to the map.
4753
4754 The map is a vector of fixed-size entries. The size of each entry depends on
4755 the length of the longest name used. The first two bytes of each entry are the
4756 group number, most significant byte first. There follows the corresponding
4757 name, zero terminated. The names are in alphabetical order.
4758
4759 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
4760 case instead of 255. Making it the same both with and without UTF-8 support
4761 means that the same test output works with both.
4762
4763 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
4764 calling malloc() with a zero argument.
4765
4766 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
4767 optimization. I've improved things by keeping a bitmap of backreferences with
4768 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
4769 fact referenced, the optimization can be applied. It is unlikely that a
4770 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
4771 the match to follow \n) will appear inside brackets with a number greater than
4772 31, but if it does, any back reference > 31 suppresses the optimization.
4773
4774 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
4775 of disabling numbered capturing parentheses. Any opening parenthesis that is
4776 not followed by ? behaves as if it were followed by ?: but named parentheses
4777 can still be used for capturing (and they will acquire numbers in the usual
4778 way).
4779
4780 41. Redesigned the return codes from the match() function into yes/no/error so
4781 that errors can be passed back from deep inside the nested calls. A malloc
4782 failure while inside a recursive subpattern call now causes the
4783 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
4784
4785 42. It is now possible to set a limit on the number of times the match()
4786 function is called in a call to pcre_exec(). This facility makes it possible to
4787 limit the amount of recursion and backtracking, though not in a directly
4788 obvious way, because the match() function is used in a number of different
4789 circumstances. The count starts from zero for each position in the subject
4790 string (for non-anchored patterns). The default limit is, for compatibility, a
4791 large number, namely 10 000 000. You can change this in two ways:
4792
4793 (a) When configuring PCRE before making, you can use --with-match-limit=n
4794 to set a default value for the compiled library.
4795
4796 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
4797 a different value is set. See 45 below.
4798
4799 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
4800
4801 43. Added a new function pcre_config(int, void *) to enable run-time extraction
4802 of things that can be changed at compile time. The first argument specifies
4803 what is wanted and the second points to where the information is to be placed.
4804 The current list of available information is:
4805
4806 PCRE_CONFIG_UTF8
4807
4808 The output is an integer that is set to one if UTF-8 support is available;
4809 otherwise it is set to zero.
4810
4811 PCRE_CONFIG_NEWLINE
4812
4813 The output is an integer that it set to the value of the code that is used for
4814 newline. It is either LF (10) or CR (13).
4815
4816 PCRE_CONFIG_LINK_SIZE
4817
4818 The output is an integer that contains the number of bytes used for internal
4819 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
4820
4821 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
4822
4823 The output is an integer that contains the threshold above which the POSIX
4824 interface uses malloc() for output vectors. See item 31 above.
4825
4826 PCRE_CONFIG_MATCH_LIMIT
4827
4828 The output is an unsigned integer that contains the default limit of the number
4829 of match() calls in a pcre_exec() execution. See 42 above.
4830
4831 44. pcretest has been upgraded by the addition of the -C option. This causes it
4832 to extract all the available output from the new pcre_config() function, and to
4833 output it. The program then exits immediately.
4834
4835 45. A need has arisen to pass over additional data with calls to pcre_exec() in
4836 order to support additional features. One way would have been to define
4837 pcre_exec2() (for example) with extra arguments, but this would not have been
4838 extensible, and would also have required all calls to the original function to
4839 be mapped to the new one. Instead, I have chosen to extend the mechanism that
4840 is used for passing in "extra" data from pcre_study().
4841
4842 The pcre_extra structure is now exposed and defined in pcre.h. It currently
4843 contains the following fields:
4844
4845 flags a bitmap indicating which of the following fields are set
4846 study_data opaque data from pcre_study()
4847 match_limit a way of specifying a limit on match() calls for a specific
4848 call to pcre_exec()
4849 callout_data data for callouts (see 49 below)
4850
4851 The flag bits are also defined in pcre.h, and are
4852
4853 PCRE_EXTRA_STUDY_DATA
4854 PCRE_EXTRA_MATCH_LIMIT
4855 PCRE_EXTRA_CALLOUT_DATA
4856
4857 The pcre_study() function now returns one of these new pcre_extra blocks, with
4858 the actual study data pointed to by the study_data field, and the
4859 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
4860 before. That is, this change is entirely upwards-compatible and requires no
4861 change to existing code.
4862
4863 If you want to pass in additional data to pcre_exec(), you can either place it
4864 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
4865 block.
4866
4867 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
4868 data string contains the escape sequence \M, pcretest calls pcre_exec() several
4869 times with different match limits, until it finds the minimum value needed for
4870 pcre_exec() to complete. The value is then output. This can be instructive; for
4871 most simple matches the number is quite small, but for pathological cases it
4872 gets very large very quickly.
4873
4874 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
4875 returns the size of the data block pointed to by the study_data field in a
4876 pcre_extra block, that is, the value that was passed as the argument to
4877 pcre_malloc() when PCRE was getting memory in which to place the information
4878 created by pcre_study(). The fourth argument should point to a size_t variable.
4879 pcretest has been extended so that this information is shown after a successful
4880 pcre_study() call when information about the compiled regex is being displayed.
4881
4882 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
4883 because what follows is always an absolute path. (Later: it turns out that this
4884 is more than cosmetic for MinGW, because it doesn't like empty path
4885 components.)
4886
4887 49. Some changes have been made to the callout feature (see 28 above):
4888
4889 (i) A callout function now has three choices for what it returns:
4890
4891 0 => success, carry on matching
4892 > 0 => failure at this point, but backtrack if possible
4893 < 0 => serious error, return this value from pcre_exec()
4894
4895 Negative values should normally be chosen from the set of PCRE_ERROR_xxx
4896 values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
4897 "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
4898 use by callout functions. It will never be used by PCRE itself.
4899
4900 (ii) The pcre_extra structure (see 45 above) has a void * field called
4901 callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
4902 pcre_callout_block structure has a field of the same name. The contents of
4903 the field passed in the pcre_extra structure are passed to the callout
4904 function in the corresponding field in the callout block. This makes it
4905 easier to use the same callout-containing regex from multiple threads. For
4906 testing, the pcretest program has a new data escape
4907
4908 \C*n pass the number n (may be negative) as callout_data
4909
4910 If the callout function in pcretest receives a non-zero value as
4911 callout_data, it returns that value.
4912
4913 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
4914 there were some redundant $(CFLAGS) in commands that are now specified as
4915 $(LINK), which already includes $(CFLAGS).
4916
4917 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
4918 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
4919 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
4920 one-byte characters throughout. Note that case-insensitive matching applies
4921 only to characters whose values are less than 256. PCRE doesn't support the
4922 notion of cases for higher-valued characters.
4923
4924 (i) A character class whose characters are all within 0-255 is handled as
4925 a bit map, and the map is inverted for negative classes. Previously, a
4926 character > 255 always failed to match such a class; however it should
4927 match if the class was a negative one (e.g. [^ab]). This has been fixed.
4928
4929 (ii) A negated character class with a single character < 255 is coded as
4930 "not this character" (OP_NOT). This wasn't working properly when the test
4931 character was multibyte, either singly or repeated.
4932
4933 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
4934 mode, for example: \x{100}{2,3}.
4935
4936 (iv) The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
4937 singly or repeated) now correctly test multibyte characters. However,
4938 PCRE doesn't recognize any characters with values greater than 255 as
4939 digits, spaces, or word characters. Such characters always match \D, \S,
4940 and \W, and never match \d, \s, or \w.
4941
4942 (v) Classes may now contain characters and character ranges with values
4943 greater than 255. For example: [ab\x{100}-\x{400}].
4944
4945 (vi) pcregrep now has a --utf-8 option (synonym -u) which makes it call
4946 PCRE in UTF-8 mode.
4947
4948 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
4949 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
4950 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
4951 value.)
4952
4953 53. The single man page has become too large. I have therefore split it up into
4954 a number of separate man pages. These also give rise to individual HTML pages;
4955 these are now put in a separate directory, and there is an index.html page that
4956 lists them all. Some hyperlinking between the pages has been installed.
4957
4958 54. Added convenience functions for handling named capturing parentheses.
4959
4960 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
4961 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
4962 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
4963 are faulted.
4964
4965 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
4966 calling configure. These values are used when compiling the dftables.c program
4967 which is run to generate the source of the default character tables. They
4968 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
4969 you will need to set these values.
4970
4971 57. Updated the building process for Windows DLL, as provided by Fred Cox.
4972
4973
4974 Version 3.9 02-Jan-02
4975 ---------------------
4976
4977 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
4978
4979 2. If --disable-static was given, the building process failed when trying to
4980 build pcretest and pcregrep. (For some reason it was using libtool to compile
4981 them, which is not right, as they aren't part of the library.)
4982
4983
4984 Version 3.8 18-Dec-01
4985 ---------------------
4986
4987 1. The experimental UTF-8 code was completely screwed up. It was packing the
4988 bytes in the wrong order. How dumb can you get?
4989
4990
4991 Version 3.7 29-Oct-01
4992 ---------------------
4993
4994 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
4995 This caused pcretest, when used on the test data, to segfault. Unfortunately,
4996 this didn't happen under Solaris 8, where I normally test things.
4997
4998 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
4999 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
5000 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
5001 this fix an hour or so after the initial 3.7 release.)
5002
5003
5004 Version 3.6 23-Oct-01
5005 ---------------------
5006
5007 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
5008 offsets passed as NULL with zero offset count.
5009
5010 2. The config.guess and config.sub files had not been updated when I moved to
5011 the latest autoconf.
5012
5013
5014 Version 3.5 15-Aug-01
5015 ---------------------
5016
5017 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
5018 had been forgotten.
5019
5020 2. By using declared but undefined structures, we can avoid using "void"
5021 definitions in pcre.h while keeping the internal definitions of the structures
5022 private.
5023
5024 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
5025 user point of view, this means that both static and shared libraries are built
5026 by default, but this can be individually controlled. More of the work of
5027 handling this static/shared cases is now inside libtool instead of PCRE's make
5028 file.
5029
5030 4. The pcretest utility is now installed along with pcregrep because it is
5031 useful for users (to test regexs) and by doing this, it automatically gets
5032 relinked by libtool. The documentation has been turned into a man page, so
5033 there are now .1, .txt, and .html versions in /doc.
5034
5035 5. Upgrades to pcregrep:
5036 (i) Added long-form option names like gnu grep.
5037 (ii) Added --help to list all options with an explanatory phrase.
5038 (iii) Added -r, --recursive to recurse into sub-directories.
5039 (iv) Added -f, --file to read patterns from a file.
5040
5041 6. pcre_exec() was referring to its "code" argument before testing that
5042 argument for NULL (and giving an error if it was NULL).
5043
5044 7. Upgraded Makefile.in to allow for compiling in a different directory from
5045 the source directory.
5046
5047 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
5048 options bits, the pointer it was passed was to an int instead of to an unsigned
5049 long int. This mattered only on 64-bit systems.
5050
5051 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
5052 generated) instead of pcre.in, which it its source. Also made the same change
5053 in several of the .c files.
5054
5055 10. A new release of gcc defines printf() as a macro, which broke pcretest
5056 because it had an ifdef in the middle of a string argument for printf(). Fixed
5057 by using separate calls to printf().
5058
5059 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
5060 script, to force use of CR or LF instead of \n in the source. On non-Unix
5061 systems, the value can be set in config.h.
5062
5063 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
5064 absolute limit. Changed the text of the error message to make this clear, and
5065 likewise updated the man page.
5066
5067 13. The limit of 99 on the number of capturing subpatterns has been removed.
5068 The new limit is 65535, which I hope will not be a "real" limit.
5069
5070
5071 Version 3.4 22-Aug-00
5072 ---------------------
5073
5074 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
5075
5076 2. Diagnose condition (?(0) as an error instead of crashing on matching.
5077
5078
5079 Version 3.3 01-Aug-00
5080 ---------------------
5081
5082 1. If an octal character was given, but the value was greater than \377, it
5083 was not getting masked to the least significant bits, as documented. This could
5084 lead to crashes in some systems.
5085
5086 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
5087 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
5088
5089 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
5090 These just pass their arguments on to (pcre_free)(), but they are provided
5091 because some uses of PCRE bind it to non-C systems that can call its functions,
5092 but cannot call free() or pcre_free() directly.
5093
5094 4. Add "make test" as a synonym for "make check". Corrected some comments in
5095 the Makefile.
5096
5097 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
5098 Makefile.
5099
5100 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
5101 command called pgrep for grepping around the active processes.
5102
5103 7. Added the beginnings of support for UTF-8 character strings.
5104
5105 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
5106 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
5107 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
5108 out for the ar command.)
5109
5110
5111 Version 3.2 12-May-00
5112 ---------------------
5113
5114 This is purely a bug fixing release.
5115
5116 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
5117 of ZA. This was just one example of several cases that could provoke this bug,
5118 which was introduced by change 9 of version 2.00. The code for breaking
5119 infinite loops after an iteration that matches an empty string was't working
5120 correctly.
5121
5122 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
5123 when matched against abbab (for example). After matching an empty string, it
5124 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
5125 caused it to match further down the string than it should.
5126
5127 3. The code contained an inclusion of sys/types.h. It isn't clear why this
5128 was there because it doesn't seem to be needed, and it causes trouble on some
5129 systems, as it is not a Standard C header. It has been removed.
5130
5131 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
5132 were reported on the Macintosh. The changes were from
5133
5134 while ((c = *(++ptr)) != 0 && c != '\n');
5135 to
5136 while ((c = *(++ptr)) != 0 && c != '\n') ;
5137
5138 Totally extraordinary, but if that's what it takes...
5139
5140 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
5141 available. Added HAVE_BCOPY and an autoconf test for it; if neither
5142 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
5143 assumes the way PCRE uses memmove() (always moving upwards).
5144
5145 6. PCRE is being used in one environment where strchr() is not available. There
5146 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
5147 faster code anyway.
5148
5149
5150 Version 3.1 09-Feb-00
5151 ---------------------
5152
5153 The only change in this release is the fixing of some bugs in Makefile.in for
5154 the "install" target:
5155
5156 (1) It was failing to install pcreposix.h.
5157
5158 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
5159
5160
5161 Version 3.0 01-Feb-00
5162 ---------------------
5163
5164 1. Add support for the /+ modifier to perltest (to output $` like it does in
5165 pcretest).
5166
5167 2. Add support for the /g modifier to perltest.
5168
5169 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
5170 matches null strings.
5171
5172 4. Fix perltest so that it doesn't do unwanted things when fed an empty
5173 pattern. Perl treats empty patterns specially - it reuses the most recent
5174 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
5175 effect.
5176
5177 5. The POSIX interface was broken in that it was just handing over the POSIX
5178 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
5179 required a bigger vector, with some working space on the end. This means that
5180 the POSIX wrapper now has to get and free some memory, and copy the results.
5181
5182 6. Added some simple autoconf support, placing the test data and the
5183 documentation in separate directories, re-organizing some of the
5184 information files, and making it build pcre-config (a GNU standard). Also added
5185 libtool support for building PCRE as a shared library, which is now the
5186 default.
5187
5188 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
5189 09 are not valid octal constants. Single digits will be used for minor values
5190 less than 10.
5191
5192 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
5193 existing programs that set these in the POSIX interface can use PCRE without
5194 modification.
5195
5196 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
5197 return all that pcre_info() returns, plus additional data. The pcre_info()
5198 function is retained for compatibility, but is considered to be obsolete.
5199
5200 10. Added experimental recursion feature (?R) to handle one common case that
5201 Perl 5.6 will be able to do with (?p{...}).
5202
5203 11. Added support for POSIX character classes like [:alpha:], which Perl is
5204 adopting.
5205
5206
5207 Version 2.08 31-Aug-99
5208 ----------------------
5209
5210 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
5211 trying to match at the startoffset position, but instead was moving forward to
5212 the next newline as if a previous match had failed.
5213
5214 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
5215 and could get into a loop if a null string was matched other than at the start
5216 of the subject.
5217
5218 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
5219 be distinguished at compile time, and for completeness also added PCRE_DATE.
5220
5221 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
5222 in GnuWin32 environments.
5223
5224
5225 Version 2.07 29-Jul-99
5226 ----------------------
5227
5228 1. The documentation is now supplied in plain text form and HTML as well as in
5229 the form of man page sources.
5230
5231 2. C++ compilers don't like assigning (void *) values to other pointer types.
5232 In particular this affects malloc(). Although there is no problem in Standard
5233 C, I've put in casts to keep C++ compilers happy.
5234
5235 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
5236 should be (const char *).
5237
5238 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
5239 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
5240 However, I haven't made this a standard facility. The documentation doesn't
5241 mention it, and the Makefile doesn't support it.
5242
5243 5. The Makefile now contains an "install" target, with editable destinations at
5244 the top of the file. The pcretest program is not installed.
5245
5246 6. pgrep -V now gives the PCRE version number and date.
5247
5248 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
5249 causing the entire string to be ignored, instead of just the last character.
5250
5251 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
5252 non-matching string, it can take a very, very long time, even for strings of
5253 quite modest length, because of the nested recursion. PCRE now does better in
5254 some of these cases. It does this by remembering the last required literal
5255 character in the pattern, and pre-searching the subject to ensure it is present
5256 before running the real match. In other words, it applies a heuristic to detect
5257 some types of certain failure quickly, and in the above example, if presented
5258 with a string that has no trailing " it gives "no match" very quickly.
5259
5260 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
5261 other alternatives are tried instead.
5262
5263
5264 Version 2.06 09-Jun-99
5265 ----------------------
5266
5267 1. Change pcretest's output for amount of store used to show just the code
5268 space, because the remainder (the data block) varies in size between 32-bit and
5269 64-bit systems.
5270
5271 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
5272 start matching at. This allows lookbehinds to work when searching for multiple
5273 occurrences in a string.
5274
5275 3. Added additional options to pcretest for testing multiple occurrences:
5276
5277 /+ outputs the rest of the string that follows a match
5278 /g loops for multiple occurrences, using the new startoffset argument
5279 /G loops for multiple occurrences by passing an incremented pointer
5280
5281 4. PCRE wasn't doing the "first character" optimization for patterns starting
5282 with \b or \B, though it was doing it for other lookbehind assertions. That is,
5283 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
5284 the letter 'x'. On long subject strings, this gives a significant speed-up.
5285
5286
5287 Version 2.05 21-Apr-99
5288 ----------------------
5289
5290 1. Changed the type of magic_number from int to long int so that it works
5291 properly on 16-bit systems.
5292
5293 2. Fixed a bug which caused patterns starting with .* not to work correctly
5294 when the subject string contained newline characters. PCRE was assuming
5295 anchoring for such patterns in all cases, which is not correct because .* will
5296 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
5297 DOTALL is set at top level; otherwise it knows that patterns starting with .*
5298 must be retried after every newline in the subject.
5299
5300
5301 Version 2.04 18-Feb-99
5302 ----------------------
5303
5304 1. For parenthesized subpatterns with repeats whose minimum was zero, the
5305 computation of the store needed to hold the pattern was incorrect (too large).
5306 If such patterns were nested a few deep, this could multiply and become a real
5307 problem.
5308
5309 2. Added /M option to pcretest to show the memory requirement of a specific
5310 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
5311
5312 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
5313 compiled in such a way that the backtracking after subsequent failure was
5314 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
5315 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
5316
5317
5318 Version 2.03 02-Feb-99
5319 ----------------------
5320
5321 1. Fixed typo and small mistake in man page.
5322
5323 2. Added 4th condition (GPL supersedes if conflict) and created separate
5324 LICENCE file containing the conditions.
5325
5326 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
5327 Perl, that is the internal \ allows the delimiter to be included in the
5328 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
5329 the final delimiter, add \ to the end of the pattern (to test the error).
5330
5331 4. Added the convenience functions for extracting substrings after a successful
5332 match. Updated pcretest to make it able to test these functions.
5333
5334
5335 Version 2.02 14-Jan-99
5336 ----------------------
5337
5338 1. Initialized the working variables associated with each extraction so that
5339 their saving and restoring doesn't refer to uninitialized store.
5340
5341 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
5342 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
5343 fix the problem.
5344
5345 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
5346 calls, and wasn't printing the correct value for compiling calls. Increased the
5347 default value of LOOPREPEAT, and the number of significant figures in the
5348 times.
5349
5350 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
5351
5352 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
5353 a building problem on Windows NT with a FAT file system.
5354
5355
5356 Version 2.01 21-Oct-98
5357 ----------------------
5358
5359 1. Changed the API for pcre_compile() to allow for the provision of a pointer
5360 to character tables built by pcre_maketables() in the current locale. If NULL
5361 is passed, the default tables are used.
5362
5363
5364 Version 2.00 24-Sep-98
5365 ----------------------
5366
5367 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
5368 it any more.
5369
5370 2. Allow quantification of (?>) groups, and make it work correctly.
5371
5372 3. The first character computation wasn't working for (?>) groups.
5373
5374 4. Correct the implementation of \Z (it is permitted to match on the \n at the
5375 end of the subject) and add 5.005's \z, which really does match only at the
5376 very end of the subject.
5377
5378 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
5379
5380 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
5381 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
5382 localized options. All options to pcre_study() were also removed.
5383
5384 7. Add other new features from 5.005:
5385
5386 $(?<= positive lookbehind
5387 $(?<! negative lookbehind
5388 (?imsx-imsx) added the unsetting capability
5389 such a setting is global if at outer level; local otherwise
5390 (?imsx-imsx:) non-capturing groups with option setting
5391 (?(cond)re|re) conditional pattern matching
5392
5393 A backreference to itself in a repeated group matches the previous
5394 captured string.
5395
5396 8. General tidying up of studying (both automatic and via "study")
5397 consequential on the addition of new assertions.
5398
5399 9. As in 5.005, unlimited repeated groups that could match an empty substring
5400 are no longer faulted at compile time. Instead, the loop is forcibly broken at
5401 runtime if any iteration does actually match an empty substring.
5402
5403 10. Include the RunTest script in the distribution.
5404
5405 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
5406 discrepancies, some of which were old and were also with respect to 5.004. They
5407 have now been fixed.
5408
5409
5410 Version 1.09 28-Apr-98
5411 ----------------------
5412
5413 1. A negated single character class followed by a quantifier with a minimum
5414 value of one (e.g. [^x]{1,6} ) was not compiled correctly. This could lead to
5415 program crashes, or just wrong answers. This did not apply to negated classes
5416 containing more than one character, or to minima other than one.
5417
5418
5419 Version 1.08 27-Mar-98
5420 ----------------------
5421
5422 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
5423
5424 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
5425 latter must appear before anything that relies on it in the pattern.
5426
5427
5428 Version 1.07 16-Feb-98
5429 ----------------------
5430
5431 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
5432 repeat of a potentially empty string).
5433
5434
5435 Version 1.06 23-Jan-98
5436 ----------------------
5437
5438 1. Added Markus Oberhumer's little patches for C++.
5439
5440 2. Literal strings longer than 255 characters were broken.
5441
5442
5443 Version 1.05 23-Dec-97
5444 ----------------------
5445
5446 1. Negated character classes containing more than one character were failing if
5447 PCRE_CASELESS was set at run time.
5448
5449
5450 Version 1.04 19-Dec-97
5451 ----------------------
5452
5453 1. Corrected the man page, where some "const" qualifiers had been omitted.
5454
5455 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
5456 input syntax.
5457
5458 3. Fixed memory leak which occurred when a regex with back references was
5459 matched with an offsets vector that wasn't big enough. The temporary memory
5460 that is used in this case wasn't being freed if the match failed.
5461
5462 4. Tidied pcretest to ensure it frees memory that it gets.
5463
5464 5. Temporary memory was being obtained in the case where the passed offsets
5465 vector was exactly big enough.
5466
5467 6. Corrected definition of offsetof() from change 5 below.
5468
5469 7. I had screwed up change 6 below and broken the rules for the use of
5470 setjmp(). Now fixed.
5471
5472
5473 Version 1.03 18-Dec-97
5474 ----------------------
5475
5476 1. A erroneous regex with a missing opening parenthesis was correctly
5477 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
5478 on some systems.
5479
5480 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
5481 it was reported that one broken compiler failed on the former because "code" is
5482 also an independent variable.
5483
5484 3. The erroneous regex a[]b caused an array overrun reference.
5485
5486 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
5487 fail on data ending with that character. (It was going on too far, and checking
5488 the next character, typically a binary zero.) This was specific to the
5489 optimized code for single-character negative classes.
5490
5491 5. Added a contributed patch from the TIN world which does the following:
5492
5493 + Add an undef for memmove, in case the the system defines a macro for it.
5494
5495 + Add a definition of offsetof(), in case there isn't one. (I don't know
5496 the reason behind this - offsetof() is part of the ANSI standard - but
5497 it does no harm).
5498
5499 + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
5500 most of the places where whitespace preceded '#'. I have given up and
5501 allowed the remaining 2 cases to be at the margin.
5502
5503 + Rename some variables in pcre to eliminate shadowing. This seems very
5504 pedantic, but does no harm, of course.
5505
5506 6. Moved the call to setjmp() into its own function, to get rid of warnings
5507 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
5508
5509 7. Constructs such as \d{8,} were compiling into the equivalent of
5510 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
5511 outcome, but in this particular case used more store than had been allocated,
5512 which caused the bug to be discovered because it threw up an internal error.
5513
5514 8. The debugging code in both pcre and pcretest for outputting the compiled
5515 form of a regex was going wrong in the case of back references followed by
5516 curly-bracketed repeats.
5517
5518
5519 Version 1.02 12-Dec-97
5520 ----------------------
5521
5522 1. Typos in pcre.3 and comments in the source fixed.
5523
5524 2. Applied a contributed patch to get rid of places where it used to remove
5525 'const' from variables, and fixed some signed/unsigned and uninitialized
5526 variable warnings.
5527
5528 3. Added the "runtest" target to Makefile.
5529
5530 4. Set default compiler flag to -O2 rather than just -O.
5531
5532
5533 Version 1.01 19-Nov-97
5534 ----------------------
5535
5536 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
5537 like /([ab]*)*/, that is, for classes with more than one character in them.
5538
5539 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
5540 as /((?>a*))*/ (a PCRE_EXTRA facility).
5541
5542
5543 Version 1.00 18-Nov-97
5544 ----------------------
5545
5546 1. Added compile-time macros to support systems such as SunOS4 which don't have
5547 memmove() or strerror() but have other things that can be used instead.
5548
5549 2. Arranged that "make clean" removes the executables.
5550
5551
5552 Version 0.99 27-Oct-97
5553 ----------------------
5554
5555 1. Fixed bug in code for optimizing classes with only one character. It was
5556 initializing a 32-byte map regardless, which could cause it to run off the end
5557 of the memory it had got.
5558
5559 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
5560
5561
5562 Version 0.98 22-Oct-97
5563 ----------------------
5564
5565 1. Fixed bug in code for handling temporary memory usage when there are more
5566 back references than supplied space in the ovector. This could cause segfaults.
5567
5568
5569 Version 0.97 21-Oct-97
5570 ----------------------
5571
5572 1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
5573
5574 2. Optimized negated single characters not to use a bit map.
5575
5576 3. Brought error texts together as macro definitions; clarified some of them;
5577 fixed one that was wrong - it said "range out of order" when it meant "invalid
5578 escape sequence".
5579
5580 4. Changed some char * arguments to const char *.
5581
5582 5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
5583
5584 6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
5585 pcretest.
5586
5587
5588 Version 0.96 16-Oct-97
5589 ----------------------
5590
5591 1. Added a simple "pgrep" utility to the distribution.
5592
5593 2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
5594 unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
5595 where "ddd" means "one or more decimal digits".
5596
5597 3. Fixed serious bug. If a pattern had a back reference, but the call to
5598 pcre_exec() didn't supply a large enough ovector to record the related
5599 identifying subpattern, the match always failed. PCRE now remembers the number
5600 of the largest back reference, and gets some temporary memory in which to save
5601 the offsets during matching if necessary, in order to ensure that
5602 backreferences always work.
5603
5604 4. Increased the compatibility with Perl in a number of ways:
5605
5606 (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
5607 to request this handling. The option can be set at compile or exec time.
5608
5609 (b) $ matches before a terminating newline by default; an option
5610 PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
5611 mode). The option can be set at compile or exec time.
5612
5613 (c) The handling of \ followed by a digit other than 0 is now supposed to be
5614 the same as Perl's. If the decimal number it represents is less than 10
5615 or there aren't that many previous left capturing parentheses, an octal
5616 escape is read. Inside a character class, it's always an octal escape,
5617 even if it is a single digit.
5618
5619 (d) An escaped but undefined alphabetic character is taken as a literal,
5620 unless PCRE_EXTRA is set. Currently this just reserves the remaining
5621 escapes.
5622
5623 (e) {0} is now permitted. (The previous item is removed from the compiled
5624 pattern).
5625
5626 5. Changed all the names of code files so that the basic parts are no longer
5627 than 10 characters, and abolished the teeny "globals.c" file.
5628
5629 6. Changed the handling of character classes; they are now done with a 32-byte
5630 bit map always.
5631
5632 7. Added the -d and /D options to pcretest to make it possible to look at the
5633 internals of compilation without having to recompile pcre.
5634
5635
5636 Version 0.95 23-Sep-97
5637 ----------------------
5638
5639 1. Fixed bug in pre-pass concerning escaped "normal" characters such as \x5c or
5640 \x20 at the start of a run of normal characters. These were being treated as
5641 real characters, instead of the source characters being re-checked.
5642
5643
5644 Version 0.94 18-Sep-97
5645 ----------------------
5646
5647 1. The functions are now thread-safe, with the caveat that the global variables
5648 containing pointers to malloc() and free() or alternative functions are the
5649 same for all threads.
5650
5651 2. Get pcre_study() to generate a bitmap of initial characters for non-
5652 anchored patterns when this is possible, and use it if passed to pcre_exec().
5653
5654
5655 Version 0.93 15-Sep-97
5656 ----------------------
5657
5658 1. /(b)|(:+)/ was computing an incorrect first character.
5659
5660 2. Add pcre_study() to the API and the passing of pcre_extra to pcre_exec(),
5661 but not actually doing anything yet.
5662
5663 3. Treat "-" characters in classes that cannot be part of ranges as literals,
5664 as Perl does (e.g. [-az] or [az-]).
5665
5666 4. Set the anchored flag if a branch starts with .* or .*? because that tests
5667 all possible positions.
5668
5669 5. Split up into different modules to avoid including unneeded functions in a
5670 compiled binary. However, compile and exec are still in one module. The "study"
5671 function is split off.
5672
5673 6. The character tables are now in a separate module whose source is generated
5674 by an auxiliary program - but can then be edited by hand if required. There are
5675 now no calls to isalnum(), isspace(), isdigit(), isxdigit(), tolower() or
5676 toupper() in the code.
5677
5678 7. Turn the malloc/free funtions variables into pcre_malloc and pcre_free and
5679 make them global. Abolish the function for setting them, as the caller can now
5680 set them directly.
5681
5682
5683 Version 0.92 11-Sep-97
5684 ----------------------
5685
5686 1. A repeat with a fixed maximum and a minimum of 1 for an ordinary character
5687 (e.g. /a{1,3}/) was broken (I mis-optimized it).
5688
5689 2. Caseless matching was not working in character classes if the characters in
5690 the pattern were in upper case.
5691
5692 3. Make ranges like [W-c] work in the same way as Perl for caseless matching.
5693
5694 4. Make PCRE_ANCHORED public and accept as a compile option.
5695
5696 5. Add an options word to pcre_exec() and accept PCRE_ANCHORED and
5697 PCRE_CASELESS at run time. Add escapes \A and \I to pcretest to cause it to
5698 pass them.
5699
5700 6. Give an error if bad option bits passed at compile or run time.
5701
5702 7. Add PCRE_MULTILINE at compile and exec time, and (?m) as well. Add \M to
5703 pcretest to cause it to pass that flag.
5704
5705 8. Add pcre_info(), to get the number of identifying subpatterns, the stored
5706 options, and the first character, if set.
5707
5708 9. Recognize C+ or C{n,m} where n >= 1 as providing a fixed starting character.
5709
5710
5711 Version 0.91 10-Sep-97
5712 ----------------------
5713
5714 1. PCRE was failing to diagnose unlimited repeats of subpatterns that could
5715 match the empty string as in /(a*)*/. It was looping and ultimately crashing.
5716
5717 2. PCRE was looping on encountering an indefinitely repeated back reference to
5718 a subpattern that had matched an empty string, e.g. /(a|)\1*/. It now does what
5719 Perl does - treats the match as successful.
5720
5721 ****

Properties

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

webmaster@exim.org
ViewVC Help
Powered by ViewVC 1.1.12